spectral and energy efficiencies in mmwave cellular

Research ArticleSpectral and Energy Efficiencies in mmWave CellularNetworks for Optimal Utilization

Abdulbaset M Hamed and Raveendra K Rao

Faculty of Engineering Department of Electrical and Computer Engineering University of Western Ontario LondonON Canada N6A 5B9

Correspondence should be addressed to Raveendra K Rao rraouwoca

Received 29 September 2017 Revised 18 February 2018 Accepted 28 February 2018 Published 29 April 2018

Academic Editor Ahmed M Al-Samman

Copyright copy 2018 Abdulbaset M Hamed and Raveendra K Rao This is an open access article distributed under the CreativeCommons Attribution License which permits unrestricted use distribution and reproduction in any medium provided theoriginal work is properly cited

Millimeter wave (mmWave) spectrum has been proposed for use in commercial cellular networks to relieve the already severelycongested microwave spectrumThus the design of an efficient mmWave cellular network has gained considerable importance andhas to take into account regulations imposed by government agencies with regard to global warming and sustainable developmentIn this paper a densemmWave hexagonal cellular network with each cell consisting of a number of smaller cells with their own BaseStations (BSs) is presented as a solution tomeet the increasing demand for a variety of high data rate services and growing number ofusers of cellular networks Since spectrum and power are critical resources in the design of such a network a framework is presentedthat addresses efficient utilization of these resources in mmWave cellular networks in the 28 and 73GHz bands These bands arealready an integral part of well-known standards such as IEEE 802153c IEEE 80211ad and IEEE 802161 In the analysis a well-known accurate mmWave channel model for Line of Sight (LOS) and Non-Line of Sight (NLOS) links is usedThe cellular networkis analyzed in terms of spectral efficiency bits energy efficiency bitJ area spectral efficiency bitsm2 area energy efficiencybitJm2 and network latency sbit These efficiency metrics are illustrated using Monte Carlo simulation as a function of Signal-to-Noise Ratio (SNR) channel model parameters user distance from BS and BS transmission power The efficiency metrics foroptimum deployment of cellular networks in 28 and 73GHz bands are identified Results show that 73GHz band achieves betterspectrum efficiency and the 28GHz band is superior in terms of energy efficiency It is observed that while the latter band isexpedient for indoor networks the former band is appropriate for outdoor networks

1 Introduction

The traffic in cellular networks has exponentially increaseddue to high demand for existing and a variety of new wirelessservices and growing number of users of networksThereforeit has become important to utilize the available bandwidthefficiently Also it has becomemandatory due to governmentregulations on greenhouse gas emissions to pay careful atten-tion to design of networks that are energy efficient as wellThus a key issue in the design and implementation of nextgeneration of cellular networks is the joint optimization ofenergy and available spectral resources The mmWave spec-trum is known to offer tremendous bandwidth and can beutilized not only to meet the ever increasing demand for highdata rate services but also to accommodate a large numberof users This spectrum is also a serious contender for Ultra

High Definition Video (UHDV) transmission Several well-known network standards such as ECMA-387 IEEE 80211adand IEEE 802153c are being redefined to fit this spectrumDespite the massive bandwidth potential in the mmWaveband there exist a number of challenges (i) modelingand characterization of propagation environment (ii) designof antenna system (iii) transceiver integration (iv) digitalsignal processing technology

ThemmWave bands suffer from high path loss and hencetechniques are required to enhance power received at thereceiver The modeling of propagation environment plays animportant role in the design and analysis of communicationsystems required in cellular networks In [1] characteristicsof mmWave propagation environment are studied and itis observed that bands in this spectrum offer lower delay

HindawiWireless Communications and Mobile ComputingVolume 2018 Article ID 3097094 11 pageshttpsdoiorg10115520183097094

2 Wireless Communications and Mobile Computing

spread and it is possible to effectively suppress intersymbolinterference Wells [2] has shown that the 28 and 73GHzbands are the most natural choices for deployment in themmWave spectrum Also the 28GHz band is being activelypursued for Local Multipoint Distribution System (LMDS)and is very attractive for use in mmWave cellular networksgiven that this band occupies relatively lower range offrequencies within the mmWave spectrum

In recent years there have beenmany efforts to accuratelymodel themmWave channel for LOS andNLOS transmissionlinks The communication industry has devoted extensiveresources to the development of accurate channelmodels Forinstance Samsung conducted channel measurements in the28GHz band and came up with a model to fit ITU-R andFITU-R standards [3]This model can be used for a mmWavecommunication link over a distance of up to 200 meters Inaddition Nokia Huawei and Deutsch Telekom have shownkeen interest in the 73GHz band and have developed proto-type hardware using the concept of Multiple Input MultipleOutput (MIMO) system to support throughput of 20GbpsA channel model based on stochastic path loss has beenstudied for urban propagation environment in [4] A spatialstatistical model of mmWave channel has been developedas a function of channel parameters including path loss in[5] This model has been derived based on real measure-ments at New York City in 28 and 78GHz bands for LOSand NLOS links

A general framework has been proposed for evaluatingcoverage probability and rate performance in a mmWavecellular network in [6] The short wavelength of mmWavesignals can accommodate a very large number of antennasat the receiver and at the transmitter which makes beam-forming a key enabler technique in obtaining high antennagain Beamforming antenna systems formmWave bands havebeen studied for 5G networks and a novel hybrid beamform-ing algorithm has been proposed for indoor and outdoortransmission links [7] The design and characterization ofmmWave antennas for wireless links are discussed in [8 9]

Shannonrsquos capacity limits and signal processing chal-lenges for giga bit mmWave communication are discussedin [10] Also Madhow has addressed opportunities andchallenges of utilizing MIMO in LOS mmWave communi-cation systems An overview of Impulse Radio-Ultra WideBand (IR-UWB) technology for design of transmitters andreceivers is presented in [11] In [12] coverage and capacityof a dense mmWave network are analyzed using a theoreticalmodel as a function of blockage probability and beamformingbandwidth Also the spectral efficiency (SE) (bitss) of ammWave network is investigated with LOS and NLOS linksin 28 and 73GHz mmWave bands Since minimal powerconsumption has received significance in recent years due togovernment regulations it is important to address this issuefor mmWave cellular networks This issue has not receivedmuch attention in the literature The efficiency metrics forLOS and NLOS links are evaluated for 28 and 73GHz bandsin terms of maximum achievable throughput and maximumpower efficiency in [13] However the work considered onlypath loss to represent the channel effects for transmission dis-tances up to 200m Tan et al have investigated the achievable

downlink spectral efficiency of multiuser mmWave cellularsystem considering both small- and large-scale fading effectsin [14] In [15] energy coverage probability was derived formmWave transmission as a function of network densitybeamforming bandwidth and channel parameters

The trade-off between energy efficiency and spectralefficiency is examined for different access networks in [16]It is demonstrated that for best utilization of mmWave bandsit is important to understand how to trade tolerable delay forlow power The delay-power trade-off is examined for designof an energy efficient cellular network in [17] Levanen et alhave proposed technologies to reduce the delay in mmWavecommunication systems in [18] The challenges of designingenergy efficient mmWave systems are discussed in [19] Theenergy efficiency (EE) (bitJ) metric is presented and evalu-ated for 28 and 73GHz mmWave bands for LOS and NLOSlinksThe investigation is presented as a function of BS trans-mission power Signal-to-NoiseRatio (SNR) network param-eters and mmWave channel parameters The spatial char-acteristics of cellular networks play a significant role in theperformance of mmWave cellular networksThe area spectralefficiency (ASE) and the area energy efficiency (AEE) areimportant metrics that need to be considered in the design ofmmWave networks

Thus the intent of this paper is to examine energyefficiency (EE) spectral efficiency (SE) network latency areaspectral efficiency (ASE) and area energy efficiency (AEE) ofmmWave cellular network in 28 and 73GHz bands for LOSand NLOS links The investigation is presented as a functionof SNR channelmodel parameters network parameters userdistance from BS and BS transmission power This paper isorganized as follows In Section 2 mmWave cellular systemis introduced including cellular network model mmWavesignal propagation and wireless channel models and NR andSINR calculations Section 3 presents evaluation of networkperformance metrics spectral efficiency energy efficiencyspatial spectral and energy efficiencies network latency andoutage probability Also the evaluation of these efficienciesmetrics is discussed Simulation algorithm and the assump-tions that have been made for our analysis are listed inSection 4Moreover the network and channel parameters aretabulated and the numerical results with discussion are alsopresented Finally the work is concluded in Section 5

2 Millimeter Wave Cellular System

The model of mmWave cellular system is shown in Figure 1The macrocells are divided into small cells shown in thefigureThe 119869macrocells in the network are (119895 = 1 119869)usingthe same frequency and therefore create intercell interferenceEach of the 119895th cells comprises 119868 small cells each with its ownBS (119894 = 1 119868) and uses the same frequency thereby causingintracell interference It is assumed that each small cell hasone BS and it serves 119870 Mobile Users (MUs) (119896 = 1 119870)The BSs are located at the center of small cells and connectedto the gateway of the macrocell The MUs are spatiallydistributed and each one is associated with a single BS at anytime

Wireless Communications and Mobile Computing 3

Intercellinterference

Intracellinterference R

d

Figure 1 Model of mmWave cellular network in which the links ldquosdotndashsdotndashrdquo are for intracell interference ldquo rdquo for intercell interference and ldquomdashrdquofor transmission between BSs where the cochannel macrocells are remarked in blue colour

21 mmWaveWireless Channel Model ThemmWave signal issusceptible to high path loss fading noise and interferenceall these cause serious degradation in Signal-to-Noise Ratio(SNR) at the receiver leading to poor overall system perfor-mance In general mmWave communication channel can berepresented by the double directional channel model given in[20] and is given by

H = 119876sum119902=1

119862sum119888=1

119867119902119888119890120601119902119888120575 (120591 minus 120591119902119888) 120575 (120601119877 minus 120601AoA119902119888)sdot 120575 (120601119879 minus 120601AoD119902119888) 120575 (120579119877 minus 120579ZoA119902119888) 120575 (120579119879 minus 120579ZoD119902119888)

(1)

where 119876 is the number of multipath components 119862 is thenumber of rays within the cluster Each ray is represented bypath gain119867119902119888 phase 120601119902119888 time-of-arrival (ToA) 120591119902119888 azimuthangle-of-arrival (AoA) 120601AoA119902119888 azimuth angle-of-departure(AoD) 120579AoD119902119888 zenith angle-of-arrival (ZoA) 120579ZoA119902119888 andzenith angle-of-departure (ZoD) 120579ZoD119902119888 The task of channelmodeling is thus to find all these parameters for a mmWavecommunication channel For 28 and 73GHz frequencybands a model has been obtained to fit the LOS and NLOStransmission links These bands were selected since they arethe bands likely to be deployed in mmWave cellular networkAccording to this model [4] small-scale fading has a lessimpact on mmWave signal propagation and hence the large-scale fading effect is considered in measurements In [5]

mmWave channel path loss effects are considered and astatisticalmodel based on realisticmeasurements is presentedfor an urban environment LOS and NLOS links and is givenby119871119873LOS [dB] = 120572 + 12057310 log10 (119903) + 120585 [dB]

120585 sim N (0 1205902119904 ) (2)

where 119871119873LOS is the path loss in dB 119903 is the distance betweentransmitter and receiver inmeters120572 and120573 are the least squarefits of floating intercept and slope over themeasured distancesup to 200m and 1205902119904 is the variance of the lognormal shadow-ing 120585 In this paper the channelmodel given by (2) [5] is usedand the values of model parameters 120572 120573 and 1205902119904 are given inTable 1

22 Signal-to-Noise Ratio Calculation The path loss modelgiven by (2) is used to determine the received signal power atthe MU and then used to estimate SNR of the 119896th MU in the119894th BS That is

SNR119894119896[dB]

= 10 log10 1003816100381610038161003816ℎ11989411989610038161003816100381610038162 + 119875119894[dBm] + 119866119894[dBi] + 119866119896[dBi]minus 119871 119894119896[dB] (119903119894119896)minus (KT[dBmHz] + 10 log10 (119882119898119906) + NF[dB])

(3)


Table 1 Parameters of statistical channel model [5]

Path loss parameter 120572 120573 1205902119904 V28GHz

NLOS 72 292 87 1LOS 614 2 58 5

73GHzNLOS 866 245 80 1LOS 698 2 58 8

where 119875119894 is the BS119894 transmission power KT is the noise powerdensity at theMU and NF is the noise figureThe transmitterand receiver antenna gains 119866119894 and 119866119896 are calculated using119866 = 20log10(120587119897120582) for the 28 and 73GHz frequency bandsand antenna length of 119897 Assuming an independent Nakagamifading for each link the small-scale fading component |ℎ119894119896|2can be modeled as a normalized Gamma random variablethat is |ℎ119894119896|2 sim G(V 1V) [15] The small-scale fading isconsidered less severe in mmWave band [12] and hence V isvery large integer for LOS link and V = 1 when the link isNLOS The path loss component 119871 119894119896 can be calculated using(2)

23 Signal-to-Interference-Plus-Noise Ratio (SINR) ModelBased on the network model shown in Figure 1 for the119896th MU in the 119894th small cell serving BS in the 119895th macrocell the interference consists of two components which areintracell interference and intercell interference The first oneis caused by the active BSs located in the same 119895th cell whichis more severe since it is close to the MU While the secondcomponent is generated from the BSs in the other macro cellswhich is less intense When the cochannel interference dueto first tier macrocell is considered the downlink aggregatedinterference in the 119904 small cell BS and 119901macro cell for the 119896thMU can be expressed as

Λ 119895119894119896 = 119868sum119894=1119894 =119904

119875119901119894119866119901119894119866119901119904119896 10038161003816100381610038161003816ℎ119901119894119896100381610038161003816100381610038162 119871119901119894119896 (119903119894119896)

+ 119869sum119895=1119895 =119901

119868sum119894=1

119875119895119894119866119895119894119866119895119894119896 10038161003816100381610038161003816ℎ119895119894119896100381610038161003816100381610038162 119871119895119894119896 (119903119895119894119896) (4)

Therefore the SINR at the MU119896 for BS119894 is given by

SINR119895119894119896[dB] = SNR119894119896[dB] minus 10 log10 (Λ 1198941198961205902119896

+ 1) (5)

where 1205902119896 is the thermal noise power at the 119896th MU Thedistance between BS and MU (3) and (4) is lt200m andis a random variable with probability distribution function119891119889(119903) = 211990311988923 Energy and Spectral Efficiency ofmmWave Network

Energy and spectrum are system resources that have tobe appropriately traded to efficiently design a mmWave

communication system Minimal use of energy resources inthe systems has become an important issue however efficientuse of mmWave spectrum requires more power due to highpath loss Therefore a balance between the use of energy andspectral resources of the system deserves a careful studyThisis examined in this section for a mmWave cellular networkAnalysis of energy and spectral efficiency metrics for suchnetwork is considered which can play an important role inthe standardization of mmWave cellular network

31 Network Spectral Efficiency The downlink spectral effi-ciency quantifies the amount of data rate delivered by the BStoMUs over a certain bandwidth119882119896This indicates how effi-ciently the mmWave spectrum is utilized Spectral efficiencyis bounded by Shannonrsquos limit 119882119896log2(1 + SNR119896) [21] for agiven SNR For the case of mmWave network SNR is replacedby SINR Thus spectral efficiency of the network depictedin Figure 1 can be evaluated using

120578SE = 119869sum 119868sum 119870sum119882119896E119903 [Eℎ [log2 (1 + SINR119895119894119896)]]sdot (1 minus 119875out119895 (119877)) (bits)

(6)

where SINR can be obtained from (5) and 119882119896 is theallocated mmWave bandwidth for the MUs E119903 and Eℎ arethe expectation values taken for over MUrsquos random locationand the small-scale fading effect respectively The outageprobability within a communication range 119877 119875out119895 is givenby [22]

119875out119895 (119877) = exp(minus21205871205881198951205742119895 (1 minus (1 + 120574119895119877) 119890minus120574119895119877)) (7)

where 120574 is the parameter that depends on the propagationenvironment density and 120588 is the BS density The outageprobability is plotted as a function of 120574 and 119877 for a given 120588 =10minus4 in Figure 2 It is observed that 119875out is nearly constant for119877 gt 300m Since the proposed network dimensions are largewith macro cell radius 119877 ⩾ 500m the outage probability canbe approximated by 119875out119895 = 119890minus21205871205881198951205742119895 32 Network Energy Efficiency The energy efficiency isdefined as a ratio between what the system delivers to what itconsumes and hence the efficiency function can be describedas 119891(119909)119909 where 119909 isin [0 119883] indicates the systemrsquos resourcesconstrained by 119883 [23] In this paper 119891(119909) represents the


Table 2 Parameters used in the simulation of cellular network

Symbol Descriptions Value119882119896 User bandwidth 01GHz119875119894 BS transmission power minus10 dBmndash40 dBmKT Noise power density minus174 dBmHzNF Noise figure 6 dB119889 Maximum BS serving distance 100m119877 Macro cell radius 500m119869 Maximum number of cells sharing the spectrum 7119868 Maximum number of BSs in the macro cell 119906119899119894119891119900119903119898 sim 1120588119870 Number of simultaneous users in BSs 20120588 BS density (119889119877)2120574 Propagation environment density 10minus2119897119905 Transmitter antenna length 01m119897119903 Receiver antenna length 0005m

02

0005

04

001

Out

age p

roba

bilit

y

06

4003000015

08

200100002

02040608

R (m)

Figure 2 Outage probability as a function of 120574 and 119877 for 120588 = 10minus4

overall downlink data rate efficiency that can be reliablytransmitted by a mmWave communication system and 119909denotes the Signal-to-Noise Ratio which is a function ofBSrsquos transmission power noise and channel effects For themmWave network the energy efficiency can be formulatedas

120578EE = 119869sum 119868sum 119870sum119882119896E119903 [Eℎ [ log2 (1 + SINR119895119894119896)SNR119895119894119896

]]sdot (1 minus 119875out119895 (119877)) (bitJ)

(8)

where SNR and SINR can be determined using (3) and(5) respectively The energy efficiency indicates the amountof data rate that can be delivered per unit of SNR perHertz of bandwidth The efficiency metrics are useful in thestandardization of mmWave cellular networks as spectralefficiency provides the maximum rate for each frequencyband and transmission range and energy efficiency givesinsight into how to utilize the energy resources in cells as afunction of data rate

33 Spatial Spectral and Energy Efficiency Spatial character-istics ofmmWave network are crucial in the evaluation of sys-tem performance For example the propagation loss is highlyaffected by network dimensions and hence SNR and SINRare also impactedTherefore studying the spectral and energyefficiencies is not enough for efficient design of mmWavenetworks The spatial spectral and energy efficiencies arebetter metrics for design of networks [24] Accordingly thesemetrics are examined for mmWave networks These metricsare evaluated for macrocell as a sum of each small cell spatialefficiency in which the serving distance 119903 determines thecellsrsquo area

4 Numerical Results

Monte Carlo simulations have been performed in MATLABenvironment to evaluate spectral energy and spatial spectraland energy efficiencies for the 28 and 73GHz mmWavebands The efficiency metrics are investigated in terms of BStransmission power BS serving distance and SNRThe stepsused in the simulation are given below

(1) The path loss for 28 and 73GHz bands is calculatedfor LOS and NLOS using (2) for each value of 119875 and119903 and the model parameters are given in Table 1

(2) SNR is computed using (3) in which small-scalefading is modelled using Gamma distribution |ℎ|2 simG(V 1V) and (3) parameters are given in Table 2

(3) For the network model Figure 1 the aggregatedintercell and intracell interferences are obtained using(4)

(4) The SINR is determined (5) for 1205902119896 = KT119882119896(5) Spectral and energy efficiency metrics are computed

using (6) and (8) with 119875out calculated using (7) seeFigure 2

(6) The spatial spectral and energy efficiency metrics arecomputed using sum(120578SE119895119894119896119860119895119894) and sum(120578EE119895119894119896119860119895119894)


respectively with 119860119895119894 = 1205871198892119895119894 being the coverage areaof each small cell BS [25]

The following assumptions are made in Monte Carlo simula-tion

(1) Channel fading coefficients ℎ119894119895119896 in (3) and (4)are independent and identical distributed randomvariables

(2) Transmission power119875119894 in (3) is identical for all smallcell BSs

(3) Antenna gain119866119894 is equal for all BSs that are assumedto be same and antenna gain 119866119896 for all MUs isassumed to be identical

(4) Intercell interference is based on only the first tier ofthe macrocells

(5) Intercell distance 119903119894119895119896 = 3119877 + 119889 and the intracelldistance 119903119894119896 = 2119889 represent the worst MUs in smallcells

(6) Macro cell radius of 119877 = 500m is assumed and themaximum serving distance of a small cell BS is set to119889 = 100m

The cellular network parameters used in the simulations aregiven in Table 2 Simulation and numerical process of thiswork are summarized in the flow chart shown in Figure 3

41 Spectral Efficiency The spectral efficiency of the networkis evaluated in terms of achievable data rate for a specificallocated bandwidth to MU This efficiency metric (bits)is plotted in Figure 4 as a function of maximum servingdistances for LOS and NLOS links in 28 and 73GHz bandswith transmission power fixed at 10 dBm It is observed thatthe maximum achievable data rate in the network can reach1Tbits as 119889 is diminished In this case the macrocell willhave a large number of small cells and intracell interferenceis increased Also it is noted that the spectral efficiencyslightly deteriorates for LOS link while for NLOS link itdecreases rapidly to 015 G bits for 119889 = 100m The spectralefficiency is also investigated as a function of BS power usedfor transmission for a specific value of serving distance 119889 =45m Figure 5 demonstrates that the spectral efficiency isenhanced as power transmitted by the BS is increased Itis observed that there is no significant improvement in thespectral efficiency when the transmission power is above40 dBm This indicates that pumping more power is waste ofBS power resource and hence impacts the energy efficiencyThus for optimum mmWave band utilization transmissionpower over 40 dBm is not required The figure also illustratesthe superiority of high frequency band 73GHz for the LOSlink

For high power transmission the achieved spectral effi-ciencies of the two transmission bands are above 1TbitsSince the SNR at MUs is a function of the BS transmissionpower receiver noise and channel effects the spectral effi-ciency for the 73GHz band for LOS link is presented inFigure 6 as a function of serving distance and SNR It is notedthat the spectral efficiency is an increasing function of SNR

which indicates the mmWave system can overcome the noiseand channel impairmentsThe spectral efficiency ofmmWavecellular networks has been investigated in [14 26] and thenumerical results presented in these works are consistentwith the results illustrated in Figure 5 However the spectralefficiency results of this paper are based on real channelmeasurements

42 Energy Efficiency Energy efficiency metric is a ldquogreencommunicationrdquo indicator since the transmitted power bythe BS plays an important role in its determination Theenergy efficiency metric is a function of the data rate overthe mmWave links and SNR at MUs The energy efficiency isexamined as a function of BS serving distance the transmis-sion power and the SNR The metric is plotted as a functionof transmission power for two mmWave bands and links inFigure 7 It is observed that the 73GHz band is less energyefficient compared to the 28GHz bandThe energy efficiencyis degraded as the transmission power is increased Figure 8shows a comparison of energy efficiency between the twofrequency bands for LOS transmission link as a function of 119889and SNR It is observed that 28GHz band Figure 8(b) attainshigher energy efficiency than the 73GHz band Figure 8(a)In fact the SNR for LOS transmission at 73GHz band isgreater than the SNR forNLOS link at the 28GHz band Alsoit is noted that the efficiency improves up to certain servingdistance 119889 = 30m and then becomes almost constantfor 119889 gt 30m It is noted that trade-off between servingdistance and the transmission power has to be consideredin dimensioning of the network and for optimum utilizationfor the two mmWave bands Energy efficiency of mmWavecellular network has been examined in terms of number ofusers and BS antenna size for full-duplex relaying backhaulsmall cell mmWave networks in [19] for 28GHz and 60GHzbands The results show an increasing function of the trans-mitted powerwhich contradicts the energy efficiency conceptillustrated in Figure 7 Also this metric has been investigatedin terms of propagation environment parameters and BS den-sity for heterogeneous mmWave network overlaid by smallcell BSs in [22] The results are consistent with the resultsillustrated in Figure 8

43 Network Latency Delay is critical metric for real timeapplications in mmWave networks Thus it is importantto address the network delay as a function of networkparameters The delay in mmWave networks is due to severalcomponents such as radio access network backhaul corenetwork and data centers In this paper we consider thetransmission delay between MU and BS For a given userbandwidth and information quantity 119876119896 the transmissiondelay119863 is given by119863 = 119876119896119862 [27] For the network shown inFigure 1 the delay-per-bit per user can be expressed as119863119896

= 119876119896119882119896E119903 [Eℎ [log2 (1 + SINR119895119894119896)]] (1 minus 119875out119895 (119877)) (9)

where 119876119896 = 1 and SINR119895119894119896 is given by (5) The trade-off among latency and efficiency metrics of the mmWave


Problem formulationOptimum utilization of mmWave bands

MmWave cellularnetwork Figure 1

Propose

Select

The propagationenvironment

Model

Path loss model(2)

Fading effectGamma model

ModelingSNR (3)SINR (5)

Models parameters Table 1

Network parameters Table 2

Use

Assumptions forevaluation

Consider

Outage probability(7)

Evaluate

Spectral efficiency(6)

Energy efficiency(8)

EvaluateEvaluate

Area spectralefficiency (10)

Area energyefficiency (11)

Spatialproperties

Calculate

Evaluate

Spatialproperties

Network latency(9)

Trade-off amongThese metrics for optimum utilization

End

If optimization done

Yes

No Change networkparametersassumptions

Frequency bands

transmission links LOS and NLOS

28 GHz 73GHz and

Figure 3 Flow chart used for evaluation of trade-off between spectral and energy efficiencies for optimum utilization in mmWave network

8 Wireless Communications and Mobile ComputingSE

(bit

s)

F = 28 GHz LOSF = 28 GHz NLOS

1013

1012

1011

1010

109

108

10 20 30 40 50 60 70 80 90 100

d (m)

F = 73GHz LOSF = 73GHz NLOS

Figure 4 Spectral efficiency of 28GHz and 73GHzmmWave bandsfor NLOS and LOS links as a function of the maximum servingdistance

SE (b

its)

1013

1012

1011

1010

109

108

107

Pt (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40


Figure 5 Spectral efficiency of 28GHz and 73GHzmmWave bandsfor NLOS and LOS links as a function of transmission power

2040 302060

SNR (dB)1080 0100

051152251012

1011

1010

d (m)minus10

times1012

(bit

s)Sp

ectr

al effi

cien

cy

Figure 6 Spectral efficiency of 73GHz band LOS link as a functionof 119889 and SNR

Ener

gy effi

cien

cy (b

itJ)

P (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40


1012

1011

1010

109

108

107

Figure 7 Energy efficiency of 28GHz and 73GHz mmWave bandsfor NLOS and LOS links as a function of BS power

network can be discussed in terms of delay-per-bit and energyefficiency Delay-per-bit is obtained by simulating (9) usingchannel parameters given in Table 1 and network parametersgiven in Table 2 The delay is plotted as a function of BSpower and user distance in Figures 9(a) and 9(b) respectivelyThese figures show that 73GHz band for LOS link achievesless latency than the 28GHz band for NLOS link Figure 9(a)shows that the delay decreases as the BS power increases Forexample the network latency is less than 1 nsbit when theBS power is greater than 20 dBm The energy efficiency ishighly impacted for such large values of BS power as shownin Figure 7 In terms of user distance the network latencyincreases as the distance increases as is clear fromFigure 9(b)In contrast the spectral efficiency is reduced as a functionof user distance which indicates that the two metrics are inconflictTherefore trade-offs among network latency energyefficiency and spectral efficiency metrics must be consideredfor optimum utilization of the mmWave bands

44 Area Spectral Efficiency The area spectral efficiency(ASE) is defined as the ratio of the average spectral efficiencyto the utilized area for a given frequency band For thenetwork in Figure 1 The ASE is calculated as the sum of thespectral efficiencies of each small cell divided by its servingarea and is given by

sum 120578SE119895119894119896119860119895119894 (10)

This ASE metric links the network spectral efficiency and theservice area and is investigated as a function of119875 119889 and SNRIn Figure 10(a) ASE is plotted in terms of 119889 It is noted thatASE is a decreasing function of the distance since the area isproportional to 1198892 Also ASE is plotted in terms of 119889 and SNRin Figure 10(b) which depicts how both factors affect the ASEIt is clear that for large coverage area the network requiresmore BS power to achieve acceptable ASE


20

Ener

gy effi

cien

cy (b

itJ)

40 302060

SNR (dB)1080 0100

24681012

1010

108

106

d (m) minus10

times1010

(a) 73GHz bands for LOS

2468101214

times1010

20

Ener

gy effi

cien

cy (b

itJ)

40 2060

SNR (dB)10

80 0100

1010

108

106

d (m) minus20minus10

(b) 28GHz bands for LOS

Figure 8 Energy efficiency of 73 and 28GHz bands for LOS link as a function of 119889 and SNR

10minus7

10minus8

10minus9

10minus10

Db

(sb

it)

minus10 0 10 20 30 40

Pt (dBm)



(a) Delay-per-bit as a function of transmitted power 119901119905

10minus8

10minus9

10minus10

Db

(sb

it)

10 20 30 40 50 60 70 80 90 100

d (m)



(b) Delay-per-bit as a function of user distance 119889

Figure 9 Delay-per-bit for 73 and 28GHz bands for LOS and NLOS links

45 Area Energy Efficiency (AEE) The AEE is used to char-acterize the area data rate per unit BS power expenditure andcan be defined as

AEE = sum 120578EE119895119894119896119860119895119894 (11)

where 120578EE119895119894119896 is given by (8) and 119860119895119894 = 1205871198892119895119894 This metricprovides insight into the deployment and implementationchallenges for optimum mmWave band utilization The AEEis evaluated as a function of BS power and is shown inFigure 11(b) The metric severely deteriorates for high valuesof BS power however it maintains a certain level of perfor-mance at lower values of power Also it is observed that AEEslightly decreases as the radius of small cell increases as shownin Figure 11(a)

From the above results it is clear that there is alwaysa conflict between spectral efficiency and energy efficiency

in terms of power transmission The optimum utilization ofmmWave bands requires trade-off between these two per-formance metrics The 28GHz band achieves higher energyefficiency for a wide range of power transmission level whileit provides acceptable spectral efficiency for 119875119905 gt 20 dBm Onthe other hand the 73GHz is superior in terms of spectralefficiency however it performs poorly with respect to energyefficiency Regarding the transmission environment it canbe said that the 28GHz band is more advantageous thanthe 73GHz band for NLOS links such as indoor networksHowever the 73GHz bands are more suitable for LOS linksuch as outdoor mmWave links

5 Conclusions

In this paper mmWave cellular network is considered forindoor and outdoor applications in which the macrocells are


1014

1012

1010

108

106

104

AEE

(bit

Joul

eＧ

2)

0 10 20 30 40 50 60 70 80 90 100

d (m)



(a) ASE of 73 and 28GHz bands for LOS and NLOS links as afunction of BS serving distance

202040

10

SNR (dB)60 080 minus10100 minus20

1234567

d (m)

1010

108

106ASE

(bit

sＧ

2)

times1011

(b) ASE of 73GHz band LOS as function of d and SNR

Figure 10 Area spectral efficiency of mmWave bands

EE (b

itJo

ule)

1014

1012

2040

6080

100d (m) minus10

010

2030

40

Pt (dBm)

98765

34

21

times1015

(a) AEE 73GHZ band LOS

Pt (dBm)

1012

1011

1010

109

108

107

AEE

(bit

Joul

eＧ

2)

minus10 minus5 0 5 10 15 20 25 30 35 40



(b) AEE of 73 and 28GHZ bands for LOS and NLOS

Figure 11 Area energy efficiency of mmWave bands as a function of 119889 and 119875

overlaid with multiple small cells A channel model basedon real measurements available in literature for LOS andNLOS transmission links was used to derive expressionsfor SNR and SINR in network considering intercell andintracell interference components The paper presented aframework for analysis of spectral and energy efficiencymetrics of the network for optimum utilization of networkresources The framework also included an investigation ofspectral efficiency energy efficiency and spatial performancemetricsThe analysis was done usingMonte Carlo simulationconsidering the effects of interference the path loss andsmall-scale fading The investigation of metrics has been

carried out as a function of BS power SNRmmWave networkparameters and channel model parametersThe results showthat the spectral efficiency is always in conflict with theenergy efficiency with regard to BS power and the maximumBS serving distance The 28GHz band can be expedient forNLOS links in the network and the 73GHz band is moreappropriate for LOS links

Conflicts of Interest

The authors declare that they have no conflicts of interest


Acknowledgments

The first author would like to acknowledge University ofTripoli Libya and the LibyanMinistry of Education for theirsupport and scholarship

References

[1] Y Li P Wang F Chen G Wang and Y Liu ldquoSimulationandAnalysis ofMillimeter-Wave PropagationCharacteristics inComplex Office Environmentrdquo Journal of Computer and Com-munications vol 03 no 03 pp 56ndash60 2015

[2] J Wells ldquoFaster than fiber The future of multi-Gs wirelessrdquoIEEE Microwave Magazine vol 10 no 3 pp 104ndash112 2009

[3] J Ko Y-S Noh Y-C Kim et al ldquo28 GHz millimeter-wavemeasurements and models for signal attenuation in vegetatedareasrdquo in Proceedings of the 11th European Conference onAntennas and Propagation EUCAP 2017 pp 1808ndash1812 FranceMarch 2017

[4] T Bai V Desai and R W Heath Jr ldquoMillimeter wave cellularchannel models for system evaluationrdquo in Proceedings of the2014 International Conference on Computing Networking andCommunications ICNC 2014 pp 178ndash182 USA February 2014

[5] M R Akdeniz Y Liu M K Samimi et al ldquoMillimeter wavechannel modeling and cellular capacity evaluationrdquo IEEE Jour-nal on SelectedAreas in Communications vol 32 no 6 pp 1164ndash1179 2014

[6] L Zhao J Cai and H Zhang ldquoRadio-efficient adaptive mod-ulation and coding Green communication perspectiverdquo inProceedings of the 2011 IEEE 73rd Vehicular Technology Confer-ence VTC2011-Spring Hungary May 2011

[7] W Roh J-Y Seol J Park et al ldquoMillimeter-wave beamformingas an enabling technology for 5G cellular communications the-oretical feasibility and prototype resultsrdquo IEEECommunicationsMagazine vol 52 no 2 pp 106ndash113 2014

[8] X Wu G V Eleftheriades and T E Van Deventer-PerkinsldquoDesign and characterization of single- and multiple-beamMM-wave circularly polarized substrate lens antennas for wire-less communicationsrdquo IEEE Transactions on Microwave Theoryand Techniques vol 49 no 3 pp 431ndash441 2001

[9] Y P Zhang and D Liu ldquoAntenna-on-chip and antenna-in-package solutions to highly integrated millimeter-wave devicesfor wireless communicationsrdquo IEEE Transactions on Antennasand Propagation vol 57 no 10 pp 2830ndash2841 2009

[10] U Madhow ldquoMultiGigabit millimeter wave communicationSystem concepts and challengesrdquo in Proceedings of the 2008Information Theory and Applications Workshop - ITA pp 193ndash196 USA February 2008

[11] C Stallo E Cianca S Mukherjee T Rossi M de Sanctisand M Ruggieri ldquoUWB for multi-gigabits communicationsbeyond 60GHzrdquo Telecommunication Systems vol 52 no 1 pp161ndash181 2013

[12] T Bai A Alkhateeb and R W Heath ldquoCoverage and capacityof millimeter-wave cellular networksrdquo IEEE CommunicationsMagazine vol 52 no 9 pp 70ndash77 2014

[13] A M Hamed and R K Rao ldquoEvaluation of capacity andpower efficiency in millimeter-wave bandsrdquo in Proceedings ofthe 2016 International Symposium on Performance Evaluation ofComputer and Telecommunication Systems SPECTS 2016Canada July 2016

[14] W Tan P J Smith H A Suraweera M Matthaiou and SJin ldquoSpectral efficiency of multi-user mmwave systems with

uniform linear arrays andMRTrdquo in Proceedings of the 83rd IEEEVehicular Technology Conference VTC Spring 2016 China May2016

[15] T A Khan A Alkhateeb and RW Heath ldquoEnergy coverage inmillimeter wave energy harvesting networksrdquo in Proceedings ofthe IEEE GlobecomWorkshops GCWkshps 2015 USA Decem-ber 2015

[16] A Mesodiakaki F Adelantado A Antonopoulos L Alonsoand C Verikoukis ldquoEnergy and spectrum efficient user associ-ation in 5G heterogeneous networksrdquo in Proceedings of the 27thIEEE Annual International Symposium on Personal Indoor andMobile Radio Communications PIMRC 2016 Spain September2016

[17] Y Chen S Zhang S Xu and G Y Li ldquoFundamental trade-offson green wireless networksrdquo IEEE Communications Magazinevol 49 no 6 pp 30ndash37 2011

[18] T Levanen J Pirskanen and M Valkama ldquoRadio interfacedesign for ultra-low latency millimeter-wave communicationsin 5GErardquo in Proceedings of the 2014 IEEEGlobecomWorkshopsGC Wkshps 2014 pp 1420ndash1426 USA December 2014

[19] Z Wei X Zhu S Sun Y Huang A Al-Tahmeesschi and YJiang ldquoEnergy-Efficiency of Millimeter-Wave Full-DuplexRelaying Systems Challenges and Solutionsrdquo IEEE Access vol4 pp 4848ndash4860 2016

[20] M Steinbauer A F Molisch and E Bonek ldquoThe double-directional radio channelrdquo IEEE Antennas and PropagationMagazine vol 43 no 4 pp 51ndash63 2001

[21] C E Shannon ldquoA mathematical theory of communicationrdquoBibliometrics vol 5 no 1 pp 3ndash55 2001

[22] J Choi ldquoEnergy efficiency of a heterogeneous network usingmillimeter-wave small-cell base stationsrdquo in Proceedings of the26th IEEE Annual International Symposium on Personal Indoorand Mobile Radio Communications PIMRC 2015 pp 293ndash297China September 2015

[23] A M Hamed and R K Rao ldquoEnergy and spectral efficiencyin cellular networks considering fading path loss and inter-ferencerdquo in Proceedings of the 30th IEEE Canadian Conferenceon Electrical and Computer Engineering CCECE 2017 CanadaMay 2017

[24] H Pervaiz L Musavian and Q Ni ldquoArea energy and areaspectrum efficiency trade-off in 5G heterogeneous networksrdquoin Proceedings of the IEEE International Conference on Commu-nication Workshop ICCW 2015 pp 1178ndash1183 UK June 2015

[25] M-S Alouini and A J Goldsmith ldquoArea spectral efficiency ofcellular mobile radio systemsrdquo IEEE Transactions on VehicularTechnology vol 48 no 4 pp 1047ndash1066 1999

[26] A Mesodiakaki F Adelantado L Alonso M Di Renzo andC Verikoukis ldquoEnergy-and Spectrum-Efficient User Associa-tion in Millimeter-Wave Backhaul Small-Cell Networksrdquo IEEETransactions on Vehicular Technology vol 66 no 2 pp 1810ndash1821 2017

[27] J-M Gorce R Zhang K Jaffres-Runser and C GoursaudldquoEnergy latency and capacity trade-offs in wireless multi-hopnetworksrdquo in Proceedings of the 2010 IEEE 21st InternationalSymposium on Personal Indoor and Mobile Radio Communica-tions PIMRC 2010 pp 2757ndash2762 Turkey September 2010

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018


Active and Passive Electronic Components

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of



Journal ofEngineeringVolume 2018

SensorsJournal of



RotatingMachinery


Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation


Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom


spread and it is possible to effectively suppress intersymbolinterference Wells [2] has shown that the 28 and 73GHzbands are the most natural choices for deployment in themmWave spectrum Also the 28GHz band is being activelypursued for Local Multipoint Distribution System (LMDS)and is very attractive for use in mmWave cellular networksgiven that this band occupies relatively lower range offrequencies within the mmWave spectrum

In recent years there have beenmany efforts to accuratelymodel themmWave channel for LOS andNLOS transmissionlinks The communication industry has devoted extensiveresources to the development of accurate channelmodels Forinstance Samsung conducted channel measurements in the28GHz band and came up with a model to fit ITU-R andFITU-R standards [3]This model can be used for a mmWavecommunication link over a distance of up to 200 meters Inaddition Nokia Huawei and Deutsch Telekom have shownkeen interest in the 73GHz band and have developed proto-type hardware using the concept of Multiple Input MultipleOutput (MIMO) system to support throughput of 20GbpsA channel model based on stochastic path loss has beenstudied for urban propagation environment in [4] A spatialstatistical model of mmWave channel has been developedas a function of channel parameters including path loss in[5] This model has been derived based on real measure-ments at New York City in 28 and 78GHz bands for LOSand NLOS links

A general framework has been proposed for evaluatingcoverage probability and rate performance in a mmWavecellular network in [6] The short wavelength of mmWavesignals can accommodate a very large number of antennasat the receiver and at the transmitter which makes beam-forming a key enabler technique in obtaining high antennagain Beamforming antenna systems formmWave bands havebeen studied for 5G networks and a novel hybrid beamform-ing algorithm has been proposed for indoor and outdoortransmission links [7] The design and characterization ofmmWave antennas for wireless links are discussed in [8 9]

Shannonrsquos capacity limits and signal processing chal-lenges for giga bit mmWave communication are discussedin [10] Also Madhow has addressed opportunities andchallenges of utilizing MIMO in LOS mmWave communi-cation systems An overview of Impulse Radio-Ultra WideBand (IR-UWB) technology for design of transmitters andreceivers is presented in [11] In [12] coverage and capacityof a dense mmWave network are analyzed using a theoreticalmodel as a function of blockage probability and beamformingbandwidth Also the spectral efficiency (SE) (bitss) of ammWave network is investigated with LOS and NLOS linksin 28 and 73GHz mmWave bands Since minimal powerconsumption has received significance in recent years due togovernment regulations it is important to address this issuefor mmWave cellular networks This issue has not receivedmuch attention in the literature The efficiency metrics forLOS and NLOS links are evaluated for 28 and 73GHz bandsin terms of maximum achievable throughput and maximumpower efficiency in [13] However the work considered onlypath loss to represent the channel effects for transmission dis-tances up to 200m Tan et al have investigated the achievable

downlink spectral efficiency of multiuser mmWave cellularsystem considering both small- and large-scale fading effectsin [14] In [15] energy coverage probability was derived formmWave transmission as a function of network densitybeamforming bandwidth and channel parameters

The trade-off between energy efficiency and spectralefficiency is examined for different access networks in [16]It is demonstrated that for best utilization of mmWave bandsit is important to understand how to trade tolerable delay forlow power The delay-power trade-off is examined for designof an energy efficient cellular network in [17] Levanen et alhave proposed technologies to reduce the delay in mmWavecommunication systems in [18] The challenges of designingenergy efficient mmWave systems are discussed in [19] Theenergy efficiency (EE) (bitJ) metric is presented and evalu-ated for 28 and 73GHz mmWave bands for LOS and NLOSlinksThe investigation is presented as a function of BS trans-mission power Signal-to-NoiseRatio (SNR) network param-eters and mmWave channel parameters The spatial char-acteristics of cellular networks play a significant role in theperformance of mmWave cellular networksThe area spectralefficiency (ASE) and the area energy efficiency (AEE) areimportant metrics that need to be considered in the design ofmmWave networks

Thus the intent of this paper is to examine energyefficiency (EE) spectral efficiency (SE) network latency areaspectral efficiency (ASE) and area energy efficiency (AEE) ofmmWave cellular network in 28 and 73GHz bands for LOSand NLOS links The investigation is presented as a functionof SNR channelmodel parameters network parameters userdistance from BS and BS transmission power This paper isorganized as follows In Section 2 mmWave cellular systemis introduced including cellular network model mmWavesignal propagation and wireless channel models and NR andSINR calculations Section 3 presents evaluation of networkperformance metrics spectral efficiency energy efficiencyspatial spectral and energy efficiencies network latency andoutage probability Also the evaluation of these efficienciesmetrics is discussed Simulation algorithm and the assump-tions that have been made for our analysis are listed inSection 4Moreover the network and channel parameters aretabulated and the numerical results with discussion are alsopresented Finally the work is concluded in Section 5

2 Millimeter Wave Cellular System

The model of mmWave cellular system is shown in Figure 1The macrocells are divided into small cells shown in thefigureThe 119869macrocells in the network are (119895 = 1 119869)usingthe same frequency and therefore create intercell interferenceEach of the 119895th cells comprises 119868 small cells each with its ownBS (119894 = 1 119868) and uses the same frequency thereby causingintracell interference It is assumed that each small cell hasone BS and it serves 119870 Mobile Users (MUs) (119896 = 1 119870)The BSs are located at the center of small cells and connectedto the gateway of the macrocell The MUs are spatiallydistributed and each one is associated with a single BS at anytime




d



H = 119876sum119902=1

119862sum119888=1


(1)



120585 sim N (0 1205902119904 ) (2)



SNR119894119896[dB]


(3)




NLOS 72 292 87 1LOS 614 2 58 5

73GHzNLOS 866 245 80 1LOS 698 2 58 8



Λ 119895119894119896 = 119868sum119894=1119894 =119904

119875119901119894119866119901119894119866119901119904119896 10038161003816100381610038161003816ℎ119901119894119896100381610038161003816100381610038162 119871119901119894119896 (119903119894119896)

+ 119869sum119895=1119895 =119901

119868sum119894=1

119875119895119894119866119895119894119866119895119894119896 10038161003816100381610038161003816ℎ119895119894119896100381610038161003816100381610038162 119871119895119894119896 (119903119895119894119896) (4)



+ 1) (5)






(6)







02

0005

04

001

Out

age p

roba

bilit

y

06

4003000015

08

200100002

02040608

R (m)





(8)



4 Numerical Results




























Propose

Select


Model

Path loss model(2)





Use


Consider


Evaluate



EvaluateEvaluate



Spatialproperties

Calculate

Evaluate

Spatialproperties

Network latency(9)


End


Yes


Frequency bands


28 GHz 73GHz and



(bit

s)


1013

1012

1011

1010

109

108

10 20 30 40 50 60 70 80 90 100

d (m)



SE (b

its)

1013

1012

1011

1010

109

108

107

Pt (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40



2040 302060

SNR (dB)1080 0100

051152251012

1011

1010

d (m)minus10

times1012

(bit

s)Sp

ectr

al effi

cien

cy


Ener

gy effi

cien

cy (b

itJ)

P (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40


1012

1011

1010

109

108

107




sum 120578SE119895119894119896119860119895119894 (10)



20

Ener

gy effi

cien

cy (b

itJ)

40 302060

SNR (dB)1080 0100

24681012

1010

108

106

d (m) minus10

times1010


2468101214

times1010

20

Ener

gy effi

cien

cy (b

itJ)

40 2060

SNR (dB)10

80 0100

1010

108

106




10minus7

10minus8

10minus9

10minus10

Db

(sb

it)

minus10 0 10 20 30 40

Pt (dBm)




10minus8

10minus9

10minus10

Db

(sb

it)

10 20 30 40 50 60 70 80 90 100

d (m)






AEE = sum 120578EE119895119894119896119860119895119894 (11)




5 Conclusions



1014

1012

1010

108

106

104

AEE

(bit

Joul

eＧ

2)

0 10 20 30 40 50 60 70 80 90 100

d (m)




202040

10


1234567

d (m)

1010

108

106ASE

(bit

sＧ

2)

times1011



EE (b

itJo

ule)

1014

1012

2040

6080

100d (m) minus10

010

2030

40

Pt (dBm)

98765

34

21

times1015


Pt (dBm)

1012

1011

1010

109

108

107

AEE

(bit

Joul

eＧ

2)

minus10 minus5 0 5 10 15 20 25 30 35 40










Acknowledgments


References































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia





d



H = 119876sum119902=1

119862sum119888=1


(1)



120585 sim N (0 1205902119904 ) (2)



SNR119894119896[dB]


(3)




NLOS 72 292 87 1LOS 614 2 58 5

73GHzNLOS 866 245 80 1LOS 698 2 58 8



Λ 119895119894119896 = 119868sum119894=1119894 =119904

119875119901119894119866119901119894119866119901119904119896 10038161003816100381610038161003816ℎ119901119894119896100381610038161003816100381610038162 119871119901119894119896 (119903119894119896)

+ 119869sum119895=1119895 =119901

119868sum119894=1

119875119895119894119866119895119894119866119895119894119896 10038161003816100381610038161003816ℎ119895119894119896100381610038161003816100381610038162 119871119895119894119896 (119903119895119894119896) (4)



+ 1) (5)






(6)







02

0005

04

001

Out

age p

roba

bilit

y

06

4003000015

08

200100002

02040608

R (m)





(8)



4 Numerical Results




























Propose

Select


Model

Path loss model(2)





Use


Consider


Evaluate



EvaluateEvaluate



Spatialproperties

Calculate

Evaluate

Spatialproperties

Network latency(9)


End


Yes


Frequency bands


28 GHz 73GHz and



(bit

s)


1013

1012

1011

1010

109

108

10 20 30 40 50 60 70 80 90 100

d (m)



SE (b

its)

1013

1012

1011

1010

109

108

107

Pt (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40



2040 302060

SNR (dB)1080 0100

051152251012

1011

1010

d (m)minus10

times1012

(bit

s)Sp

ectr

al effi

cien

cy


Ener

gy effi

cien

cy (b

itJ)

P (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40


1012

1011

1010

109

108

107




sum 120578SE119895119894119896119860119895119894 (10)



20

Ener

gy effi

cien

cy (b

itJ)

40 302060

SNR (dB)1080 0100

24681012

1010

108

106

d (m) minus10

times1010


2468101214

times1010

20

Ener

gy effi

cien

cy (b

itJ)

40 2060

SNR (dB)10

80 0100

1010

108

106




10minus7

10minus8

10minus9

10minus10

Db

(sb

it)

minus10 0 10 20 30 40

Pt (dBm)




10minus8

10minus9

10minus10

Db

(sb

it)

10 20 30 40 50 60 70 80 90 100

d (m)






AEE = sum 120578EE119895119894119896119860119895119894 (11)




5 Conclusions



1014

1012

1010

108

106

104

AEE

(bit

Joul

eＧ

2)

0 10 20 30 40 50 60 70 80 90 100

d (m)




202040

10


1234567

d (m)

1010

108

106ASE

(bit

sＧ

2)

times1011



EE (b

itJo

ule)

1014

1012

2040

6080

100d (m) minus10

010

2030

40

Pt (dBm)

98765

34

21

times1015


Pt (dBm)

1012

1011

1010

109

108

107

AEE

(bit

Joul

eＧ

2)

minus10 minus5 0 5 10 15 20 25 30 35 40










Acknowledgments


References































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia





NLOS 72 292 87 1LOS 614 2 58 5

73GHzNLOS 866 245 80 1LOS 698 2 58 8



Λ 119895119894119896 = 119868sum119894=1119894 =119904

119875119901119894119866119901119894119866119901119904119896 10038161003816100381610038161003816ℎ119901119894119896100381610038161003816100381610038162 119871119901119894119896 (119903119894119896)

+ 119869sum119895=1119895 =119901

119868sum119894=1

119875119895119894119866119895119894119866119895119894119896 10038161003816100381610038161003816ℎ119895119894119896100381610038161003816100381610038162 119871119895119894119896 (119903119895119894119896) (4)



+ 1) (5)






(6)







02

0005

04

001

Out

age p

roba

bilit

y

06

4003000015

08

200100002

02040608

R (m)





(8)



4 Numerical Results




























Propose

Select


Model

Path loss model(2)





Use


Consider


Evaluate



EvaluateEvaluate



Spatialproperties

Calculate

Evaluate

Spatialproperties

Network latency(9)


End


Yes


Frequency bands


28 GHz 73GHz and



(bit

s)


1013

1012

1011

1010

109

108

10 20 30 40 50 60 70 80 90 100

d (m)



SE (b

its)

1013

1012

1011

1010

109

108

107

Pt (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40



2040 302060

SNR (dB)1080 0100

051152251012

1011

1010

d (m)minus10

times1012

(bit

s)Sp

ectr

al effi

cien

cy


Ener

gy effi

cien

cy (b

itJ)

P (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40


1012

1011

1010

109

108

107




sum 120578SE119895119894119896119860119895119894 (10)



20

Ener

gy effi

cien

cy (b

itJ)

40 302060

SNR (dB)1080 0100

24681012

1010

108

106

d (m) minus10

times1010


2468101214

times1010

20

Ener

gy effi

cien

cy (b

itJ)

40 2060

SNR (dB)10

80 0100

1010

108

106




10minus7

10minus8

10minus9

10minus10

Db

(sb

it)

minus10 0 10 20 30 40

Pt (dBm)




10minus8

10minus9

10minus10

Db

(sb

it)

10 20 30 40 50 60 70 80 90 100

d (m)






AEE = sum 120578EE119895119894119896119860119895119894 (11)




5 Conclusions



1014

1012

1010

108

106

104

AEE

(bit

Joul

eＧ

2)

0 10 20 30 40 50 60 70 80 90 100

d (m)




202040

10


1234567

d (m)

1010

108

106ASE

(bit

sＧ

2)

times1011



EE (b

itJo

ule)

1014

1012

2040

6080

100d (m) minus10

010

2030

40

Pt (dBm)

98765

34

21

times1015


Pt (dBm)

1012

1011

1010

109

108

107

AEE

(bit

Joul

eＧ

2)

minus10 minus5 0 5 10 15 20 25 30 35 40










Acknowledgments


References































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia





02

0005

04

001

Out

age p

roba

bilit

y

06

4003000015

08

200100002

02040608

R (m)





(8)



4 Numerical Results




























Propose

Select


Model

Path loss model(2)





Use


Consider


Evaluate



EvaluateEvaluate



Spatialproperties

Calculate

Evaluate

Spatialproperties

Network latency(9)


End


Yes


Frequency bands


28 GHz 73GHz and



(bit

s)


1013

1012

1011

1010

109

108

10 20 30 40 50 60 70 80 90 100

d (m)



SE (b

its)

1013

1012

1011

1010

109

108

107

Pt (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40



2040 302060

SNR (dB)1080 0100

051152251012

1011

1010

d (m)minus10

times1012

(bit

s)Sp

ectr

al effi

cien

cy


Ener

gy effi

cien

cy (b

itJ)

P (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40


1012

1011

1010

109

108

107




sum 120578SE119895119894119896119860119895119894 (10)



20

Ener

gy effi

cien

cy (b

itJ)

40 302060

SNR (dB)1080 0100

24681012

1010

108

106

d (m) minus10

times1010


2468101214

times1010

20

Ener

gy effi

cien

cy (b

itJ)

40 2060

SNR (dB)10

80 0100

1010

108

106




10minus7

10minus8

10minus9

10minus10

Db

(sb

it)

minus10 0 10 20 30 40

Pt (dBm)




10minus8

10minus9

10minus10

Db

(sb

it)

10 20 30 40 50 60 70 80 90 100

d (m)






AEE = sum 120578EE119895119894119896119860119895119894 (11)




5 Conclusions



1014

1012

1010

108

106

104

AEE

(bit

Joul

eＧ

2)

0 10 20 30 40 50 60 70 80 90 100

d (m)




202040

10


1234567

d (m)

1010

108

106ASE

(bit

sＧ

2)

times1011



EE (b

itJo

ule)

1014

1012

2040

6080

100d (m) minus10

010

2030

40

Pt (dBm)

98765

34

21

times1015


Pt (dBm)

1012

1011

1010

109

108

107

AEE

(bit

Joul

eＧ

2)

minus10 minus5 0 5 10 15 20 25 30 35 40










Acknowledgments


References































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia






















Propose

Select


Model

Path loss model(2)





Use


Consider


Evaluate



EvaluateEvaluate



Spatialproperties

Calculate

Evaluate

Spatialproperties

Network latency(9)


End


Yes


Frequency bands


28 GHz 73GHz and



(bit

s)


1013

1012

1011

1010

109

108

10 20 30 40 50 60 70 80 90 100

d (m)



SE (b

its)

1013

1012

1011

1010

109

108

107

Pt (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40



2040 302060

SNR (dB)1080 0100

051152251012

1011

1010

d (m)minus10

times1012

(bit

s)Sp

ectr

al effi

cien

cy


Ener

gy effi

cien

cy (b

itJ)

P (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40


1012

1011

1010

109

108

107




sum 120578SE119895119894119896119860119895119894 (10)



20

Ener

gy effi

cien

cy (b

itJ)

40 302060

SNR (dB)1080 0100

24681012

1010

108

106

d (m) minus10

times1010


2468101214

times1010

20

Ener

gy effi

cien

cy (b

itJ)

40 2060

SNR (dB)10

80 0100

1010

108

106




10minus7

10minus8

10minus9

10minus10

Db

(sb

it)

minus10 0 10 20 30 40

Pt (dBm)




10minus8

10minus9

10minus10

Db

(sb

it)

10 20 30 40 50 60 70 80 90 100

d (m)






AEE = sum 120578EE119895119894119896119860119895119894 (11)




5 Conclusions



1014

1012

1010

108

106

104

AEE

(bit

Joul

eＧ

2)

0 10 20 30 40 50 60 70 80 90 100

d (m)




202040

10


1234567

d (m)

1010

108

106ASE

(bit

sＧ

2)

times1011



EE (b

itJo

ule)

1014

1012

2040

6080

100d (m) minus10

010

2030

40

Pt (dBm)

98765

34

21

times1015


Pt (dBm)

1012

1011

1010

109

108

107

AEE

(bit

Joul

eＧ

2)

minus10 minus5 0 5 10 15 20 25 30 35 40










Acknowledgments


References































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



(bit

s)


1013

1012

1011

1010

109

108

10 20 30 40 50 60 70 80 90 100

d (m)



SE (b

its)

1013

1012

1011

1010

109

108

107

Pt (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40



2040 302060

SNR (dB)1080 0100

051152251012

1011

1010

d (m)minus10

times1012

(bit

s)Sp

ectr

al effi

cien

cy


Ener

gy effi

cien

cy (b

itJ)

P (dBm)


minus10 minus5 0 5 10 15 20 25 30 35 40


1012

1011

1010

109

108

107




sum 120578SE119895119894119896119860119895119894 (10)



20

Ener

gy effi

cien

cy (b

itJ)

40 302060

SNR (dB)1080 0100

24681012

1010

108

106

d (m) minus10

times1010


2468101214

times1010

20

Ener

gy effi

cien

cy (b

itJ)

40 2060

SNR (dB)10

80 0100

1010

108

106




10minus7

10minus8

10minus9

10minus10

Db

(sb

it)

minus10 0 10 20 30 40

Pt (dBm)




10minus8

10minus9

10minus10

Db

(sb

it)

10 20 30 40 50 60 70 80 90 100

d (m)






AEE = sum 120578EE119895119894119896119860119895119894 (11)




5 Conclusions



1014

1012

1010

108

106

104

AEE

(bit

Joul

eＧ

2)

0 10 20 30 40 50 60 70 80 90 100

d (m)




202040

10


1234567

d (m)

1010

108

106ASE

(bit

sＧ

2)

times1011



EE (b

itJo

ule)

1014

1012

2040

6080

100d (m) minus10

010

2030

40

Pt (dBm)

98765

34

21

times1015


Pt (dBm)

1012

1011

1010

109

108

107

AEE

(bit

Joul

eＧ

2)

minus10 minus5 0 5 10 15 20 25 30 35 40










Acknowledgments


References































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



20

Ener

gy effi

cien

cy (b

itJ)

40 302060

SNR (dB)1080 0100

24681012

1010

108

106

d (m) minus10

times1010


2468101214

times1010

20

Ener

gy effi

cien

cy (b

itJ)

40 2060

SNR (dB)10

80 0100

1010

108

106




10minus7

10minus8

10minus9

10minus10

Db

(sb

it)

minus10 0 10 20 30 40

Pt (dBm)




10minus8

10minus9

10minus10

Db

(sb

it)

10 20 30 40 50 60 70 80 90 100

d (m)






AEE = sum 120578EE119895119894119896119860119895119894 (11)




5 Conclusions



1014

1012

1010

108

106

104

AEE

(bit

Joul

eＧ

2)

0 10 20 30 40 50 60 70 80 90 100

d (m)




202040

10


1234567

d (m)

1010

108

106ASE

(bit

sＧ

2)

times1011



EE (b

itJo

ule)

1014

1012

2040

6080

100d (m) minus10

010

2030

40

Pt (dBm)

98765

34

21

times1015


Pt (dBm)

1012

1011

1010

109

108

107

AEE

(bit

Joul

eＧ

2)

minus10 minus5 0 5 10 15 20 25 30 35 40










Acknowledgments


References































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



1014

1012

1010

108

106

104

AEE

(bit

Joul

eＧ

2)

0 10 20 30 40 50 60 70 80 90 100

d (m)




202040

10


1234567

d (m)

1010

108

106ASE

(bit

sＧ

2)

times1011



EE (b

itJo

ule)

1014

1012

2040

6080

100d (m) minus10

010

2030

40

Pt (dBm)

98765

34

21

times1015


Pt (dBm)

1012

1011

1010

109

108

107

AEE

(bit

Joul

eＧ

2)

minus10 minus5 0 5 10 15 20 25 30 35 40










Acknowledgments


References































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



Acknowledgments


References































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


spectral and energy efficiencies in mmwave cellular

Documents