· pdf fileproprietary rights statement this document contains information, which is...

PROPRIETARY RIGHTS STATEMENT

This document contains information, which is proprietary to the Flex5Gware Consortium.

Research and Innovation Action

Flex5Gware Flexible and efficient hardware/software platforms for

5G network elements and devices

H2020 Grant Agreement Number: 671563

WP 2 – RF front-ends and antennas

D 2.1 - Requirements and concepts for the analogue HW in 5G mobile systems

Contractual Delivery Date: December 31st, 2015

Actual Delivery Date: December 22nd, 2015

Responsible Beneficiary: IMC

Contributing Beneficiaries: ALUD, CTTC, EAB, F-IAF, iMinds, IMC, KU Leuven

Dissemination Level: Public

Version: V1.0

Ref. Ares(2015)6009101 - 22/12/2015

PROPRIETARY RIGHTS STATEMENT

This document contains information, which is proprietary to the Flex5Gware Consortium.

This page is left blank intentionally

H2020 Grant Agreement Number: 671563 Document ID: WP2 / D2.1


Document Information

Document ID: WP2 / D2.1

Version Date: December 14th, 2015

Total Number of Pages: 54

Abstract: This deliverable describes the analysis of requirements and

envisaged concepts for RF front-ends and antennas designated

for network elements and devices in 5G mobile systems. Key

performance parameters are identified based on use cases

elaborated in work package 1. Concepts and solutions of

analogue hardware components envisaged for detailed

evaluation during the project are presented. They are defined to

significantly improve the capability of mobile systems to meet

the challenges coming from the expected increase of data

volume and connected devices.

Keywords: 5G wireless communication, network element, devices, RF front

end, transceiver, power amplifier, filter, antenna, versatility,

multi band, multi antennas, millimetre wave, hardware

impairments.

Authors

Full Name Beneficiary / Organisation

e-mail Role

Dieter Ferling ALUD [email protected] Section Editor / Contributor

Ana Collado CTTC [email protected] Contributor

Apostolos Georgiadis CTTC [email protected] Contributor

Fermin Mira CTTC [email protected] Contributor

Henrik Sjöland EAB [email protected] Contributor

Stefan Andersson EAB [email protected] Section Editor / Contributor

Markus Mußer F-IAF [email protected] Section Editor / Contributor

Hendrik Rogier iMinds [email protected] Contributor

Sam Agneessens iMinds [email protected] Contributor

Camila Priale IMC [email protected] Overall Editor/ Contributor

Paramartha Indirayanti

KU Leuven [email protected] Contributor

Patrick Reynart KU Leuven [email protected] Contributor

mailto:[email protected]














Reviewers

Full Name Beneficiary / Organisation

e-mail Date

Stefan Andersson EAB [email protected] 12.11.2015

Markus Mußer F-IAF [email protected] 16.11.2015

Hendrik Rogier iMinds [email protected] 13.11.2015

Patrick Reynart KU Leuven [email protected] 16.11.2015

Pablo Serrano UC3M [email protected] 07.12.2015

Vincent Berg CEA [email protected] 03.12.2015

Miquel Payaró CTTC [email protected] 03.12.2015

Version history

Version Date Comments

V1.0 14/12/2015










Executive Summary

This document comprises an analysis of requirements and envisaged concepts for RF front-ends and antennas designated for network elements and devices in 5G mobile systems. Key performance parameters have been identified for several functional blocks to be considered for the definition of strategies, concepts, and solutions supporting the design of key building blocks for the analogue hardware, which show significant progression compared to state-of-the-art and a significant step towards 5G mobile applications. Concepts and solutions envisaged for detailed evaluation during the project are defined and the supported use cases, elaborated in work package 1 (5G Architecture requirements, specifications, and use cases), are indicated. The hardware concepts build on expanded RF front-end solutions for substantial extension of operation bandwidth below 6 GHz, and on exploiting new radio bands expected for mobile operation between 6 and 100 GHz. Several solutions expand the performance on throughput, user data rate, power consumption or costs by exploiting the significantly increased aggregated operation bandwidth for the purpose of 5G mobile systems.



Table of Contents

1. Introduction ..................................................................................................... 11

2. Requirements on 5G RF Front Ends and Antennas ..................................... 14

2.1 RF Bandwidth and Operated Frequency Bands ...............................................14 2.1.1 Radio bands below 6 GHz .................................................................................14 2.1.2 Radio bands above 6 GHz ................................................................................15

2.2 Base Station Transceivers for Multiband and Multi-Antenna Applications ....16 2.3 RF Impairments of mmWave Transceivers .......................................................18 2.4 Requirements on Amplifiers for 5G Network Elements and User Equipment 20

2.4.1 PAs for versatile base station transceivers below 6 GHz ...................................20 2.4.2 28 GHz CMOS PAs ...........................................................................................21 2.4.3 K-Band off-the-shelf PAs ...................................................................................23

2.5 PLL Requirements for Multi-Antenna mmWave Transceivers .........................25 2.6 Requirements on Filters .....................................................................................26 2.7 Requirements on Antennas for 5G Base Stations ............................................26

2.7.1 Flexible low-profile antenna requirements (< 6 GHz) .........................................26 2.7.2 Flexible low-profile active antenna requirements (< 6 GHz)...............................28 2.7.3 Flexible low-profile antenna array requirements (< 6 GHz) ................................28 2.7.4 Passive planar antenna requirements (20-40 GHz) ...........................................29

2.8 Requirements on Integrated Active Transmit Antennas (20-40 GHz) ..............30

3. Concepts for Versatile RF Front Ends ........................................................... 31

3.1 Multiband RF Front Ends for Base Stations below 6 GHz ................................31 3.2 Power Amplifiers below 6 GHz ...........................................................................33

3.2.1 Efficient simulation and performance evaluation of multiband power amplifiers .33 3.2.2 Multiband high power amplifier for base stations below 6 GHz ..........................34 3.2.3 Power amplifiers for massive MIMO ..................................................................35

3.3 Multiband Filter ...................................................................................................35 3.4 5G Antenna Systems below 6 GHz ....................................................................36

3.4.1 Flexible low-profile antenna ...............................................................................36 3.4.2 Flexible low-profile active antenna ....................................................................37 3.4.3 Flexible low-profile antenna array ......................................................................39

4. Concepts for RF Front Ends above 6 GHz .................................................... 41

4.1 Power Amplifiers up to 28 GHz ..........................................................................41 4.1.1 K-Band power amplifier .....................................................................................41 4.1.2 28 GHz CMOS PA topology selection ...............................................................42

4.2 On-Chip Frequency Generation for mmWave Transceivers ............................43 4.3 RF Impairments Analysis ...................................................................................44 4.4 5G Antenna Systems above 6 GHz ....................................................................46 4.5 Integration of Power Amplifier with Antenna up to 28 GHz .............................46

4.5.1 Flip-chip interconnect ........................................................................................46 4.5.2 Antenna–PA co-optimization .............................................................................47 4.5.3 Integrated active transmit antenna vs. interconnected antenna and PA (20-40 GHz band) ...............................................................................................................47

5. Conclusions ..................................................................................................... 49

6. References ....................................................................................................... 50



List of Figure

Figure 2-1: Summary of total available licensed spectrum available for mobile broadband in 2013 (in MHz) [FCC13]. ........................................................................................................14 Figure 2-2: Amount of spectrum below 6 GHz available for unlicensed broadband use in 2013 (in MHz) [FCC13]. ........................................................................................................15 Figure 2-3: Generic block diagram of a transceiver with a single transmit and receive path.17 Figure 2-4: Direct conversion transmitter architecture..........................................................18 Figure 3-1: Generic block diagram on functional level of a multiband transceiver with N individual transmit and receive paths. ...................................................................................32 Figure 3-2: Generic block diagram of a multiband transceiver with multiband (MB) functional blocks in transmit and receive paths. ....................................................................................32 Figure 3-3: Substrate integrated waveguide. .......................................................................37 Figure 3-4: Full-wave/circuit co-optimization. .......................................................................38 Figure 3-5: Full-wave/circuit co-design implementation. ......................................................38 Figure 3-6: Switched-beam array system. ...........................................................................39 Figure 3-7: Adaptive array system. ......................................................................................39 Figure 4-1: Potential 28 GHz transformer-based Doherty PA implementation. ....................42 Figure 4-2: RF and LO phase shifters. .................................................................................45



List of Tables

Table 1-1: Use cases and KPIs listed in three families with their relevance for RF front-ends assessed in three categories (low, medium, and high). ........................................................11 Table 2-1: List of possible frequency bands above 6 GHz [MET51]. .....................................16 Table 2-2: State-of-the-art multiband power amplifiers. ........................................................20 Table 2-3: State-of-the-art silicon K/Ka band PA ..................................................................21 Table 2-4: 28 GHz CMOS PA, best effort requirements ........................................................23 Table 2-5: Selected reported power amplifiers from 20 to 40 GHz. .......................................24 Table 2-6: 5G Antenna deployment use cases. ....................................................................26 Table 2-7: K-and Ka-band antennas. ....................................................................................29



List of Acronyms and Abbreviations

Term Description

5G Fifth Generation

ACLR Adjacent Channel Leakage Ratio

ADC Analogue Digital Converter

AM-AM Amplitude Modulation to Amplitude Modulation

AM-PM Amplitude Modulation to Phase Modulation

BER Bit Error Rate

CA Aggregation

CMOS Complementary Metal-Oxide-Semiconductor

CPE Common Phase Error

DAC Digital Analogue Converter

DC Direct Current

DFN Dual Flat No-leads

DSP Digital Signal Processing

EMC Electromagnetic Compatibility

EMDB Electromagnetic Database

E-UTRA Evolved Universal Terrestrial Radio Access

EVM Error Vector Magnitude

FDD Frequency Division Duplex

GaN Gallium Nitride

HEMT High Electron Mobility Transistor

HJ-FET Heterojunction Field Effect Transistor

HPA High Power Amplifier

IC Integrated Circuit

ICI Intercarrier Interference

IoT Internet of Things

IQ Real and Imaginary

KPI Key Performance Indicator

LDMOS Laterally Diffused MOSFET

LNA Low Noise Amplifier

LO Local Oscillator

LTCC Low Temperature Co-fired Ceramics

LTE Long Term Evolution

MB Multiband

MIMO Multiple Input Multiple Output

MMIC Monolithic Microwave Integrated Circuit

mmWave Millimetre Wave

NE Network Element

OFDM Orthogonal Frequency Division Multiplexing

PA Power Amplifier

PAE Power Added Efficiency

PAPR Peak to Average Power Ratio



PCB Printed Circuit Board

PLL Phase Locked Loop

PSAT Saturated Output Power

PTFE Polytetrafluoroethylene

Q Quality factor

QAM Quadrature Amplitude Modulation

QFN Quad Flat No-leads

RAT Radio Access Technology

RF Radio Frequency

RX Receive

SiC Silicon Carbide

SiGe Silicon Germanium

SIW Substrate Integrated Waveguide

SNR Signal to Noise Ratio

SOI Silicon On Insulator

TDD Time Division Duplex

TX Transmit

UE User Equipment

WP Work Package



1. Introduction The purpose of this deliverable is to compile the requirements and expected operational conditions for analogue hardware components designated for 5G wireless systems, and to offer a baseline of potential solutions. The key challenges are discussed in the document, and will be used as starting point for further evaluations for the coming tasks within this project. The requirements and concepts for analogue hardware components are based on use cases elaborated in work package (WP) 1, and described in further detail in [F5G11]. Table 1-1 provides an overview on seven use cases defined in the project and allocated to three use case families. For each use case, the relevant key performance indicators (KPI) are listed in the table. The numbers of building blocks for 5G RF front-ends which impact the KPIs are included by dividing them in low, medium or high, with respect to the impact on expected performance improvement for 5G systems.

Table 1-1: Use cases and KPIs listed in three families with their relevance for RF front-ends assessed in three categories (low, medium, and high).

Use case families and use cases

KPIs Low Medium High

Broad Band Access in Dense Areas

Crowded Venues

User data rate 3

Mobile data volume 1 6

Number of users (connected devices) 3

Bandwidth 1 1 4

Flexibility / versatility / re-config. 1 4 2

Resilience 1 1

Latency 3

Dynamic Hotspots User data rate 2 3


Energy efficiency 1 5

Bandwidth 1 1 6

Mobile data volume 1 2 6

Number of users 1 3

Resilience and continuity 3

Broad Band Access Everywhere

50+ Mbps Everywhere Mobile data volume 5

Bandwidth 1 3

Flexibility / versatility / re-configurability (e.g. multi-RAT)

1 3 1


User data rate 1

Resilience (related mainly to Flexibility, versatility, re-config.)

Mobile Broadband in Vehicles

User data rate

Flexibility / versatility / re-config. (scenario driven RAT, BW etc.)

3

Mobile data volume 3

Number of users

Massive Internet of Things

Smart Cities Energy efficiency 5



Number of connected devices 1

Integration / size / footprint 2 2

Cost 4 1


Resilience, continuity 1

Latency 1

Performance Equipment

Cost 4 1

User data rate 3 2

Integration / size / footprint 2 3

Number of users (connected devices) 1

Bandwidth 1 1 3

Flexibility / versatility / re-config. (multi-RAT)

1 2 1


V2X Communications for Enhanced Driving

Latency

Resilience, continuity

User data rate 3

The structure of the document is split into the following main sections: Section 2 explains the anticipated hardware requirements for 5G RF front ends and antennas. It covers aspects of the required RF bandwidth for 5G wireless systems and aspects for both frequency radio bands below 6 GHz and above up to 100 GHz. Due to the expected demand of additional bandwidth for 5G systems, an increased usage of the spectral resources is envisaged for the lower bands together with the exploitation of RF bands operating at high frequencies, assuming that in these frequency ranges larger contiguous radio bands become available. The implementation challenges for RF circuits at high frequencies increase the difficulty to achieve the required performance at high efficiencies. Thus, the impact of hardware impairments is addressed within this section. The specific challenges for base station transceivers are highlighted, as well as for power amplifiers (PA), filters, and antennas for 5G applications, both for network and user equipment. The covered requirements reflect the KPIs identified for the addressed use cases. They show the basis for the concepts of key building blocks for 5G network elements and user equipment defined in the following chapters. Section 3 introduces the concept of a versatile RF front end, operating below 6 GHz. The proposed solutions support use cases for multiband power amplifiers and filters in RF front ends or for multi-antenna applications, intended for base stations with the target to reduce hardware complexity. The focus of the study is to identify key building blocks and consider various aspects in terms of losses, bandwidth, efficiency and linearity. Antenna concepts for deployment in different scenarios are defined as standalone devices, as active antennas, or for supporting multi-antenna systems. Section 4 presents key functional blocks for 5G RF front ends above 6 GHz. An RF impairment analysis above 6 GHz will evaluate the limitations in the key building blocks for transceivers, relevant in achieving the different KPIs. This section defines further concepts for PA implementation up to 28 GHz, for on-chip frequency generation at 30 and 60 GHz, for 5G antenna systems, and for the integration of PAs with antennas. Section 5 summarizes the most important findings regarding the identified building blocks for hardware components requested by considering the solutions proposed for investigation in



this project, which are detailed in the previous sections. It provides an outlook regarding the further study of the various concepts.



2. Requirements on 5G RF Front Ends and Antennas 2.1 RF Bandwidth and Operated Frequency Bands

The aim of significantly increased data rate and throughput in 5G mobile radio communication with respect to prior generations leads to a mandatory usage of high RF-bandwidth for the radio interface. Accordingly, the usage of new radio bands is foreseen in the range up to 6 GHz, where several applications for wireless communication are currently used, and also in the range between 6 and 100 GHz, which shows the potential availability of several GHz bandwidth. 2.1.1 Radio bands below 6 GHz The currently used radio bands are basically positioned between 700 MHz and 2.7 GHz with a total bandwidth of approx. 600 MHz distributed over several bands ( Figure 2-1, [FCC13]). Single operators own licenses for about 200 MHz, which allows them to use roughly 100 MHz for downlink and uplink, respectively when considering frequency division duplexing (FDD) as preferred against time division duplexing (TDD). Different activities are ongoing worldwide on regulatory level to increase the spectrum available for mobile communication. In different countries, new bands will be licensed below 6 GHz to achieve a total amount between 500 and 700 MHz ( Figure 2-1). Further bands are in discussion on the World Radiocommunication Conference 2015 (WRC-15) for mobile usage in the ranges 470-694 MHz, 1350-1517 MHz, 2700-2900 MHz, and 3800-4200 MHz, which we refer to as “pipeline” bandwidth in the table. Additional spectrum shall be used for mobile communication in unlicensed bands (Figure 2-2). Thus, we can expect that the available bandwidth will triple by the years following 2020 and individual operators will be able to use up to 600 MHz bandwidth below 6 GHz.

Figure 2-1: Summary of total available licensed spectrum available for mobile broadband in 2013 (in MHz) [FCC13].



Figure 2-2: Amount of spectrum below 6 GHz available for unlicensed broadband use in 2013 (in MHz) [FCC13].

In order to exploit the bands below 6 GHz for providing high data throughput and high user data rate, a concurrent operation in different bands is mandatory. Similar to the operation in up to three bands, defined for carrier aggregation (CA) in the LTE standard release 12 [ETS14] in radio bands between 700 MHz and 3.6 GHz, simultaneous operation in different bands must also be enabled for 5G by considering the extended range between 450 MHz and 6 GHz. The simultaneous operation of high aggregated bandwidth of 100 to 200 MHz is a fundamental characteristic which supports several use cases of the two broadband access families (Table 1-1). The lower frequencies up to 2 or 3 GHz are appropriate for providing Broadband Access Everywhere, while the frequencies between 2.5 and 6 GHz are more suitable for smaller cells enabling Broadband Access in Dense Areas. This can be completed by operation above 6 GHz to provide additional throughput and data rate in dense areas and hotspots. 2.1.2 Radio bands above 6 GHz Spectrum above 6 GHz with large contiguous bandwidths is required for enabling fast mobile broadband access in dense usage scenarios as Broadband Access in Dense Areas (Table 1-1). Spectrum above 6 GHz will also be used for wireless backhaul solutions for high-capacity ultra-dense small networks [MET54]. There are many frequency bands in the range from 6 to 100 GHz being considered for 5G deployments [MET51].



Table 2-1: List of possible frequency bands above 6 GHz [MET51].

Band [GHz] BW [GHz]

9.9 - 10.6 0.7

14.4-15.35 0.95

17.1 - 17.3 0.2

17.7 - 19.7 2.0

21.2 - 21.4 0.2

27.5 - 29.5 2.0

31.0 - 31.3 0.3

31.8 - 33.4 1.6

36 .0 - 37.0 1.0

40.5 – 42.5 2

42.5 – 43.5 1

43.5 – 45.5 2

45.5 – 47.0 1.5

47.2 – 50.2 3

50.4 – 52.6 2.2

55.78 – 57.0 1.22

57 – 66 7

66 – 71 5

71 – 76 5

81 – 86 5

In the spectrum range above 6 GHz there are a few frequency bands in which industries so far have shown more interest, for example the 15 GHz band where Ericsson is building a 5G testbed together with NTT Docomo. Another one is the 28 GHz band where Samsung Electronics has conducted field measurements. In the latest trials, data rates of 7.5 Gbps for

stationary users and 1.2 Gbps for mobile users moving at speeds of 100 km/h have been

achieved. The 28 GHz band is a hot candidate for the 5G system being launched at the Winter Olympics 2018 in South Korea. There is also some interest in the unlicensed 60 GHz band. All the mentioned frequency bands have bandwidths that enable to achieve data rates exceeding 1 Gbps per user. 2.2 Base Station Transceivers for Multiband and Multi-Antenna Applications

The key functionalities of a signal path in base station transceivers are signal conditioning (e.g. modulation, demodulation, crest factor reduction, digital predistortion), data conversion, frequency generation, frequency conversion, filtering, amplification, and transmission and reception via antennas. The respective functional blocks are shown in Figure 2-3 and can be realized in different ways. Some can be designed as single component or distributed over different components, like filters and amplifier, or others can be implemented as analogue or digital circuits, like IQ modulation or frequency conversion.



Figure 2-3: Generic block diagram of a transceiver with a single transmit and receive path.

Single transceiver paths are usually realized up to eight times in state-of-the-art base stations to support antenna diversity or MIMO operation. Additional up-scaling is requested by future systems for massive MIMO applications in general and for multi-antenna systems at millimetre waves in particular. In general, for both cases, a number in the range between 16 and 64 (or even more) antennas is considered. Transceivers with the same high number of signal paths as antennas show the highest degree of flexibility and performance during operation allowing full digital precoding/beamforming, but demand the highest complexity of the required hardware, leading to high manufacturing costs. To reduce this complexity, approaches like hybrid or analogue precoding/beamforming are discussed in the community to find applicable trade-offs between hardware complexity and performance, which are especially beneficial with the availability of line-of-sight conditions (which is often assumed for applications at millimetre waves). Massive MIMO transceivers Transceivers for massive MIMO applications provide an increased spectral efficiency at decreased output power per transmitter by exploiting spatial multiplexing. They lead to a more complex hardware and require transceiver architectures and hardware solutions optimized for performance, costs, and power consumption. They provide a benefit within the operated bandwidth on user data rate, mobile data volume, and number of users for the use cases Crowded Venues, Dynamic Hotspots, and 50+ Mbps Everywhere (Table 1-1). Millimetre wave transceivers The operation at millimetre wave frequencies enables large contiguous bandwidth to be exploited for mobile communication in new frequency bands and supports the use case family Broadband Access in Dense Areas (Error! Reference source not found.Table 1-1). It requires multi-antenna systems and transceiver architectures optimized for hardware complexity and performance. It shows benefit in dense areas by increased user data rate and mobile data volume. Multiband transceivers up to 6 GHz With the demand on the operation of high aggregated bandwidth at frequencies below 6 GHz (Section 2.1.1), transceivers for multiple bands are required leading to an up-scaling of signal paths with respect to the number of frequency bands and, thus, increasing the degree of hardware complexity. In order to restrict this increase of complexity, solutions to handle simultaneously multiple signal carriers positioned in different radio bands are desired for transceiver components. A direct RF signal generation for multiple radio bands can reduce the number of components required for data conversion and frequency conversion or can shift such functionalities into digital circuits supporting a single transceiver paths for all considered bands. Multiband PA, broadband low noise amplifiers (LNA), multiband filters and

Sig

nal

Co

nd

itio

nin

g Data

ConvFilter

Filter

Freq

Conv

Freq

Conv

PA

LNA

TX

RX

LO

LO

Data

Conv

BB

Filter

Filter



broadband or multiband antennas support the realisation of multiband transceivers without up-scaling the number of transceiver paths corresponding to the number of operated radio bands. Multiband transceivers below 6 GHz support the use case families Broadband Access Everywhere and Broad Band Access in Dense Areas (Table 1-1). They show the benefit of increased aggregated bandwidth enabling high mobile data volume in areas of different cell size and user density by limiting the number of transceiver paths and thus limiting the size of transceivers. 2.3 RF Impairments of mmWave Transceivers

For high frequency broadband systems - especially at mmWave range-, the nonlinearity of analogue components used in RF front ends gives increased challenges in the modelling of circuits and in anticipating the compensation measures required for performance improvements. Therefore, the principal impairments at RF front end of a transceiver and the key requirements needed for achieving high performance will be outlined.

Figure 2-4: Direct conversion transmitter architecture.

The RF front end consists of all components between the antenna and the digital baseband system of a transceiver, namely mixer or modulator, phase shifter, and power amplifier. An example of an RF front end of a transmitter can be seen in Figure 2-4. Use cases with the demand of high data rates, as defined in the Broadband Access families in Table 1-1, are increasing the need for research on how to get a good compromise of conflicting requirements. Furthermore, the request on additional bandwidth will require operation in high frequency RF bands. This creates big challenges to achieve high efficiency and good performance of the circuitry. A balance has to be found in physical size, capacity and cost, which satisfy the expected requirements for 5G RF hardware. This encompasses performance issues such as capacity increase, energy efficiency, reliability, scalability, and modularity. Therefore, hardware imperfections must be taken into account, in particular related to phase noise and IQ imbalance. 5G communications aim to allow transmitting large amount of data modulated by complex waveforms. Therefore, higher order modulation schemes are needed and a highly linear amplification stage is required to minimize distortion. The drawback of highly linear PA is the lack of efficiency, caused by choosing operational conditions resulting in low distortion, and high DC currents. At higher frequency, especially for low cost CMOS implementation, the efficiency and achievable output power decreases. This means that for transmitting a certain amount of signal power, more DC power is required by the PA, compared to the systems operating at carrier frequencies below 6 GHz. The major imperfection of the PA is its



nonlinear response and memory effects [Bou03]. The advantage at mmWave frequencies is the higher bandwidth availability compared to frequencies below 6 GHz. Nevertheless, as bandwidth requirements become larger, the nonlinearity problem of the PA becomes more evident. There is a need to operate in a linear mode to achieve a more linear behaviour. This translates into high peak to average power ratio (PAPR) of the modulated signal, meaning that the power amplifier in the transmitter is operated at a relatively low signal power level. As a result, the amplitude peaks of the signal are not distorted by a saturating amplifier. This causes further reduction of efficiency of the PA and results in low energy efficiency. Therefore, reduction of the required power back-off of the PA becomes important as it enables higher TX output power resulting in improved coverage and will improve energy efficiency of the PA operation as well. Impairments in PAs PA nonlinearity also affects the TX signal in terms of in-band distortion and spectral regrowth. The in-band distortion increases the error vector magnitude (EVM) of a transmitted signal and the spectral regrowth causes adjacent channel interference. The nonlinearity has memory effects, which are important to consider since memoryless amplitude modulation to amplitude modulation (AM-AM) and amplitude modulation to phase modulation (AM-PM) simulation models are not sufficiently accurate to describe a PA. The memory effects cause additional severe distortion in the transmitted signal and subcarriers. Memory effects imply a deviation in amplitude and phase caused by changes in modulation frequency. These can either be thermal or electrical effects. Thermal memory effects are caused by electro-thermal coupling, meaning that variation in the envelope of signal has a rapid change in temperature in active devices of a PA. In case of electrical memory effects, they are caused by the variation of terminal impedance over the input signal bandwidth around the carrier frequency. These effects - defined as phase distortion and amplitude changes over the modulation bandwidth - are challenging the modelling and compensation concepts. Impairments in components for frequency conversion The PA is not the only – but the most critical - source of TX impairments. Mixers, oscillators, quadrature modulators are also sources of impairments, due to low port isolation. They generate imperfections in the output signal. Mixing the RF signal with the local oscillator (LO) realizes frequency conversion within the RF transceivers. The phase of the LO can be non-stationary as a free running oscillator or time varying modelled as a stationary process of a PLL synthesizer. Ideally, the outcome of an LO is a single tone in the frequency domain. However, in reality, the outcome is a modulated tone and the phase shift of the carrier is referred as phase noise. The use of OFDM [Son04] in imperfect oscillators, causes effects called common phase error (CPE) and intercarrier interference (ICI) caused by LO phase noise. Moreover, phase noise causes significant degradation in the performance and reduces the effective SNR at the receiver, limiting the BER and data rate. The high sensitivity of OFDM receivers to phase noise gives several constraints on the design of oscillators. These must be taken into account to avoid that phase noise causes the system to fail. This makes the choice of transceiver architecture important, since different architectures determine efficiency, gain and linearity of transceiver. The most common architecture in wireless devices is the so called direct conversion architecture, which is simple and cost effective. In case of direct conversion transmitter architecture (Figure 2-4) the DACs in the digital parts create the analogue baseband signal, which will be filtered by low pass filter to suppress extra high frequency images created by the DACs. The analogue I and Q signals are mixed with 90 degree out of phase signal of an LO. The LO frequency is chosen as the desired



output frequency and the modulator output is based on I and Q modulator inputs. The translated I and Q components are summed (or subtracted) resulting in a final RF signal. The non-ideal output signal not only depends on the IQ imbalance but also on digital imperfections caused by limitations of waveforms, which as well limits the performance of RF front ends. Several waveforms must be evaluated to ensure optimal spectral efficiency. The recovery of the real symbols depends on the IQ symbol, including the accuracy of the amplitude, phase and frequency. These parameters become more critical in higher order modulation schemes, such as 64-QAM or higher. In a limited bandwidth, where more data needs to be transmitted, higher order modulation is a must and high accuracy becomes an important implementation requirement. The concepts mentioned in this subsection consider use cases such as dynamic hotspots and crowded venues, which address in more detail energy efficiency, flexibility, bandwidth and user data rate KPIs listed in Table 1-1. 2.4 Requirements on Amplifiers for 5G Network Elements and User Equipment

The objective of the PA is to deliver power to the antenna with sufficient bandwidth to ensure high system data rate while maintaining good signal quality (e.g., in terms of EVM). The output power should be sufficiently high to establish a good communication link with the receiver. Since 5G mobile systems are targeted, PAs have to support the radio frequencies of interest including multiband operation. PAs are key elements for enabling the use cases of the Broadband Access families shown in Error! Reference source not found.. 2.4.1 PAs for versatile base station transceivers below 6 GHz The objective of using flexible PAs for base stations is to cover several mobile communication bands, which are nowadays distributed over a large frequency range from 0.45 to 6 GHz. For multiband PAs, large bandwidth as well as increased output power are required, as the power transmitted in two or more bands passes one single PA. State-of-the-art multiband PAs are mainly available for adjacent bands (Table 2-2) with a relative bandwidth of about 20 to 40 %.

Table 2-2: State-of-the-art multiband power amplifiers.

Freq. range Drain Efficiency at 6dB back-off

Peak Power Technology Ref.

1.70 – 2.30 GHz

54 % @ 1.80 GHz 40 % @ 2.14 GHz

42.0 dBm LDMOS [Qur10]

1.70 – 2.60 GHz

45 % @ 1.80 GHz 36 % @ 2.60 GHz

42.2 dBm 42.2 dBm

GaN Asymmetric

[Bat11]

2.20 – 2.96 GHz

40 % @ 2.20 GHz 41 % @ 2.96 GHz

41.3 dBm 40.1 dBm

GaN [Sun12]

2.10 – 2.70 GHz

45 % @ 2.14 GHz 40 % @ 2.65 GHz

45.0 dBm GaN [Gre12]

1.80 – 2.17 GHz

47 % +/- 1pt Full band

50 dBm LDMOS [Mor15]

Classical push-pull [Qur14] concepts provide a large bandwidth, but they are limited in operating frequency. The main limitation for the push-pull concept in the frequency range around 1 GHz is the lack of 180° high power transformers [Cri15]. Novel PA concepts like Class J push-pull [Smi12] provide a large bandwidth. The Class J concept [Wri09] allows harmonic termination over a large bandwidth, which increases both bandwidth and efficiency.



Combining the push-pull concept and the Class J concept, the requirement of a transformer can be reduced, whereby a perfect short over the entire bandwidth is not required. Also here the limitations are suitable for 180° high power transformers. Another disadvantage of these concepts is the back-off efficiency. Qureshi [Qur14] has shown that the combination of the push-pull and Doherty architecture in the lower frequency range is possible, which combines both high efficiency over large bandwidth (push-pull) and increased back-off efficiency (Doherty). Within Flex5Gware, it will be investigated how these concepts for higher frequency and higher power levels can be used. To support multiband transceivers, as defined in Section 2.2, solutions for multiband PAs are required, which allow the operation at higher frequency and higher output power for nonadjacent radio bands with acceptable drain efficiency. The multiband operation including radio bands above 2.7 GHz, like the E-UTRA operating bands 22, 42 and 43 would provide significant additional bandwidth supporting the two Broadband Access use case families (Table 1-1).

2.4.2 28 GHz CMOS PAs The trend of CMOS scaling is advantageous from a 5G perspective. As a result of technology scaling, the miniaturization not only reduces the area, but also reduces the intrinsic capacitance, allowing mmWave operation. Moreover, the implementation in CMOS enables a system-on-chip solution, whereby the DSP, mixed-signal, and the analogue front end circuits are integrated in a chip. Such integration capability reduces the form factor and the bill-of-materials in mobile user equipment, whose volume of production is a major driver of wireless communication development. From the PA’s perspective, the miniaturization due to scaling allows high power density (power per unit area) and the integration with the DSP circuits for predistortion, enabling high linearity. Hence, despite the relatively low power handling capability compared to off-the-shelf PA in III-V compound semiconductors, these benefits and the target application of mobile user equipment motivate the research activity in CMOS PAs. In 5G context, the developed 28 GHz CMOS PA solution targets the following KPIs, namely integration, energy efficiency, bandwidth and data rate. Based on this, the relevant use cases are Performance Equipment, Crowded Venues, and Dynamic Hotspots. Table 2-3 provides a survey of the state-of-the-art of K/Ka band PA. Note that the scope of comparison excludes compound semiconductor technology, such as GaN, GaAs, InP, etc, for fair comparison.

Table 2-3: State-of-the-art silicon K/Ka band PA

Reference

Technology Freq Psat OP1dB

PAEmax

PAE @ 6-dB BO

3dB BW Gain Supply Area

(GHz) (dBm)

(dBm) (%) (%)

(GHz) (dB) (V) (mm2)

[Kuo11] CMOS 180

nm 22 17.4 15.4 12 5.4 5 12 3.6 0.4

[Hua10] CMOS 180

nm 24 22 20 20 5 8 8 3.6 0.42

[Hun10] CMOS 180

nm 21 20.1 16.2 9.3

5 22.5 3.6 0.79

[Koo12] CMOS 130

nm 24 16 13.3 17.7 9 4 15.6 1.2 0.5



[Kim14] BiCMOS 180 nm

24 19.4 13.8 22.3 1* 11.5 37.6 2.4 1.1

[Kuo12] CMOS 180

nm 24 19 15.7 24.7 4* 2* 19 3.6 0.38

[Kuo13] CMOS 180

nm 26 19.5 16 10.2 1.5* 18.2 15.2 3.6 0.86

[Kaw11] CMOS 65

nm 24 23.8 17* 25.1 1* 4 26 2.4 0.9

[Lee10] CMOS 130

nm 21 20 16.5 12.4 3 2* 19.5 12 1.68

[Che05] SiGe 200

nm 21-26

23 20 19.75 3 5 19 1.8 6

[Liu14] CMOS 90

nm 21 20.4 18.3 17.3 4 5 26.9 2.4 0.74

[Che13] CMOS 45 nm SOI

37 20.2 14.5 11.2 2* 6* 5 3.6 0.28

[Cha07] SiGe

130nm 26-40

19.4 15.5 11.2 1* 14 13 1.4 1.83

*Graphically estimated

Different key parameters are relevant for PAs in mobile communication equipment as discussed below. Output power Despite the advantages of reduced area, thus lower intrinsic capacitance, CMOS scaling is not favourable for PA output power since the breakdown voltage of transistors also scales down, limiting the maximum voltage swing. Moreover, scaling does not only happen at the device side. The metal interconnect and dielectrics, particularly the lowest metal layers in the stack, also become thinner, thus increasing power loss. Furthermore, on radio link level, even though the link loss is significantly higher at mmWave frequencies compared to at LTE bands below 6 GHz, thanks to beamforming and small cell implementation, it is possible to provide sufficient output power to meet the link budget requirement. As the cell becomes smaller, the challenge eventually is to balance the output power and the interference level. For supporting the usage of CMOS PA technology, investigations on the maximum achievable output power in 28 nm CMOS technology is required, as it ensures high bandwidth for multi-Gbps data rate. A saturated power of about 20 dBm shall be targeted. Efficiency The efficiency requirement is defined by the drain efficiency and the power-added efficiency (PAE). Integrating over modulated signal time, we get the energy efficiency. In general, PAs are the most efficient at the maximum output power point. However, the utilization of multicarrier and higher order amplitude modulation schemes result into an average power which is much lower from the peak. This forces the PA to backoff several dBs, degrading the efficiency significantly. Hence, the proposed solution should implement efficiency enhancement at power backoff. From the survey in Table 2-3, a target of 10 % and 20 % of backoff and peak efficiency, respectively, is realistic yet still extends the state-of-the-art. PA bandwidth The 5G operation at mmWaves exploits the additionally available spectrum. Referring to Table 2-1, a bandwidth of 2 GHz is available around 28 GHz, allowing Gbps data rate transmission. With the abundance of bandwidth, millimetre wave operation also provides



additional advantage. This enables the possibility of relatively lower complexity circuit implementation, e.g. single carrier modulation scheme, yet still achieving Gbps data rate with low PAPR. For example, QPSK has typically 3.7 dB PAPR. This low PAPR will result into less degradation of PA efficiency due to reduced output power backoff. Depending on the application, this can be a good cost-performance tradeoff compared to multicarrier or higher order amplitude modulation schemes whose PAPR is high (> 8 dB). Power gain The power gain is specified in terms of transducer power gain, defined as the ratio between the power delivered to the load and the power available from the source. The gain specification is determined from the driving requirement for the power amplifier. Such requirement is derived either from the previous front-end blocks in the final integration case or the measurement setup in the standalone case. A gain of approximately 20 dB is targeted. Linearity The final objectives of the linearity figures of merit are to minimize both EVM and out-of-channel emissions (e.g. spectral mask, ACLR). The specification becomes stringent as higher order modulation schemes are used to ensure high data rate for a certain bandwidth. Since EVM is difficult to evaluate in the circuit design phase, further translation into circuit specification sets is critical. The linearity specification can be categorized into amplitude linearity and phase linearity. The typical circuit metrics to quantify amplitude linearity are namely output 1-dB gain compression (OP1dB) and output intermodulation distortion (Output IM3). Amplitude distortion is often categorized into AM-AM distortion. On the other hand, the circuit metrics to quantify phase linearity is AM-PM, which is the phase distortion due to the increase of amplitude. Note that AM denotes amplitude modulation and PM denotes phase modulation. Here, we specify the linearity requirement of 1-dB gain compression at output power of 17 dBm. Following the discussed key parameters, a set of target specifications of the 28 GHz CMOS PA is summarized in Table 2-4.

Table 2-4: 28 GHz CMOS PA, best effort requirements

No. Circuit-level KPI Value Unit

1. Centre frequency 28 GHz

2. Psat 20 dBm

3. OP1dB 17 dBm

4. Transducer power gain 20 dB

5. 3-dB Bandwidth 4 GHz

6. Efficiency @Peak 20 %

7. Efficiency @ 6dB backoff 10 %

8. CMOS technology 28 nm

2.4.3 K-Band off-the-shelf PAs The realization of PAs for user equipment based on off-the-shelf components is considered for operating frequencies in K-band (18 GHz - 27 GHz). Table 2-5 shows a selected set of reported power amplifier designs in the frequency range from 20 to 40 GHz and the obtained performance in terms of power added efficiency (PAE), saturated output power (Psat) and gain. There is a wide performance range corresponding to different fabrication technologies as well as biasing classes. The use of a switching amplifier topology such as Class-F-1/F was able to lead to a state-of-the-art PAE value of 40 % at a low saturated power of 17 dBm



[Mor14]. The use of gallium nitride (GaN) technology allows one to obtain a saturated output power of 37 dBm with a PAE of 30% [CF12].

Table 2-5: Selected reported power amplifiers from 20 to 40 GHz.

Ref, Freq. (GHz)

PAE (%)

Psat (dBm)

OP1dB (dBm)

Gain (dB)

Process technology

[Mor14] 25-30 39.3-40.7 17.1 15 8-10 0.13μm SiGe

[Cam12] 29 30 37 - 25 0.15μm GaN on SiC HEMT

[Hua12] 17-35 30-40 22.5-23.5 21-22 9-12 0.15μm GaAs HEMT

[Che05] 24 13 21 18.8 15-19 0.2μm SiGe

[Kom05] 24 6.5 14.5 11 7 0.18μm CMOS

In order to implement an efficient power amplifier design in K-band using off-the-shelf components, there are several challenges which need to be addressed associated with device availability, substrate materials and fabrication techniques. The upper operating frequency limit is set by device availability. There are a limited number of high power devices operating in K-band. As an example Triquint and Cree offer GaN high electron mobility transistor (HEMT) devices on a silicon carbide (SiC) substrate as bare die devices which are rated up to 18 GHz operation frequency. There are different devices available offering saturated output powers from 6 W to 90 W. For the purposes of this project, devices capable of a saturated power of 6 W will be considered, which can be aligned with lower power (up to 1-2 W) application scenarios such as in femto cells. It should be noted that low noise hetero-junction field effect transistor (HJ-FET) devices (e.g. Renesas) are available in a micro-X plastic package with K-band operating frequency up to 25 GHz, but they can only provide a low output power level of approximately 10 dBm. Operation in K-band will be sought, by designing an amplifier with a minimum operating frequency of 18 GHz, thus utilizing a selected power device above its recommended frequency limit. The upper operating frequency limit and the precise operating bandwidth will be decided based on the device simulated performance, and will not exceed 24 GHz. Similarly, amplifier gain, efficiency and saturated power will be conditioned by the device performance. The rated device performance at the limit of 18 GHz is approximately 6 dB of saturated gain, 40 % of PAE and 36 dBm of saturated power. The performance at higher frequencies is expected to be reduced. Ultimately, the selected device will additionally be chosen based on the availability of a nonlinear device model which can be used for the amplifier design.

Several dielectric substrates are available for planar devices operating at K-band. When the lowest losses are targeted, PTFE based laminates are the best choice, with dielectric

constant around r = 2.2 and dielectric losses lower than tan = 0.001. These substrates (e.g. Rogers Duroid 5880, Arlon Diclad 880) are costly and, moreover, they pose technological challenges regarding the realization of metalized via holes. For low cost applications and easy fabrication, woven glass reinforced hydrocarbon/ceramics such as Rogers Duroid 4003C or Arlon 25N are good candidates, with moderate losses (around tan

=0.002) and leading to a smaller circuit size due to their larger dielectric permittivity (r = 3.38). For further size reduction, substrates with dielectric constant around 10, with losses

around tan =0.002 can be used, like Arlon AD1000 and Rogers Duroid 6010.

Three main fabrication techniques of planar circuits can be considered. The usage of mechanical circuit board plotters offer the capability of milling, drilling and routing of metallic



sheets. Alternatively, the structuring can be done by chemical etching or by means of a laser circuit board plotter. A specific challenge in the fabrication process consists of implementing plated through via holes. It is sometimes possible to drill via holes using laser, thus being able to finish the circuit layout without the use of mechanical machines. The metallization of via holes is usually performed using electroless deposition of copper. However, a simpler alternative exists for small size prototypes consisting of the use of a conductive paste. In summary, in addition to the inherent limitation of the device performance, a number of additional parameters need to be considered and will play an important role in the obtained performance of the designed prototype amplifier, including the accuracy of the device model, the losses of the selected substrate materials, and the resolution of the fabrication method. A saturated power more than 36 dBm and PAE better than 20 % shall be targeted. These numbers will be conditioned on the device performance which will be operated at frequencies above its frequency rating, however the combination of off-the-shelf devices, low cost substrates and fabrication techniques highlight an exciting potential for a cost effective implementation of such circuits in the 5G application framework. 2.5 PLL Requirements for Multi-Antenna mmWave Transceivers

The quality of local oscillator signals, generated by means of Phase Locked Loops (PLL), impacts the performance transceivers and has to meet several requirements. Frequency bands To cover multiple frequency bands the PLL has to be reconfigurable to cover as many frequency bands as possible. A good first design goal is to cover one octave per PLL implementation. The target frequency range here is 28 GHz to 60 GHz. Phase noise requirements Phase noise is a critical parameter, as the targeted carrier spacing in 5G systems is significantly less than in indoor millimetre wave systems. The close-in phase noise, in relation to the sub-carrier spacing, must be especially minimized, to prevent phase noise imposed on the carriers from degrading their neighbours. We therefore plan to investigate a wide-band PLL with low in-band noise fed by a high frequency reference. Frequency control resolution A high frequency control resolution is needed to be able to use different frequency bands with different channel spacing. It is also needed to be able to correct reference frequency errors. Using a fractional-N PLL with a delta-sigma modulator, the frequency resolution is only limited by the number of bits in the digital accumulators. In 28 nm technology, compact, high speed, and low power digital circuits can be realized, and frequency resolution is therefore not regarded as one of the main limitations. Phase control accuracy If possible, to accurately control antenna array pattern nulls to suppress interference, high phase control accuracy is needed. The errors should then be less than a few degrees. Given the importance of beamforming techniques in 5G systems, the phase control accuracy is a key parameter to investigate. Quadrature accuracy Quadrature accuracy is important as it affects the modulation and demodulation accuracy. Errors in the quadrature accuracy result in constellation errors. Depending on the order of the modulation used, there will be different requirements. At high frequencies there will be difficulties obtaining sufficient quadrature accuracy for high order modulation. Especially for



60 GHz transceivers investigation of high quadrature accuracy LO generation is therefore of key importance. All the performance requirements listed above will be addressed to meet KPIs such as user data rate, energy efficiency and low cost implementations for 5G equipment. 2.6 Requirements on Filters

The operation of multiband transceivers require filtering each individual radio band in front of the antenna in order to meet the specification of unwanted emissions defined in the standard documents [ETS14]. The utilisation of individual filter components per radio band is very volume and cost sensitive. Accordingly, the microwave high power filters take up to about 30 % of volume, cause 30-40 % of overall loss, and are responsible for about 40 % of thermal management in modern single RAT state-of-the-art base stations. The traditional high power filter solutions involve the low-cost coaxial cavity filters. They consist of coaxial resonators coupled to each other by apertures. The capacitive coupling of the resonator post and the top cavity surface of the filter lid is increased by a capacitive hat at the end of the post. This capacitance reduces the electrical length of the resonator to about λ/8. The inductive and capacitive coupling of the resonators is achieved by means of metallic stubs with a capacitive coupling to the open ends of the resonators. This technology offers substantial benefits in terms of volume savings in the lower edge of the spectrum (e.g., 700 MHz or 900 MHz), low-cost implementation and it is a very mature technology largely employed in current systems. Hardware systems for 5G applications will require volume, weight and cost savings along with improved performance, including the upper frequency bands of the space below 6 GHz. The fundamental questions that need to be addressed and the technical challenges ahead involve the cost-effective, reduced volume and improved performance multi-band filtering so as the new technologies can become enablers for the new systems. 2.7 Requirements on Antennas for 5G Base Stations

2.7.1 Flexible low-profile antenna requirements (< 6 GHz) The next generation of wireless technology aims for a wide diversity of application scenarios and for high data rate communication as it has been described in [F5G11]. To be able to support the large number of users, and high data volumes, small cell sizes (e.g. femto cell) are necessary, especially with the increased interest in Internet-of-Things. This will increase the number of required base stations and hence it is preferable to deploy small, invisibly integrated base stations. This poses particular challenges to the antenna, which is typically a large component and therefore difficult to integrate, especially when particular boundary conditions in terms of cost and performance must be met. The following table summarizes some relevant potential 5G antenna integration use cases.

Table 2-6: 5G Antenna deployment use cases.

Use cases Specific antenna form-factor requirements

Specific antenna performance requirements

Broadband Access Everywhere: 50+ Mbps

unobtrusive integration

invisible integration

high antenna/platform isolation



everywhere:

compact KPI: Integration, size, footprint

robust performance KPI: Resilience

wideband characteristics for specific applications KPI: Bandwidth

Massive Internet of Things: Performance equipment

low cost

invisible integration

easy deployment

high antenna/platform isolation KPI: Integration, size, footprint

robust performance KPI: Resilience

wideband characteristics for specific applications KPI: Bandwidth

Massive Internet of Things: Smart cities: Smart metering

low cost

compact

easy deployment

high antenna/platform isolation KPI: Integration, size, footprint

compact antenna for long-distance communication KPI: Bandwidth

Given the limitations imposed on antenna size, materials and form factor, it is challenging to develop high-performance antennas that offer the radiation performance required for energy-efficient high data rate communication. Indeed, antenna design for these scenarios requires that four conflicting requirements are satisfied:

1. It should be possible to realize the antenna in a cost-effective manner. 2. The antenna should exhibit stable radiation characteristics in all deployment use

cases. This requires resilient antenna topology that does not suffer from detuning from objects present within close proximity of the antenna.

3. The antenna should be unobtrusively integrated into its environment. In terms of size and footprint. This requires a low-profile antenna geometry, which, combined with the ground plane topology, inherently suffers from narrowband radiation characteristics.

4. The 5G wireless communication system should offer high data rates, large throughput and support large numbers of users. Therefore antennas with large aggregated bandwidth are required. Hence, advanced techniques must be applied to adapt the narrowband low-profile antenna topologies to wideband and multi-band operation.

These requirement rule popular base station antenna solutions, such as reflector antennas (bulky, difficult to integrate) or PCB antennas (low bandwidth). Solutions need to be investigated that enable inconspicuous integration of robust antennas with sufficient bandwidth to enable the realization of very compact base stations for small cells.



2.7.2 Flexible low-profile active antenna requirements (< 6 GHz) The antenna performs a vital function in any wireless system. By receiving or transmitting radiation it allows the transceiver to communicate. The performance of an antenna enables the achievable system’s specifications and an inadequate design may severely limit the overall functionality. As mentioned in Sections 2.1.1 and 2.2, in order to fulfil the need for high mobile data volume, large user density, and data rates, the next-generation systems require large aggregated bandwidth, which can be achieved by relying on multi-band or wideband transceivers. Furthermore, energy efficient operation is necessary to keep overall power consumption low and, as hardware is expensive, the complexity and dimensions of the system must be minimized, requiring a proper design. To be able to meet these requirements and for to reach the transceiver’s full potential, the antenna’s specifications should closely match those of the rest of the system. This can be done by separately designing the different subsystems (Figure 2-3) for a common input/output impedance of 50 Ω. Such an approach often requires matching networks as many active components require different source and load impedances for optimal performance and small antennas typically have input impedances with a large reactive component. Realization of an impedance matching network can be achieved by relying on lumped elements or by using transmission line segments. This methodology makes the design of separate blocks easier, but inadvertently leads to additional losses, large footprint, and increased system complexity. Alternatively, the matching networks can be omitted if a co-design paradigm is adopted. Here, different subsystems are designed simultaneously in order to improve the overall performance, reduce losses (energy efficiency) and decrease complexity (cost). Interesting active integrated antenna designs are the co-design of an amplifier (PA in a TX and LNA in an RX path) and the antenna, where the antenna’s complex input impedance is controlled to match the optimal value for amplifier operation. For multi-band operation, such a co-design of the antenna and active element need to be optimized simultaneously over different frequency bands. 2.7.3 Flexible low-profile antenna array requirements (< 6 GHz)

Multiple-input-multiple-output (MIMO) systems based on the deployment of multi-antenna systems at both transmit and receive sides of the wireless channel will experience an ever increasing growth, in particular in the 5G communication system. Moreover, it is expected that we will see an important increase in the number of antennas applied in arrays deployed in both base stations and user terminals, to optimally exploit their beamforming, diversity and spatial multiplexing gain. In addition, massive MIMO systems, the large number of antenna elements may be exploited to avoid interference between different users by forming pencil beams, thus allowing the system to support a larger number of users.

Although it is relatively easy to deploy arrays with large numbers of antenna elements on base stations, space and cost constraints are much more stringent for user equipments. The challenge consists in combining low-profile antennas into a compact array of suitable size and footprint. Unfortunately, when bringing multiple antennas in each other’s proximity to build an array of limited size, important mutual coupling effects will occur between the antennas. This will complicate the MIMO processing algorithms and thereby increase the overall cost of the array system. To reduce the cost of the total system, we will design a low-cost antenna array with high antenna/platform isolation, such that the antenna array may be easily integrated in different types of environments. To allow simple beamforming



algorithms, we propose a compact planar antenna array with low mutual coupling between the antenna elements.

2.7.4 Passive planar antenna requirements (20-40 GHz) Before setting out the requirements, it is important to again keep in mind the potential 5G antenna integration scenarios, specified in Table 2-6. Given the higher frequency range and the corresponding smaller wavelength, we now have less concerns with antenna size and form factor, whereas the choice of material and the antenna/environment isolation become more critical. This leads to the following set of adapted requirements:

1. The antenna geometry must be easy to fabricate in a cost-effective and reliable manner.

2. The antenna must be easily and efficiently interfaced with active components, such as a power amplifier and/or a low-noise amplifier.

3. The antenna must exhibit stable radiation characteristics in all deployment scenarios outlined in Table 1-1. This requires an antenna topology that does not suffer from detuning from objects present within close proximity of the antenna. Therefore, at least an antenna topology with ground plane is preferred.

4. The antenna must be unobtrusively integrated into its environment. This requires a low-profile antenna geometry, which, combined with the ground plane topology, inherently suffers from narrowband radiation characteristics.

5. The antenna requires a wide bandwidth for supporting the high data rates requested by the 5G wireless communication systems. Hence, advanced techniques must be applied to adapt the narrowband low-profile antenna topologies to wideband operation.

In Table 2-7 we list characteristics of some antenna designs operating in the neighbourhood of 28 GHz, being representative for the current state-of-the-art. Note that these references mainly involve directive antennas, being of horn and array topologies. Within Flex5Gware, the focus is on the co-optimization of the antenna with an amplifier and thus on a single antenna element.

Table 2-7: K-and Ka-band antennas.

Stand-alone antenna

Description No.

layers Tech-nology

fc [GHz]

BW [GHz]

Substrate Gain [dBi]

Eff. [%]

Dimension

s Ref.

SIW horn antenna

2 PCB 21 6.6 2 mm Arlon

8 55 22x57 mm2

[Cai15]

SIW horn with dielectric loading

1 LTCC 21 3.6 1.8 mm Ferro A6

7 x 7x14 mm2

[Tan14]

Antenna array

Description No.

layers Technology

fc [GHz]

BW [GHz]

Substrate Gain [dBi]

Eff. [%]

Dimension

s Ref.

SIW slot antenna array (2x2)

2 PCB 20.5 2.3 1.5 mm RO5880

10 x 15x15 mm2

[Gua15]

SIW slot antenna array (16x16)

2 PCB 20.5 2.5 1.5 mm RO5880

29 x 60x60 mm2

[Gua15]

SIW slot array 2 PCB 33 1.03 1.52 mm 15.64 72.4 60x17 [Kim11]



(1x16) RO3035 mm2

SIW slot array (8x8)

1 PCB 31.5 2.34 0.5 mm 18.74 x 140x1

30 mm2

[Liu09]

Patch antenna array (2x2)

1 PCB 28 1.6 0.750 mm 13 x 20x10 mm2

[Chi11]

Stacked patch array (2x2)

2 LTCC 28 2.85 0.6 mm + 0.16mm

10.35 x 20x10 mm2

[Chi11]

Steerable patch array with branch line coupler (2x3)

1 PCB 28 10 0.254 mm RO5880

14 70 26x14 mm2

[Ora15]*

* Only simulation results presented.

2.8 Requirements on Integrated Active Transmit Antennas (20-40 GHz)

Although most electronic components have been drastically reduced in size over the past years, antenna dimensions are still (and always will be) dictated by the wavelength corresponding to the operating frequency. Hence, the antenna typically requires a large area to achieve decent radiation efficiency and to cover a sufficiently large bandwidths. Therefore, the endeavour to move to higher operating frequencies for exploiting larger bandwidth is also welcome due to the smaller wavelengths, which allow for smaller antenna size. This would benefit the achievable mobile data volume, user data rate and supported number of users. Yet, increasing frequencies typically lead to higher losses in interconnections and to signal integrity issues and electromagnetic compatibility (EMC) problems due to undesired frequency dispersion and radiation. This requires careful design not only of the antenna as a stand-alone component, but also of the interface between the antenna and the active electronic circuitry, such as amplifiers. An optimal design adopts a full-wave/circuit co-optimization paradigm that simultaneously optimizes active and passive parts in a single optimization step, to minimize unwanted effects and optimize the overall performance and coverage. As part of the transmit system, PAs are critical components of the RF front end for 5G wireless systems since they provide the required transmitted signal power and, hence, they consume a large portion of the system power. Directly integrating the PA onto the antenna poses important challenges in terms of cost and reduction of power losses. Yet, a successful co-design will maximize energy efficiency and eliminate unwanted spurious transmissions. Moreover, it will reduce the size and the footprint of the transmitter.



3. Concepts for Versatile RF Front Ends Several concepts are proposed for analogue components to enable powerful mobile systems operating in currently used and in the future expected frequency bands below 6 GHz. The targeted performance improvement will be investigated for solutions supporting multiband operation to maximize the aggregated operation bandwidth and for solutions supporting multi-antenna systems to maximize the mobile data volume and the number of connected users. Concept architectures for multiband transceivers with decreased hardware complexity are considered for operation at different frequency bands and different RF power level. The design and realization of multiband PAs is investigated to limit the number of transceiver chains. Activities on multiband filters and broadband antennas complete the list of key investigations on multiband transceiver solutions. To support the realization of multi-antenna systems, analogue hardware components for multi-chain transceivers of decreased complexity, power consumption, and cost are required. The integration of PAs is investigate in the cost efficient GaN on Si technology, enabling high RF performance in low cost components. Solutions for active antennas and antenna arrays support the realization of large scale multi-antenna systems. 3.1 Multiband RF Front Ends for Base Stations below 6 GHz

Following the requirements on base stations for future broadband performance in mobile communication below 6 GHz, concept architectures for multiband transceivers are targeted. They shall provide a high aggregated bandwidth by considering the already defined radio bands and such which are expected for 5G applications (Section 2.1.1). These concepts shall satisfy a certain level of generality with respect to the addressed radio bands, as it is not clear yet which sets of radio bands will be of highest interest by operators for multiple band operation. Individual operators own different bands and there are regional differences, thus strategies may differ and the availability of additional new bands after 2020 is still unknown. A basic approach, where for each supported radio band an individual transceiver path is realized, is shown in Figure 3-1. It shows the highest level of flexibility regarding the selection of supported radio bands as there is no restriction to consider bands, which can be close together or far apart. The disadvantage is the high hardware complexity due to the multiplication of the transmit paths by the number (N) of radio bands. Solutions to reduce the hardware complexity and cost shall focus on key building blocks of transceiver components which handle simultaneously multiple signal carriers positioned in different radio bands. This includes direct RF signal generation for multiple radio bands with the benefit to integrate the data conversion and frequency conversion for e.g. three bands into one component. For amplifiers, filters, and antennas, multiband solutions are considered, which allow reducing the number of components for transmitter and receiver. This approach supports transceivers for multiband operation with only one transmit and one receive path as shown in Figure 3-2.



Figure 3-1: Generic block diagram on functional level of a multiband transceiver with N individual transmit and receive paths.

Figure 3-2: Generic block diagram of a multiband transceiver with multiband (MB) functional blocks in transmit and receive paths.

A number of N=3 radio bands is proposed for detailed evaluation. It is considered as challenging, but still promising with regards to the addressed key building blocks. It does not imply that each of them must exhibit three dedicated frequency bands, as some bands can be covered by broadband capabilities of components for instance of PAs or antennas. Such multiband transceiver paths can be multiplied in base stations for the same radio bands to support antenna diversity or MIMO applications or for different sets of radio bands to increase the overall aggregated bandwidth.

.

.

.

Sig

nal

Co

nd

itio

nin

g

Bd.1Data Conv

Bd. 1Filter

Bd. 1

Filter

Bd. 1 Freq Conv

Bd. 1Freq Conv

B.1PA

B. 1LNA

Band 1 TX

Band 1 RX

LO 1

LO 1

Bd. 1 Data Conv

BB

Bd. 1

Filter

Bd. 1

Filter

Bd.NDataConv

Bd. NFilter

Bd. N Freq Conv

B.NPA

Band N TX

LO N

Bd. N

Filter

Bd. N

Filter

Bd. NFreq Conv

B. NLNA

Band N RX

LO N

Bd. N Data Conv

Bd. N

Filter

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.S

ign

alC

on

dit

ion

ing MB

Data Conv

MBFilter

MB

Filter

MB Freq Conv

MBFreq Conv

MBPA

MBLNA

MB TX

MB RX

LO

LO

MB Data Conv

BB

MB

Filter

MB

Filter



The work on a multiband transceiver for three radio bands, which could support a total signal bandwidth of 6x20 MHz, will be aligned to the research on key building blocks defined for the WPs 2 and 3. These are direct RF signal generation based on RF-DACs in WP3 (5G Mixed-signal technologies), multiband PA, multiband filters, and optionally multiband or broadband antennas in WP2. Multiband transceivers below 6 GHz show the benefit of increased aggregated bandwidth enabling high mobile data volume in areas of different size and user density by limiting the number of transceiver paths and thus limiting the size of transceivers. As concepts of a certain level of universality, they support all use cases of the broadband access families shown in Table 1-1. 3.2 Power Amplifiers below 6 GHz

Power amplifiers impact the performance of analogue front ends significantly on bandwidth, power level, efficiency, size and cost. Thus, the investigations will target on one hand on multiband PA solutions for maximum aggregated operation bandwidth for high power level and efficiency and on the other hand on PAs for massive MIMO applications, allowing for cost efficient solutions of high efficiency and low size. 3.2.1 Efficient simulation and performance evaluation of multiband power amplifiers Multiband power amplifier design provides an exciting possibility for 5G systems as it can lead to a reduction of the number of required individual power amplifier circuits necessary to satisfy multi-standard operation, which in turn leads to a reduced circuit area and bill of materials and consequently a more compact circuit profile, and more importantly reduced cost. Multiband operation however entails a greater design challenge in order to satisfy the more stringent bandwidth requirements and maintain high performance in terms of gain, saturation power, and power added efficiency in all operating frequency bands. At the same time, concurrent operation in multiple frequency bands unavoidably leads to intermodulation products, which place corresponding filtering requirements in the input and output amplifier networks and also provide a challenge in maintaining a low distortion within the operating frequency bands. A realistic model of a highly efficient dual band power amplifier will be developed, which can be used towards developing new PAPR reduction and power amplifier pre-distortion techniques tailored to 5G systems. The added circuit and operational complexity of multiband amplifiers is reflected in the simulation tools used in their design and optimization through an increased computational load and simulation time but also in terms of the ability of traditional circuit simulation tools to accurately and efficiently handle complex operational scenarios such as concurrent operation employing digitally modulated signals. Simulation of microwave amplifier circuits is typically done in the frequency domain using a harmonic balance simulator, due to the ease of representation of distributed components such as transmission lines in the frequency domain [Maa03]. Harmonic balance simulators represent nonlinear devices such as transistors in the time-domain and employ a fast Fourier transform algorithm to switch between time and frequency domains. Gain and s-parameter simulation and optimization is typically done using a single tone frequency input signal and consequently a harmonic balance simulation using a single fundamental frequency basis, whereas intermodulation distortion is studied using a two-tone input signal and a two-fundamental frequency basis corresponding to a quasi-periodic scenario. While these simple input signal scenarios can be treated efficiently using harmonic balance simulation, the use of complex digitally modulated signals which are not periodic require the use of time-domain integration or envelope transient simulation in order to accurately and efficiently predict the circuit performance.



The envelope transient analysis combines properties of time domain integration and frequency domain harmonic balance in order to efficiently simulate modulated signals. Specifically, it employs a harmonic balance formulation with a frequency basis defined by the carrier frequency and, at the same time, it considers the (complex) amplitudes at the various frequency harmonics to be time-varying. In this way, the fast time variations due to the carrier signal are treated in the frequency domain, and the slow time variations due to the modulation are treated in the time domain. A traditional time domain integration technique would require the use of a sufficiently small time step in order to capture the fast signal variations due to the carrier frequency, and a sufficiently large total simulation interval in order to capture the slow variations due to the modulation. These requirements result in an inefficient simulation and optimization process. By treating fast variations due to the carrier signal in the frequency domain, envelope transient analysis does not require a small integration step to simulate the signal modulation and therefore is more efficient than traditional time domain integration. Furthermore, the realistic simulation of concurrent dual band operation of amplifier circuits requires the use of envelope transient simulation with a 2-fundamental frequency basis, and consequently presents an added simulation challenge [Maa03]. It should be noted that the targeted operation around two center frequencies permits the concurrent operation of two or more radio bands which are covered within the operating amplifier bandwidth around the respective center frequencies. The expected benefit of this work is to demonstrate a computationally efficient model of a dual band amplifier circuit based on envelope transient analysis, and help identify the performance limitations of such circuits under concurrent operation and with digitally modulated input signals. The model can be used as a tool to test and develop new PAPR reduction and pre-distortion techniques with an aim to improve energy efficiency of 5G systems. Dual band operation in the frequency bands of 2.4 GHz Industrial, Scientific and Medical (ISM) band and 3.5 GHz LTE frequency bands will be considered. The amplifier saturated output power will be 27 dBm - 30 dBm with a maximum PAE above 45 % in both frequency bands. A cost efficient solution will be considered exploring commercial off-the-shelf (COTS) components and traditional printed circuit board (PCB) fabrication exploring properties from composite right/left handed (CRLH) transmission line networks used to provide a dual band operation. The model provides a computationally efficient tool for the subsequent development and testing of novel PAPR reduction and pre-distortion techniques. The developed amplifier model specifications are suitable for the Massive Internet of Things use cases of Smart cities and Performance equipment. The targeted KPIs emphasize the requirements for flexibility and versatility on one hand, and energy efficiency on the other hand through highly efficient dual band operation. 3.2.2 Multiband high power amplifier for base stations below 6 GHz To increase flexibility of base stations multiband, high power amplifiers are required. As described in the Section 2.2, complexity thereby can also be reduced. For the realisation of multiband high power amplifiers (MB-HPA), there are two main challenges. On the one hand, a suitable architecture has to be considered e.g. Doherty. On the other hand, the assembly technology which includes the packaging and realisation of the passive components must not exhibit large tolerances. The aim of the MB-HPA design is to cover three radio bands in the frequency range of 2.6-3.8 GHz with a total signal bandwidth of 6x20 MHz. The target peak output power is around 53 dBm, which is typical for macro base station transmitting in three radio bands. In principle, the Doherty architecture is a narrow band concept. There exists trade-off between bandwidth and efficiency. A possible approach area is to investigate bandwidth limitation of the active and passive components. Targeting a frequency range below 6 GHz, the packaging of the active device plays an important role in terms of efficiency and stability.



Different packaging solutions will be investigated and a suitable package will be chosen. Concerning the peak power level and the upper frequency edge of the MB-HPA, a single chip RF power transistor, based on GaN HEMT technology, is developed and designed. Also the passive components will be investigated, mainly the power splitter and combiner, which are necessary for the Doherty architecture. The main aim here is to increase bandwidth and power capability. The MB-HPA below 6 GHz allows the realisation of versatile base stations. In consequence, the number of power amplifiers required in the base station can be reduced. Aggregation of bandwidth also allows for increased mobile data volume. It supports all use cases of the broadband access families shown in Table 1-1. 3.2.3 Power amplifiers for massive MIMO For supporting the realization of base station transceivers for massive MIMO systems, the cost efficient GaN on Si technology will be evaluated. A MMIC power amplifier will be designed, targeting an average output power level of 30 dBm at 3.5 GHz. Furthermore cost efficient packaging such as DFN/QFN is investigated. Base stations for massive MIMO incorporate numerous individual transmitters. Compared to macro base stations, a larger number of smaller PAs is needed. Owing to scaling effects, a multitude of small amplifiers exhibits higher performance than a single large HPA. Consequently the PAs can be fabricated in a low power density technology without sacrificing system efficiency. GaN-on-Si therefore is an interesting option that can provide high performance at low costs. Furthermore Si substrates allow for further integration of CMOS based components onto a single chip, e.g. driver amplifiers, baseband processing, supply regulation, etc. This allows even higher cost efficiency, better system performance and a more compact setup that can be conveniently embedded into the complete system. GaN on Si based PAs are capable of delivering high power and efficiency with highly linear amplification while occupying a small die area only. The resulting compact high performance building block allows manufacturing costs reduction and a boost in system performance. . The high efficiency obtained from GaN based PAs prevents high power dissipation and thereby heating of the system that has to be counteracted by active cooling, which is an issue particularly present in macrocell base stations. The cost efficient GaN on Si technology enable high efficiency, high linear and compact PAs for massive MIMO applications. Thus it enables the increase of mobile data volume and number of users supporting the use cases Crowded Venues, Dynamic Hotspots, and 50+ Mbps Everywhere. 3.3 Multiband Filter

Supporting the requirement on multiband filters for band filtering in front of the antenna, a hardware filter capable of concurrently supporting two or three frequency bands is targeted. The solutions and disruptive innovation will benefit the hardware systems into volume and weight reduction. It shall be able to handle the aspired power levels and provide a further improved filter performance. The considered radio bands will be defined out of the frequency range between 2.6 and 4.2 GHz aligned to the targets of further key functional blocks of multiband transceivers researched in the project. Such bands can be for example around 2.6 GHz, between 3.4 and 3.8 GHz or around 4 GHz, for both FDD and TDD, as already licensed bands or in a spectrum expected to get defined for mobile communication (Section 2.1.1). The average



power levels required for the filters are of the order of 10s of Watts per frequency band so as the combined multiband filter hardware needs to withstand the combined power levels imposing additional challenges in the hardware. Since the focus is on high power low-cost solutions, the coaxial cavity filters can offer a distinct benefit (Section 2.6). Hence, the research focus will be on this technology. The proposed technical solution will target coaxial cavity resonators that can support multiple resonances within the volume of the cavity. This solution can target frequency bands with increased ratio of frequency separation. Alternatively, another solution can be implemented, utilizing both metallic surfaces of a coaxial cavity resonator integrating the volume of a resonator at one frequency with the volume of a resonator at another frequency, achieving the optimum volume and Q ratio. For the implementation of the cavity post/resonator, solutions that can offer wide tuning range can be beneficial. This solution can involve highly coupled resonator posts that allow substantial reduction in the resonator cavity volume while simultaneously increasing tunability. This solution, once implemented, combined with other proposed solutions can also offer the ability to have multiple bands and the ability to tune one of them in a wide range. This multiband filter supports the operation of increased aggregated bandwidth as a key building block in multiband transceivers and provides a reduction of the component volume. The increased tunability can provide certain flexibility and a reduction of manufacturing costs. Due to multiband capability this envisaged filter solutions support all use cases of the broadband access families shown in Table 1-1. 3.4 5G Antenna Systems below 6 GHz

3.4.1 Flexible low-profile antenna A compact flexible low-profile antenna will be designed, fabricated and tested. The antenna design will focus on wideband operation and high antenna/platform isolation and, therefore, the resulting antenna will be suitable for deployment in different operating scenarios, such as MIMO communication systems. The antenna design starts by selecting an appropriate set of materials. The aim is to start from materials that are readily available in the environment where the antenna will be deployed:

Textile fabrics and protective foams applied in garments for, e.g., body-centric applications.

Cork agglomerates found in walls, floors and furniture.

The advantage of making use of such materials is that they already exhibit the mechanical (flexible, breathable), esthetical (non-protruding, low-profile, invisible) and ergonomical (comfortable to wear) properties required by the application at hand. This selection of materials will be fine-tuned by applying dedicated in-house characterization procedures to determine the permittivity and loss tangent of potential antenna substrates. We will select those materials that yield the best antenna radiation properties: lowest loss, largest impedance bandwidth and stable antenna parameters.

Next, the choice of an appropriate antenna topology is required that enables us to meet the four requirements set out in Section 2.7.1. Given that high antenna/platform isolation is required, we will make use of substrate integrated waveguide (SIW) technology to implement low-profile antennas in a cost-effective manner. The basic geometry of an SIW component is shown in Figure 3-3. By relying on rows of vias, the electromagnetic fields may be contained



in the antenna cavity with isolations almost as high as in metal rectangular waveguides. This enables us to design cavity-backed slot antennas that only radiate through the antenna slots, while fields remain fully confined in the antenna cavity. Therefore, the antenna’s radiation characteristics are highly robust and do not change when objects are brought in close proximity of the antenna cavity.

Figure 3-3: Substrate integrated waveguide.

The following step will consist in applying computer-aided full-wave optimization of the antenna geometry, combined with multi-moding techniques to enlarge the impedance bandwidth. The wideband operation will make the design suitable for use in scenarios that require high data rates, high mobile data volume, and large numbers of users. The possibility of integrating electronic circuitry behind the ground plane make the antenna versatile in terms of operating scenarios and helps to reduce the overall footprint of the system. It supports the use cases of the broadband access families. 3.4.2 Flexible low-profile active antenna

An active 5G antenna is designed by full-wave/circuit co-optimization [Riz06, Die14, Die15]. As schematically shown in Figure 3-4, this hybrid strategy pairs the speed of circuit simulators and their ability to analyse and optimize non-linear circuits to the accuracy of full-wave solvers that rigorously solve Maxwell’s equations and, therefore, include all electromagnetic effects, including radiation. The actual implementation of the co-design paradigm is shown in Figure 3-5. Those parts of the design that exhibit significant full-wave effects, such as the antenna, are modelled by an electromagnetic field simulator, which generates a circuit description of the component under study by means of, for example, S-parameters. This description is stored in an electromagnetic database (EMDB). The circuit simulator builds models for the active, non-linear components, such as power amplifiers and combines these models with the descriptions of the full-wave components, whose data are retrieved from the EMDB. An optimizer interacts with the circuit simulator, completing each design cycle by modifying component values in the circuit schematic and the geometry of the structure in the full-wave component layout to improve the design before a new analysis in the next design cycle starts.



Figure 3-4: Full-wave/circuit co-optimization.

Figure 3-5: Full-wave/circuit co-design implementation.

The aim here is to directly integrate an active component, being a dual-band PA with a dual-band antenna covering two bands in the frequency range 2.4-4.2 GHz. To obtain an optimal co-design in a minimal amount of time, our co-optimization strategy will proceed as follows:

1. Preliminary design of the active component, i.e., power amplifier. One of the outcomes of this step will be the antenna impedance that yields optimal performance of the PA.

2. Preliminary design of the antenna, using as a specification the input impedance obtained in step 1.

3. In the final step, the active circuit is combined with the antenna and full-wave/circuit analysis is performed to evaluate the performance of the integrated solution. At this moment, the performance of the active antenna will still be suboptimal. Therefore, a final full-wave/circuit co-optimization procedure will be started to fine-tune both the antenna and power amplifier. In this step, specific attention will be devoted to optimizing the antenna/PA interface.

The active integrated antenna, which will be validated by simulations, should be able to operate at two frequency bands fulfilling the need for large aggregated bandwidth, large mobile data volume, and high number of users. Furthermore, owing to the co-design, the system will be less complex, because there shall be no need for additional matching



networks (cost) and better performance can be obtained (energy efficiency). This concept supports the use cases of the broadband access families. 3.4.3 Flexible low-profile antenna array A compact antenna array will be designed based on the previously developed passive antennas. Special attention will be devoted to low mutual coupling between antenna elements in a compact array configuration. Again, we will exploit the high isolation provided by the SIW technology to obtain compact antenna arrays with low mutual coupling. This will specifically simplify the algorithms required to perform beam steering. To be more specific, we will present two beamforming strategies in Figure 3-6 and Figure 3-7. On the one hand, the switched-beam array in Figure 3-6 sets up of a series of fixed beams and selects the most appropriate beam for each user. The adaptive array system illustrated in Figure 3-7, on the other hand, points one beam to each user and modifies its radiation pattern in real time in accordance to the user position.

Figure 3-6: Switched-beam array system.

Figure 3-7: Adaptive array system.

Particular challenges related to the antenna array design in the 5G system are:

1. Instead of deploying the arrays on a base station, as in the above figures, we want to invisibly integrate the antennas in their environment, embedding them in walls, furniture, ceilings and floors.

2. We need arrays with more antenna elements, especially to target massive MIMO systems. These arrays must still be sufficiently compact and low-profile for non-obtrusive integration into their environment.

3. The antenna array element must exhibit wideband radiation characteristics to support high data rates.



The beamforming potential of the array will be investigated by means of simulations. The beamforming capabilities of antenna arrays facilitate the link improvement between mobile user and base station by forming pencil like beams and can be used for MIMO applications, which leads to higher mobile data volume, user data rate, and number of users. It supports use cases like Crowded Venues, Dynamic Hotspots, and 50+ Mbps Everywhere.



4. Concepts for RF Front Ends above 6 GHz The requested increase on mobile data volume is significantly supported by the exploitation of frequency bands above 6 GHz, expected to become available for 5G mobile communication. For exploiting these spectrum resources, key building blocks will be investigated to promote the realization of RF front ends suitable for wireless communication in this frequency ranges. The focus will be on components for frequency generation, amplifiers and antennas. Components on cost effective 28 nm CMOS technology will be evaluated together with off-the-shelf devices. The analysis of RF impairments for transceivers is required to show the limitations of key functional blocks and for defining solutions to dynamically increase the performance. The integration of active transmit antennas proves solutions supporting the realization of large transceiver arrays. 4.1 Power Amplifiers up to 28 GHz

As it has been pointed out previously, the PA represents a critical component of the 5G wireless system front-end since it is responsible for generating the required output power of the system and since it presents one of the most power consuming components. Therefore, similarly as in the situation presented in Section 3 for bands below 6 GHz, the design of efficient power amplifiers remains an important challenge in terms of maximizing the overall energy efficiency of the system. There are important design trade-offs between linearity, efficiency, output power and gain that need to be addressed by the designer. Furthermore, operation in high frequency bands such as 20 to 40 GHz on one hand provides a significant advantage in terms of available frequency bandwidth, but, on the other hand introduces further challenges in terms of performance (efficiency, output power, linearity, gain) and cost. In an effort to address the above challenges, two different power amplifier circuits will be developed. One based on cost effective integrated circuit process technologies (CMOS) and one based on off-the-shelf available transistor devices. 4.1.1 K-Band power amplifier A high efficiency topology such as Class-F will be targeted in order to maximize the efficiency of the obtained power amplifier [Mor14]. Distributed circuit terminations will be used in order to implement the required 2nd and 3rd harmonic loading to achieve the desired operating class. The substrate parameters such as thickness, permittivity and loss will be carefully selected in order to minimize the losses associated with the input and output networks of the amplifier as well as the supply and bias networks of the device. Co-design with the antenna array, which will be used at the output of the amplifier, is critical in order to ensure the proper loading for the output amplifier network necessary to maximize efficiency. Furthermore, different technologies such as microstrip, coplanar waveguide and substrate integrated waveguide technology will be considered in order to optimize the circuit layout, performance and facilitate interconnection with the antenna array. Substrate integrated waveguide technology will be used to implement the antenna array and presents an exciting alternative to microstrip technology owing to its superior performance at high frequency (SIW has reduced radiation losses compared to a microstip line) [Boz11]. However, the flexibility and compact form of distributed microstrip matching networks has to be considered carefully. Ultimately, as noted in Section 2.4.3, the obtained performance of the power amplifier will be dictated by the selected device as well as the substrate fabrication technology limitations. The expected benefit from the designed amplifier is towards reducing fabrication cost while increasing the obtained energy efficiency and at the same time taking advantage of the large frequency bandwidth available at high frequencies, which can accommodate larger user data rates. These KPIs are essential in use cases such as Smart cities, Performance equipment and V2X Communications for Enhanced Driving under the Massive Internet of Things use case.



4.1.2 28 GHz CMOS PA topology selection

As outlined in Section 2.4.2, our approach is to explore how much output power we can generate from the 28 nm CMOS technology at 28 GHz, while maintaining acceptable efficiency, linearity and high bandwidth. By the same token, with the foreseen enormous amount of data rate and the operation in mmWave, efficient usage of energy becomes paramount. This is exacerbated with the high PAPR as a result of multicarrier and/or higher order amplitude modulation schemes, necessitating the adoption of efficiency enhancement techniques at power backoff. Based on these requirements, the relevant use cases are Performance Equipment, Crowded Venues, and Dynamic Hotspots with KPIs such as integration, energy efficiency, bandwidth and data rate. A possible solution for this is the proposed 28 GHz transformer-based Doherty amplifier, depicted in Figure 4-1.

Figure 4-1: Potential 28 GHz transformer-based Doherty PA implementation.

In order to boost the output power, power combining techniques are utilized. The challenge is to come up with a compact, low loss and broad band solution. Since the implementation is on chip, the size of the passives is critical to minimize the area. Transformer-based power combining shown in Figure 4-1 is a good candidate. The power combining can be performed at the primary side, for instance distributed active transformer (DAT), at the secondary side (Figure-of-eight-shaped transformers), or a hybrid of both [Kay15a]. This method can be combined with slow-wave transmission lines. Further layout co-optimization of active and passive components results in high output power at good efficiency. The transformer-based power combining method can be complemented with smart dynamic switching of the PA periphery to facilitate efficiency enhancement at power backoff, which ultimately improves average modulated-signal efficiency. This concept is called the transformer-based Doherty technique [Kay15a, Kay15b, Aga13]. It is essentially a segmented PA which is composed of: 1) a class B/AB amplifier as the main amplifier and 2) a class-C amplifier as the auxiliary amplifier. The main amplifier is responsible for the amplification at average power levels (backoff point) whereas the auxiliary amplifier switches on and provides amplification at the event of peak power. Such segmentation and ‘on-demand’ activation ensures that the auxiliary amplifier delivers power to the output only when



necessary, thus saving energy. Hence, the targeted KPIs are energy efficiency and re-configurability.

Another potential method of efficiency enhancement at power backoff is adaptive biasing. In this method, the operating point of the transistor is adjusted according to the signal to save power and to extend the linearity. This technique can serve as an alternative or a complementary method to the Doherty implementation. In the transformer-based Doherty PA, adaptive biasing is used for extending the backoff range. A high back-off range in a segmented PA configuration requires an auxiliary amplifier whose aspect ratio is larger than that of the main causing a mismatch in the turn-on point of the auxiliary amplifier, distorting the composite output power profile, and thereby deteriorating the linearity. The adaptive bias then plays a role of dynamically adjusting the bias point of the amplifier to activate at the right output power point. To conclude, in our solution we plan to incorporate circuit techniques such as power combining and a transformer-based Doherty for the PA. These concepts are implemented to achieve high average efficiency at wide band, facilitating high data-rate communication. Moreover, the implementation in CMOS enables a system-on-chip solution, in which the PA can be integrated with the DSP circuits. Such integration capability reduces the form factor and the bill-of-materials in mobile user equipment. These benefits are in line with use cases such as Performance Equipment, Crowded Venues and Dynamic Hotspots. The relevant KPIs in this case are integration, bandwidth, data rate and energy efficiency. 4.2 On-Chip Frequency Generation for mmWave Transceivers

Different architectures for on-chip frequency generation in 28 GHz transceivers will be investigated. A suitable architecture for a PLL will be identified for implementation in a test circuit. It will generate a 19 GHz phase-controlled output signal for use in a sliding-IF 28 GHz transceiver. For high frequency resolution, the PLL will use a delta-sigma modulator controlling the fractional frequency. It will also feature a digital calibration of the voltage-controlled oscillator (VCO) to reduce the VCO tuning sensitivity and thereby PLL in-band noise. Two PLLs will be implemented on a test chip in 28 nm SOI technology, to investigate potential interaction between PLLs, and also to evaluate a technique that produces independent delta-sigma noise in both PLL output signals. This PLL work will serve as a baseline for coming PLLs and key parts will be re-used in PLL concepts intended for higher frequency bands up to 60 GHz. Generation of LO signals for transceivers operating at 60 GHz is another important activity. Oscillators operating directly at 60 GHz will be implemented in a test chip in 28 nm SOI technology, where the advanced technology offers possibilities for realizing good 60 GHz varactors, which otherwise form a well-known bottleneck to millimetre wave VCO performance. Architecture investigations of using lower frequency VCOs will also be performed in combination with sliding-IF transceivers and/or frequency multipliers. Important considerations in this work are the possibilities of implementing efficient frequency dividers and quadrature generation. Another test chip will then be fabricated and tested to prove the concepts in silicon. In conclusion, the work described above on on-chip frequency generation will address the special needs on PLLs for realizing the use cases: Crowded Venues and Dynamic Hotspots for Broadband Access in Dense Areas as well as Performance Equipment for Massive Internet of Things, all listed in Table 1-1. A successful implementation of the PLL is necessary to fulfil KPIs such as user data rates, energy efficiency and low cost. The expected benefit of this work is to demonstrate an integrated phase controlled PLL for 5G mmWave beamforming transceivers at low cost and low power consumption.



4.3 RF Impairments Analysis

This analysis studies the RF impairments for RF transceivers for frequencies above 6 GHz and provides an overview of the limitations of key transceiver building blocks. At high frequencies it is necessary to create large antenna arrays to better control the propagation characteristics. The study will indicate how to overcome the limitations by solutions that should be dynamic, manufactured at low cost and deployed widely. To perform a concept study of large transceiver arrays, it is useful to start with an explanation of massive MIMO and beamforming. The advantage at higher frequencies is that antennas are smaller, enabling more antenna elements, and the beams gets narrower. Furthermore, at frequencies above 6 GHz as well as in mmWave range, path and penetration loss must be considered. Hence, it becomes important to steer the transmission accurately into the direction of the receiver. Beamforming is also known as spatial filtering, where steering a beam containing a digital or analogue signal along a well-chosen direction prevents interference. For this reason, beamforming is an attractive technique that relies on an independent and narrow beam, while suppressing mutual interference between multiple terminals. It is energy efficient, it saves power, and reduces interference. While beamforming is more user-specific, another concept called “massive-MIMO” develops the idea of transmitting to several users at the same time. At frequencies above 6 GHz the concept of massive MIMO enables the possibility to have systems that consist of antenna arrays with up to a few hundred antennas. These systems can simultaneously serve many tens of terminals within same time-frequency resource. This concept increases the capacity with an aggressive spatial multiplexing. More antennas increase the possible data rate that can be served simultaneously. This enhances capacity since there are more paths where signals can be propagated. Massive MIMO has the advantage that it can be built with cheap, low power components. This means it reduces accuracy and linearity requirements of each individual amplifier and RF chain, while it relies on the importance of the combined component performance to work. In early system generations in the implementation of antenna arrays, a hybrid capability solution may be proposed within different scenarios. At high frequencies, the channel does not exhibit rich scattering, hence the rank of the channel matrix is low. Herewith, it is not possible to obtain much gain in terms of capacity but still get gain in terms of capacity when serving multiple users. In a hybrid capability solution, for long range get the highest possible directivity relying on all antennas, whereas for short range, we address multiple users with limited MIMO capability. This hybrid solution would enable a trade-off between flexibility and performance. The requirement of higher sampling rates leads to higher energy consumption. Power consumption is a very critical aspect. Full digital beamforming receivers consume a large amount of power due to the ADCs used at each antenna element. Therefore, analogue beamforming is a good approach that enables steering of array elements. The antenna arrays are connected to power amplifiers or low noise amplifiers, mixers and phase shifters. The mixer is a key element for direct conversion but sensitive to imbalance problems between phase and quadrature branches. Another problem is the coupling of the two mixers requiring a phase shifter. Phase shifters are another source of imbalance and they can be implemented by power divider and all-pass filters producing 90 degrees of phase shift. The accuracy depends on the components such that the 90 degree phase shift is not guaranteed. There are different phased-array architectures which differ in functionality of phase shifting and signal combining and have as critical factors power consumption, losses, linearity and bandwidth.



Figure 4-2: RF and LO phase shifters.

RF phase-shifter solutions switch the phase in the LO or receive chain. In the receiver element the phase shifted signals are combined before arriving at the downconverter. This setup can phase out the interference and then relax power consumption on the blocks down the chain. Due to the signal insertion loss of the phase shifter it requires an RF gain stage prior signal combining for noise reasons. Additionally, it should meet tougher linearity specifications, which increases the losses of RF phase shifter. A way to cancel out the losses is to use an amplifier stage. Since the RF phase shifter is located at the signal path, the bandwidth of the RF signal aggregates challenge and demands higher power consumption to the RF phase shifter. LO phase shifters, compared to RF phase shifters, have an LO path, which is less sensitive to bandwidth because a single tone LO is delivered. The bandwidth only translates to a conversion gain difference in the mixer, when using different LO frequencies, and this can be compensated by increasing the gain in the LO driver. It has the advantage of very accurately rotating the phase for transmitter and receiver, but the penalty for this is the high power consumption. The topology of LO phase shifters have the drawback of a high number of mixers, since every element needs a mixer to down convert before phase shifting and combining. The advantage of the RF phase shifter topology over the LO phase shifter is that the output signal after the RF combiner has a high pattern directivity and can reject interference before it arrives at the receiver units, maximizing the value of the phased array as a spatial filter. Another advantage is the elimination of LO distribution network resulting in a simple system architecture. Nevertheless, the RF phase shifter concept suffers from an increase chip size. An example of both RF- and LO phase shifters can be seen in Error! Reference source not found.. In conclusion, this subsection analyses several trade-offs to be taken at high frequencies in a more detailed manner. It addresses the requirements to achieve KPIs (i.e., user data rate), as can be seen in Error! Reference source not found., and enlightens the limitations. It pictures the drawbacks and how to better utilize the advantage in use cases, such as Broadband Access in Dense Areas and Broadband Access Everywhere. Furthermore, distinguishing among several compensation techniques needed, while addressing the way to maintain low cost and improving performance.



4.4 5G Antenna Systems above 6 GHz

A compact planar antenna operating in the 20 to 40 GHz band will be designed, fabricated and tested. As this antenna needs to interface with a PA, the targeted frequency bands will be according to the PA specifications. An antenna topology is required that, on the one hand, facilitates easy interfacing with active electronics, such as a low-noise or a power amplifier and, on the other hand, provides sufficient isolation between the amplifier and the antenna radiation to avoid instability. Moreover, the developed active antenna element must be easily extendable to a compact array configuration. Therefore, we will again leverage substrate integrated waveguide technology to realize a cavity-backed slot antenna. The via walls of the cavity will act as an enclosure that shields the generated electromagnetic fields from their environment. As radiation will only occur through the slot on the top of the cavity, it is very easy to isolate the integrated amplifier from the antenna radiation by placing the component at the bottom of the cavity. A first design will be implemented on a thin low-cost high-frequency laminate. A trade-off will have to be made between substrate thickness, on the one hand, and substrate losses and antenna operation bandwidth, on the other hand. Next, we will study the feasibility of realizing this antenna on an ultrathin flexible substrate. We will determine the relationship between substrate thickness and operation bandwidth. We shall also compare the fabrication cost for both solutions. In simulations, we will study the integration of a power amplifier on both substrates. This passive antenna operating at mmWave frequencies will provide the necessary bandwidth which will allow the support of large numbers of connected users, high mobile data volume and large mobile data volume required for uses cases such as dynamic hot spots, smart cities and performance equipment. Owing to the small wavelength at mm-wave frequencies and the low-profile topology the antenna can combine compact dimensions (KPI: integration, size, footprint) with efficient operation (KPI: energy efficiency). 4.5 Integration of Power Amplifier with Antenna up to 28 GHz

Co-design of the amplifier with the antenna array is critical in order to maximize the obtained performance of the active antenna system. Specifically, the termination provided by the antenna array strongly affects the obtained PAE and gain, and additionally the input matching network of the amplifier. The challenge in jointly optimizing a large passive radiating structure such as an antenna array and a strongly nonlinear active circuit such as a power amplifier lies in the fact that electromagnetic simulation is a computationally expensive process especially due to the fact that the performance must not only be analysed at the fundamental operating frequency but that also the second and third harmonic frequencies need to be considered. Therefore a stepwise approach is considered. The design process will consist of several stages of different complexity, for example approximating the substrate integrated waveguide sections and the antenna by ideal circuit elements providing desired impedance at different frequencies, and progressively creating a more realistic model for simulation. The interconnection between the amplifier and the antenna array will be carefully designed in order to minimize loss, optimize amplifier gain, efficiency and at the same time allow integration of both the active and passive circuitry on the same substrate, potentially on different sides (faces) of the substrate in order to minimize the effect of the active circuitry on the antenna radiation pattern.

4.5.1 Flip-chip interconnect In flip chip technology, the chip is placed upside down and the connection between the chip and the laminate board is established by the stud bumps. Flip chip is preferred mainly owing to its superior performance in mmWave compared to wire bonding. First, the parasitic



inductance and the resistance are lower. Second, the RF performance varies less owing to a more robust mechanical structure. Third, the structure itself allows less transition discontinuities. The parameters involved are:

Bump diameter

Bump material

Height

Pad geometry (size and pitch)

Underfill material (dielectric constant)

Routing pitch on the laminate

Laminate material (dielectric constant, height) Such parameters determine the bandwidth and loss of the structure. The pad pitches and the transmission line pitches on the substrate may differ to allow stand-alone chip characterization by probing. Hence, a suitable transition may need to be designed for antenna feeding. 4.5.2 Antenna–PA co-optimization The approach we take in this project is first to conduct stand-alone measurements with 50 Ω characteristic impedance followed by co-optimization in the second iteration for the final proof-of-concept. This approach minimizes the risk and allows step-by-step debugging, so that we can identify potential problems which may occur. In the second iteration, a full-wave/circuit co-design shall be applied to an integrated antenna and power amplifier system operating in the frequency range from 20 to 40 GHz. This is because at the PA side, the optimum impedance for output power/gain is often not 50 Ω and the same is true for the antenna design. To demonstrate the benefits of the full-wave/circuit co-optimized active antenna, the integration of two different power amplifier designs shall be considered, one based on an integrated (IC) solution and one based on commercial-off-the-shelf components, as alternatives to produce cost effective yet high performance solutions. By co-designing the amplifier with the antenna, with specific focus on their interconnection, the dissipation losses can be minimized and a proper interfacing of the two circuits can be achieved for maximum energy efficiency. The full-wave/circuit co-design of this active antenna can demonstrate the benefits in terms of operation bandwidth and energy efficiency (in terms of transducer power gain and total gain) compared to a simple concatenation of a stand-alone antenna and a stand-alone power amplifier. 4.5.3 Integrated active transmit antenna vs. interconnected antenna and PA (20-40 GHz band) Different PA antenna systems, consist of an antenna realized in substrate integrated waveguide technology (resilient) and a PA realized in CMOS (KPI: Integration, size, footprint) or off-the-shelf components (KPI: cost), will be investigated. These proof-of-principles can be split up into two groups:

system composing of stand-alone components

co-designed PA and antenna For the first group, the design of separate components is done individually and the impedances at the interface are chosen to be 50 Ω. This results in a quicker design process and easy concatenation. The downside, however, is that the overall performance is suboptimal and that losses and parasitic influences (which cannot be ignored at mmWave frequencies) can further degrade performance.



To boost the complete system’s performance, components are no longer considered independently and a co-design relying on full-wave/circuit simulations is adopted. This omits the need to match individual designs to a fixed reference impedance and helps minimize parasitic side effects and interconnection losses. Since the system must be seen as an undividable entity, the subsystems can no longer be tested separately. Owing to the co-design advantages, the antenna-PA co-design addresses the use cases Smart Cities, Performance Equipment, and Connected Vehicles for Massive Internet of Things. The performance of these antenna PA combinations will be compared in function of bandwidth (KPI: bandwidth), integration/size/footprint and realized gain (KPI: energy).



5. Conclusions

The requirements on analogue hardware components aimed at 5G mobile systems target a significantly increase of capacity, throughput, and user data rate. A basic approach for that is the usage of substantially extended aggregated operation bandwidth. Already used or new expected radio bands below 6 GHz will provide up to about 600 MHz of total radio bandwidth for single operators in licensed and unlicensed bands after 2020. A further major increase of radio bandwidth by up to several GHz is expected to become available between 6 and 100 GHz. To exploit the total bandwidth expected below 6 GHz for increased throughput and data rate, concepts supporting simultaneous multiband operation are proposed, showing a certain level on flexibility, versatility and re-configurability. Aligned to approaches for multiband RF signal generation elaborated in WP3 (5G mixed-signal technologies), broadband or multiband concepts are considered for amplifiers, filters and antennas. As multiband operation shows an up scaling of signal carriers dedicated to individual radio bands, a corresponding up scaling of hardware transceiver chains shall be prevented by these approaches with advantages on integration, size and cost. The evaluation of key building blocks for mmWave transceivers is proposed for exploiting the radio bandwidth expected between 6 and 100 GHz, with the focus on frequency generation, amplifiers and antennas. The requirements on performance parameters of hardware components request for new architectures for on-chip frequency generation and for progressions in power amplifiers and antennas. A co-design of PAs and antennas by considering promising interconnection technologies for instance substrate integrated wave guides and amplifiers based on integrated solution as well as on commercial-off-the-shelf components, reduces power losses at interconnection level and component size. The analysis of RF impairments for transceivers will show the limitations of key functional blocks and allows for defining solutions to dynamically increase the performance with low manufacturing costs supporting the realization of large transceiver arrays. In addition to solutions supporting the operation in multiple and new radio bands, solutions supporting multi-antenna systems like MIMO and massive MIMO are considered for increasing the mobile data volume within a defined bandwidth by significantly increase of spectral efficiency. Cost efficient transceiver components of small footprint and high efficiency are required for such transceiver systems for integrating a high number of signal chains. This is supported by on-chip solutions for power amplifiers and frequency generation to be evaluated in new semiconductor technologies, for instance 28 nm CMOS bulk and SOI for millimetre wave frequencies and lower power level or GaN on Si for the lower frequency bands enabling high efficiency and high linearity at higher power level. The requirements for RF front ends and antennas define performance parameters for assessing the analogue hardware concepts by KPIs, which are aligned to the use cases defined in WP1 (5G Architecture requirements, specifications, and use cases). The presented concepts and solutions for analogue hardware components define key building blocks to be further evaluated for paving the way towards 5G mobile systems.



6. References

[Aga13] A. Agah, H.-T. Dabag, B. Hanafi, P.M. Asbeck, J.F.Buckwalter, L.E. Larson, "Active Millimeter-Wave Phase-Shift Doherty Power Amplifier in 45-nm SOI CMOS," in Solid-State Circuits, IEEE Journal of , vol.48, no.10, pp.2338-2350, Oct. 2013.

[Bat11] K. Bathich, et. al., “Frequency response analysis and bandwidth extension of the Doherty amplifier”, IEEE Trans. Microw. Theory Tech., vol. MTT-59, no. 4, pp. 934–944, Apr. 2011.

[Bou03] Boumaiza, Slim, and Fadhel M. Ghannouchi. "Thermal memory effects modeling and compensation in RF power amplifiers and predistortion linearizers." Microwave Theory and Techniques, IEEE Transactions on 51.12 (2003): 2427-2433.

[Boz11] M. Bozzi, A. Georgiadis, K. Wu, "Review of substrate-integrated waveguide circuits and antennas," in IET Microwaves, Antennas & Propagation, vol.5, no.8, pp.909-920, June 2011.

[Cai15] Y. Cai, Z. Qian, W. Cao, Y. Zhang, J. Jin, L. Yang, and N. Jing, “Compact Wideband SIW Horn Antenna Fed by Elevated-CPW Structure,” Antennas and Propagation, IEEE Transactions on, vol. 63, no. 10, pp. 4551–4557, Oct 2015.

[Cam12] C.F. Campbell, M.Y. Kao, S. Nayak, "High efficiency Ka-band power amplifier MMICs fabricated with a 0.15µm GaN on SiC HEMT process," in Proc. 2012 IEEE MTT-S Intl Microwave Symposium (IMS) pp.1-3, 17-22 June 2012

[Cha07] M. Chang, G.M. Rebeiz, "A 26 to 40GHz Wideband SiGe Balanced Power Amplifier IC," in Radio Frequency Integrated Circuits (RFIC) Symposium, 2007 IEEE , vol., no., pp.729-732, 3-5 June 2007.

[Che05] T.S.D. Cheung, J.R. Long, "A 21-26-GHz SiGe bipolar power amplifier MMIC," in Solid-State Circuits, IEEE Journal of , vol.40, no.12, pp.2583-2597, Dec. 2005.

[Che13] J.-H. Chen, S.R. Helmi, S. Mohammadi, "A fully-integrated Ka-band stacked power amplifier in 45nm CMOS SOI technology," in Silicon Monolithic Integrated Circuits in RF Systems (SiRF), 2013 IEEE 13th Topical Meeting on , vol., no., pp.75-77, 21-23 Jan. 2013.

[Chi11] K.-S. Chin, H.-T. Chang, J.-A. Liu, H.-C. Chiu, J. Fu, and S.-H. Chao, “28-GHz patch antenna arrays with PCB and LTCC substrates,” in Cross Strait Quad-Regional Radio Science and Wireless Technology Conference (CSQRWC), 2011, vol. 1, July 2011, pp. 355–358.

[Cri15] Cripps, S.C., "The evolution of the push-pull RFPA," in Microwave Symposium (IMS), 2015 IEEE MTT-S International , vol., no., pp.1-4, 17-22 May 2015

[Die14] A. Dierck, S. Agneessens, F. Declercq, B. Spinnewyn, G. Stockman, P. Van Torre, L. Vallozzi, D. Vande Ginste, J. Vanfleteren, T. Vervust and H. Rogier, “Active Textile Antennas in Professional Garments for Sensing, Localisation and Communication”, Int. Journal of Microwave and Wireless Technologies, DOI:10.1017/S175907871400018X, vol. 6, Special Issue 3-4, pp 331-341, Jun. 2014.



[Die15] A. Dierck, F. Declercq, T. Vervust, and H. Rogier, “Design of a circularly polarized Galileo E6-band textile antenna by dedicated multi-objective constrained Pareto-optimization”, International Journal of Antennas and Propagation, vol. 2015, Article ID 895963, 7 pages, 2015. doi:10.1155/2015/895963, 2015.

[ETS14] ETSI TS 136 104 V12.5.0 (2014-10), “LTE; Evolved Universal Terrestrial Radio Access E-UTRA); Base Station (BS) radio transmission and reception (3GPP TS 36.104 version 12.5.0 Release 12)”

[F5G11] Flex5Gware Project, “Deliverable 1.1: 5G system use cases, scenarios, and requirements break-down”, 2015.

[FCC13] Federal Communications Commission, “The Mobile Broadband Spectrum Challenge: International Comparisons,” FCC White Paper, February 26, 2013. https://apps.fcc.gov/edocs_public/attachmatch/DOC-318485A1.pdf

[Gre12] A. Grebennikov, et. al., “A Dual-Band Parallel Doherty Power Amplifier for Wireless Applications”, IEEE Trans. Microw. Theory Tech., vol. 60, no. 10, Oct. 2012.

[Gua15] D.-F. Guan, Z.-P. Qian, Y.-S. Zhang, and J. Jin, “High-Gain SIW Cavity-Backed Array Antenna With Wideband and Low Sidelobe Characteristics,” Antennas and Wireless Propagation Letters, IEEE, vol. 14, pp. 1774–1777, 2015.

[Hua12] P.-C. Huang, Z.-M. Tsai, K.-Y. Lin, H. Wang, "A 17–35 GHz Broadband, High Efficiency PHEMT Power Amplifier Using Synthesized Transformer Matching Technique," in IEEE Transactions on Microwave Theory and Techniques, vol. 60, no. 1, pp.112-119, Jan. 2012.

[Hun10] C.-C. Hung, J.-L. Kuo, K.-Y. Lin, H. Wang, "A 22.5-dB gain, 20.1-dBm output power K-band power amplifier in 0.18-µm CMOS," in Radio Frequency Integrated Circuits Symposium (RFIC), 2010 IEEE , vol., no., pp.557-560, 23-25 May 2010.

[Kaw11] Y. Kawano, A. Mineyama, T. Suzuki, M. Sato, T. Hirose, Tatsuya, K. Joshin, "A fully-integrated K-band CMOS power amplifier with Psat of 23.8 dBm and PAE of 25.1 %," in Radio Frequency Integrated Circuits Symposium (RFIC), 2011 IEEE , vol., no., pp.1-4, 5-7 June 2011.

[Kay15a] E. Kaymaksut; P. Reynaert, "Dual-Mode CMOS Doherty LTE Power Amplifier With Symmetric Hybrid Transformer," in Solid-State Circuits, IEEE Journal of , vol.50, no.9, pp.1974-1987, Sept. 2015

[Kay15b] E. Kaymaksut, D. Zhao, P. Reynaert, "Transformer-Based Doherty Power Amplifiers for mm-Wave Applications in 40-nm CMOS," in Microwave Theory and Techniques, IEEE Transactions on , vol.63, no.4, pp.1186-1192, April 2015.

[Kim11] D.-Y. Kim, W. Chung, C. Park, S. Lee, and S. Nam, “Design of a 45-Inclined SIW Resonant Series Slot Array Antenna for Ka-Band,” Antennas and Wireless Propagation Letters, IEEE, vol. 10, pp. 318–321, 2011.

[Kim14] K. Kim; C. Nguyen, "A 16.5–28 GHz 0.18- μm BiCMOS Power Amplifier With Flat 19.4±1.2 dBm Output Power," in Microwave and Wireless Components Letters, IEEE , vol.24, no.2, pp.108-110, Feb. 2014.

[Kom05] A. Komijani, A. Natarajan, A. Hajimiri, "A 24-GHz, +14.5-dBm fully integrated power amplifier in 0.18-μm CMOS," IEEE Journal of Solid-State

https://apps.fcc.gov/edocs_public/attachmatch/DOC-318485A1.pdf



Circuits, vol. 40, no. 9, pp. 1901-1908, Sep. 2005.

[Koo12] H. Koo; B. Koo; S. Hong, "Highly efficient 24-GHz CMOS linear power amplifier with an adaptive bias circuit," in Microwave Conference Proceedings (APMC), 2012 Asia-Pacific , vol., no., pp.7-9, 4-7 Dec. 2012.

[Kuo11] N.-C. Kuo, J.-C. Kao, C.C. Kuo, H. Wang, "K-band CMOS power amplifier with adaptive bias for enhancement in back-off efficiency," in Microwave Symposium Digest (MTT), 2011 IEEE MTT-S International , vol., no., pp.1-4, 5-10 June 2011.

[Kuo12] J.-L. Kuo; H. Wang, "A 24 GHz CMOS power amplifier using reversed body bias technique to improve linearity and power added efficiency," in Microwave Symposium Digest (MTT), 2012 IEEE MTT-S International , vol., no., pp.1-3, 17-22 June 2012.

[Kuo13] C.-W. Kuo; H.-K. Chiou; H.-Y. Chung, "An 18 to 33 GHz Fully-Integrated Darlington Power Amplifier With Guanella-Type Transmission-Line Transformers in 0.18 -µm CMOS Technology," in Microwave and Wireless Components Letters, IEEE , vol.23, no.12, pp.668-670, Dec. 2013.

[Lee10] J.-W. Lee; B.-S. Kim, "A K-Band High-Voltage Four-Way Series-Bias Cascode Power Amplifier in 0.13 μm CMOS," in Microwave and Wireless Components Letters, IEEE , vol.20, no.7, pp.408-410, July 2010.

[Liu09] B. Liu, W. Hong, Z. Kuai, X. Yin, G. Luo, J. Chen, H. Tang, and K. Wu, “Substrate Integrated Waveguide (SIW) Monopulse Slot Antenna Array,” Antennas and Propagation, IEEE Transactions on, vol. 57, no. 1, pp. 275–279, Jan 2009.

[Liu14] J.Y.-C. Liu, C.-T. Chan, S.S.H.Hsu , "A K-band power amplifier with adaptive bias in 90-nm CMOS," in European Microwave Integrated Circuit Conference (EuMIC), 2014 9th , vol., no., pp.432-435, 6-7 Oct. 2014.

[Maa03] S. A. Maas, Nonlinear microwave and RF circuits. Artech House, 2003.

[MET51] METIS D5.1: FP7-ICT-317669

[MET54] METIS D5.4: FP7-ICT-317669-METIS/D5.4

[Mor14] S.Y. Mortazavi, K.J. Koh, "14.4 A Class F-1/F 24-to-31GHz power amplifier with 40.7% peak PAE, 15dBm OP1dB, and 50mW Psat in 0.13μm SiGe BiCMOS," in Proc 2014 IEEE International Solid-State Circuits Conference (ISSCC),pp.254-255, 9-13 Feb. 2014.

[Mor15] X. Moronal, et. Al. “A 100 W Tri-Band LDMOS Integrated Doherty Amplifier for LTE-Advanced Applications”, in Microwave Symposium (IMS), 2015 IEEE MTT-S International , vol., no., pp.1-3, 17-22 May 2015

[Ora15] S. I. Orakwue, R. Ngah, T. Rahman, and H. M. Al-Khafaji, “A steerable 28 GHz array antenna using branch line coupler,” in Telematics and Future Generation Networks (TAFGEN), 2015 1st International Conference on, May 2015, pp. 76–78.

[Qur10] J. Qureshi, et. al., “A wideband 20 W LDMOS Doherty power amplifier”, IEEE MTT-S IMS Dig., pp. 1504–1507, June 2010.

[Qur14] J. Qureshi, et. al., “A 700-W Peak Ultra-Wideband Broadcast Doherty Amplifier”, IEEE MTT-S IMS, June 2014.

[Riz06] Rizzoli V, Costanzo A, Masotti D, Spadoni P, Neri A. Prediction of the end-to-end performance of a microwave/RF link by means of



nonlinear/electromagnetic cosimulation. IEEE Transactions on Microwave Theory and Techniques 2006; 54(12):4149–4160. DOI:10.1109/TMTT.2006.885566.

[Smi12] Smith, R.M.; Lees, J.; Tasker, P.J.; Benedikt, J.; Cripps, S.C., "A Novel Formulation for High Efficiency Modes in Push-Pull Power Amplifiers Using Transmission Line Baluns," in Microwave and Wireless Components Letters, IEEE , vol.22, no.5, pp.257-259, May 2012

[Son04] Wu, Songping, and Yeheskel Bar-Ness. "OFDM systems in the presence of phase noise: consequences and solutions." Communications, IEEE Transactions on 52.11 (2004): 1988-1996

[Sun12] G. Sun, et. al., “Broadband Doherty Power Amplifier via real frequency technique”, IEEE Trans. Microw. Theory Tech., vol. MTT-60, no. 1, pp. 99–111, Jan. 2012.

[Tan14] Y. Tang, Z. Wang, L. Xia, and P. Chen, “A novel high gain K-band H-plane SIW horn antenna using dielectric loading,” in Microwave Conference (APMC), 2014 Asia-Pacific, Nov 2014, pp. 372–374.

[Wri09] Wright, P.; Lees, J.; Tasker, P.J.; Benedikt, J.; Cripps, S.C., "An efficient, linear, broadband class-J-mode PA realised using RF waveform engineering," in Microwave Symposium Digest, 2009. MTT '09. IEEE MTT-S International , vol., no., pp.653-656, 7-12 June 2009



________________________________________________________________________

http://www.flex5gware.eu ________________________________________________________________________

· pdf fileproprietary rights statement this document contains information, which is...

Documents