from the bit to multi-carrier modulation - basics of ...the-eye.eu/public/books/electronic...

Digital modulation techniques havebeen used in radio engineering for along time. Together with the use ofsignal processing it can combine theadvantages of digital transmission atlow cost. The following overview ofdigital modulation techniques in thetime and frequency domain gives aninsight into modern multi-carriertechniques. As an example DigitalRadio Mondiale (DRM) is digital shortwave broadcasting.

1.Basic techniques of digitaltransmission

How does the bit become the RF carrier?The three standard practices ofamplitude, frequency and phasemodulation can be transferred directly tothe digital world. In the simplest formmodulation methods that contain onlyone bit are used, e.g. radio telegraphyusing two frequencies for zero and one. Ifthe transmission path is errors free amodulation method can contain more ofthan two conditions and transfer morethan one bit; there are communicationtechniques with 16 frequencies meaningthat the symbol contains 4 bits.

Amplitude Shift Keying (ASK):The simplest form of amplitudemodulation is pulsing the carrier fortelegraphy. Since the zero conditioncannot be differentiated from aninterrupted connection, in practice twoamplitude values are used: Duringtransmission of DCF77 for example,100% and 25% amplitude are used.

Advantages: • Modulation spectrum is only

upward and symmetrical to thecarrier frequency

• Modulation can take placesomewhere between the oscillatorand the antenna

• Amplitude can be directly measuredat the receiver

Disadvantages: • Amplitude errors on the

transmission (fluctuations in level,malfunctions) produce errors

• Frequency Shift Keying (FSK):The modulation method usesdifferent frequencies, e.g. RTTYand packet radio use twofrequencies for zero and one. It canbe interpret either as two phaseamplitude modulated channels or asfrequency modulation with a non-existent centre frequency with adeviation of ΔF. If the spacing of

Jochen Jirmann, DB1NV

From the bit to multi-carriermodulation - basics of digitalmodulation techniquesFrom a lecture at the VHF conference inWeinheim/Bensheim 2006

VHF COMMUNICATIONS 2/2008

88

the two frequencies is equal to halfthe data rate, then it is calledMinimum Shift Keying (MSK).This is the standard modulation forthe GSM mobile phone system.

Advantages: • Amplitude does not contain any

information therefore it isinterference proof

• No linear transmitter amplifiers orlinear receivers necessary

• Frequency can be directly measuredat the receiver

Disadvantages: • The frequency modulated carrier is

theoretically infinitely wide,broader than AM (B ≈ 2Δ + 2fmod)

• Modulation requires modulating theoscillator thus worse frequencystability

Phase Shift Keying (PSK): Using mathematics it can be shownthat frequency and phase modulationare very similar and can be convertedto one another using suitable filters.The most common method uses two-phase shift keying (0/180°, BinaryPhase Shift Keying) and a four-phaseshift keying (0/90/180/270°,Quadrature Phase Shift Keying). Thetwo-phase shift keying can be treatedas a special case of the amplitudemodulation where the carrierfrequency is first multiplied by a factor+1 and then by a factor -1. The four-phase shift keying can be consideredas two BPSK signals with carriers thatare phase shifted by 90°. It can beshown mathematically that these twoequivalent carriers do not affect eachother and they can be used for BPSKor AM and modulated with different

Fig 1: An amplitude modulated 10MHz signal.


89

information.

Advantages: • The phase modulator comes after

the oscillator • Amplitude does not contain any

information therefore it isinterference proof

• No linear transmitter amplifiers orlinear receivers necessary

Disadvantages: • Phase cannot be directly measured

at the receiver, only phase changes • The frequency spectrum is wider

than the comparable AM or FMsignals

Combined methods : Quadrature Amplitude Modulation(QAM): If a transmit path is linear andhas a good signal-to-noise ratio, thenthe data rate can be increased by a

combination of amplitude and phasemodulation: The amplitude and phaseof the carrier is changed at the sametime in small steps. Quadratureamplitude modulation accommodates4 to 15 bits. QAM with 16 states (4bits) and 64 states (6 bits) is incommon use. The main application for higher QAMmethods with 1024 to 32768 states aremodems and DSL technologies onfixed line connections.

Advantages: • Best utilisation of the transmit path

in terms of bits/Hz .

Disadvantages: • Good signal-to-noise ratio

necessary • Linear transmit path from the

transmitter to the receiver necessary

Fig 2: The spectrum of an amplitude modulated signal.


90

2.Modulation in the time domainand frequency domain andfiltering of the data signal

It is known that switching the carrier of atelegraph transmitter produces clickingdisturbances in neighbouring radio links.A low-pass filter in the transmitter solvesthe problem if the modulator and thefollowing transmit amplifier are linear.How is the low-pass designed? HarryNyquist found the answer years ago withthe investigation of machine producedtelegraph signals:The most unfavourable case is the bitpattern (01010101….) corresponding to asymmetrical square wave pulse. A datastream of 1kBit/sec represents a square

wave pulse with a frequency of 500HzIf the base frequency of the square waveis fitted to the transmission path theharmonics are filtered out. The datastream then becomes a sine wave and it isdifficult to find the centre of the bits,which is a problem for c lockregeneration.

Broadband modulation: A digital data stream has a very fastrise time (74HC logic: Rise time <10nS), its spectrum extends far beyonddata rate. An un-filtered data signalshould not be fed to a modulatorbecause it produces broadbandmodulation that makes neighbouringradio links useless. The same thinghappens if an SSB transmitter is overdriven producing a square signalfeeding the output stage, some conteststations can be heard +/- 300kHz and

Fig 3: The impulse response of a 19kHz cutoff, 3rd order low-pass Butterworthfilter fed with a 40kBit/sec data stream.


91

that is not because of a bad receiver!An amplitude modulator is used for thefollowing examples. A diode ringmixer is used as an AM modulator fora 10MHz carrier. The data stream hasa data rate of 40kBit/sec and a 32767(215 –1) bit pseudo random generator isused to represent the data.The modulated signal is shown in thetime domain in Fig 1, the bit length is25µsec from the 40kBit/sec data rate.The associated spectrum is shown inFig 2; the carrier frequency is in thecentre and the sidebands extend200kHz to the left and right. For thoseinterested in the mathematics: theamplitude of the sidebands follow (sinx)/x function. With such a signal theneighbouring radio links would besubstantial disturbed. The modulationsignal must be filtered in a suitableway! Subjectively it is assumed that

with reduced rise time of the datasignal the range occupied becomessmaller. At first sight a sharp low-passfilter is the solution, but this isdiscussed further.Correct filtering of the modulationsignal: Most IF bandpass filters inamateur radio equipment are optimisedfor a rectangular transmission curve,because the Shape Factor is animportant advertising feature but mostspecial CW filters under 250Hz ringand are hardly useful. A CW signalonly occupies a few Hz so why doesn’tit fit in a 100Hz filter? Then there arethe strange transmission curves of theIF bandpass filters in a spectrumanalyser.

This is where filter theory helps: If a low-pass filter is fed with a shortrectangular voltage pulse or a short RF

Fig 4: The impulse response of a 19kHz cutoff, 10th order, super-steep, lowpass Tschebyscheff filter fed with a 40kBit/sec data stream.


92

pulse at the centre frequency of abandpass filter, then the output is notonly a reduced version of the originalinput but also echoes that are the ringingof the CW filter. The “better” the filterthe more echoes can be seen. Theimpulse response of a 19kHz cutoff, 3rdorder low-pass Butterworth filter fedwith a 40kBit/sec data stream is shown inFig 3. The second trace is the clock. Thepulse has been beautifully “rounded”:One bit time later the pulse responsecrosses zero, which means thatneighbouring bits in the bit stream arenot affectedA negative echo of the origin pulse canbe seen, its amplitude is approximately10 times smaller than the first pulse.Fig 4 shows a 10th order, super-steep,low pass Tschebyscheff filter with analmost rectangular transmission curve.

The data rate is again 40kBit/sec and thecutoff frequency 19kHz. 12 echoes of theorigin pulse can be seen fading awaysinusoidally so this filter is not useful.The following requirements result forfiltering digital signals:

• A low pass filter with a cutofffrequency of half the data rate canbe used to filter digital signals andleave the bits recognisable.

• Low-pass filters must have agradual transition into the cutoffarea so that the filter does not ring.Theoretically the optimal solutionsare Gaussian filters (as used for IFbandpass filters in spectrumanalysers) and Cosine roll off filterswith a cosine shaped filter curve.

• The 3rd order Butterworth low passfilter represents a goodapproximation.

• By small variations to the filter

Fig 5: An amplitude modulated signal filtered with a Butterworth low-passfilter.


93

cutoff frequency the pulse responseof the filter can be designed to makethe zero cross over in the centre ofthe following bit so thatneighbouring bits affect each otheras little as possible.

The advantage of the last point can onlybe achieved if the bit timing is recoveredin the correct phase at the receiver andthe bit stream is scanned in the centre ofthe bit timing. This highlights twoadditional tasks: The transmitter mustsend sufficient bit changes independentlyof the bit stream to ensure that the bittiming can be recovered. The receivermust be able to recover the bit timing,usually using a PLL circuit.The success of filtering with aButterworth filter is show in Fig 5 in thetime domain and Fig 6 in the frequencydomain. The formerly rectangular RFenvelope is rounded and the powerdensity at more than 40kHz from the

carrier is practically zero.

Disturbed transmission paths: Digital transmission techniques wereintroduced with the goals of improvingquality and of reducing the equipmentcost. To start with digital transmissiontechniques were concerned with theincrease of the bit rates into theGigabit region so that the giganticcapacity of fibre optic cable could beused. That can be achieved withcarefully planned transmission pathsthat have a good, well-behaved andsmooth frequency response and lownoise. These conditions are fulfilled bycable but not by radio paths. • On long cable runs the

characteristic impedance jumps ifthe connections are notimplemented correctly causinghumps in the frequency response.

• Multi-path propagation in a hop

Fig 6: The spectrum of an amplitude modulates signal filtered with aButterworth low-pass filter.


94

produces sharp "holes" in thefrequency response caused byinterference

• Foreign transmissions disturb bothcable runs (cross talk from othercable cores or unsatisfactoryshielding) and radio links.

• The frequency response of DRMreception taken with stronglydisturbed frequency is shown in Fig7 and Fig 8.

The following is applicable: • A frequency response equaliser

gives good results so that thefrequency response is not toouneven.

• Some radio systems use automaticchannel selection and look a goodradio link

• Using frequency diversitytechnology the information istransmitted on two frequencies in

Fig 7: Frequencyresponses of aDRM transmissionon short wave

Fig 8: Frequencyresponses of aDRM transmissionon short wave


95

the hope that one is good.Area diversity uses a second receivingantenna shifted by some wavelengths thatoften supplies unimpaired receptionduring multi-path propagation.A difficult radio link in California hasused double frequency diversity andquadruple area diversity for some years.

Advantages of multi-carriermodulation: The frequency diversity method can beused with more consistency. Before thedigital age telephone calls wereconverted using frequency-multiplexingequipment in to s ingle s idebandmodulation; the method meant that thecalls could be transmitted using bothcable and radio (radio relay link orsatellite communication). If a subchannel was disrupted it was not used.With a disrupted channel unused only thetransmission rate was affected and theconnection was not broken, the sameapplies to a digital transmission.This advantage has been rediscoveredwith the availability of special signalprocessors thus overcoming thearguments against frequency divisionmultiplexing. The modulators and filtersfor each sub channel can be realised

digitally using cheap signal processors.There are many cable and radio systemsusing digital frequency multiplexingequipment:

• ADSL technology (AsymmetricDigital Subscriber Line) uses 4kHzwide sub channels with 4.3125 kHzspacing

• DAB (Digital Audio Broadcasting)uses 1536 sub channels in a1.5MHz spread spectrum

• DVB-T (Digital VideoBroadcasting - Terestrial) uses 2048or 8192 sub channels in one 7MHzwide Television channel

• DRM (Digital Radio Mondiale)uses the 9kHz wide AM radiochannels with 40 to 110Hz(!) spreadsub channels.

Even the portable radio planners thinkafter the failure of UMTS and spreadingspectrum modulation multi-carriermodulation (3G) that this is the 4thgeneration portable radio system.It must not be forgotten that a multi-carrier modulation system has aconsiderable administration overhead.With a two-way system users can stilloperate with the loss of a single channelof the distributed channels. Broadcastsystems send enough redundant

Fig 9: Carriers for an orthogonal frequency multiplexed system (OFDM).


96

information so that it does not present aproblem if a certain portions of thechannels are not usable.A recognised disadvantage of multi-c a r r i e r s y s t e m s i s t h e l o n gsynchronisation time: a DSL modemdoes not need many seconds tosynchronise, depending upon line quality,a transmitter search can be used fordigital television broadcasts. Thedevelopments to a DRM similar amateurradio procedure run however!

Summary of the pros and cons ofmulti-carrier systems:

Advantages: • Well suited to bad transmission

paths with poor frequency responseor sharp breaks

• The frequency response only needsto be good in the sub channel

• Data with a low rate is transferredin the sub channel. Thus the

transmission becomes insensitive tomulti-path propagation. Therequirement is that the transit timedifference between the multi pathwaves at the receiver must besmaller than the symbol length

• Individual sub channels can failcompletely during transmission

Disadvantages: • The modulation is generated with

considerable amounts of digitalsignal processors

• Pilot channels must be inserted forrecovery

• The synchronisation procedure islong therefore it is suitable forpermanent connections

Practical realization: DMT andOFDM:The term Digi ta l Mul t i -car r ie rModulation (DMT) is self-descriptive.

Fig 10: Spectrum of a DRM signal as seen on a spectrum analyser.


97

The term Orthogonal Frequency DivisionMultiplex (OFDM) is also used inc o n n e c t i o n w i t h m u l t i - c a r r i e rmodulation. Orthogonal means right-angled, but why is this used for atransmission technique? Sine and cosinewaves are orthogonal for the samefrequency, if they are multiplied togetherthen the dc component is zero. That hasbeen used for many years in quadratureamplitude modulation for PAL colourtelevision transmissions to transfer twochannels on a carrier.The frequency spacing of the sub carriersfor a multi-carrier modulation can bechosen freely as long as the spectra donot overlap themselves. There are carrierspacings where the mutual disturbancebecomes minimal thus saving onfiltering.Looking again at Fig 2: Although thedata signal is not filtered, the spectrumfalls to zero symmetrically around thecarrier at a spacing of the bit rate.Therefore it is best to select thefrequency spacing of the sub carriers atinteger multiples of the symbol rate. Thefirst case (standard OFDM) does not givethe closest packing of the sub carriers buttheoretically the sub channels dointerfere. This is because the informationis carried in the range of the carrierfrequency plus/minus half the symbolrate as shown in Fig 9 from [1]. Thiscannot be seen on a spectrum analyser;only the rectangular form of the wholespectrum can be seen as shown in Fig 10.

3.Digitally Radio Mondiale -current state

In the last few years some digitalbroadcast and television systems havebeen introduced, whose use for theordinary consumer is either doubtful orwho does not fulfil expectations created

by advertising: • Digital Satellite Radio (DSR) failed

after 5 years due of lack ofparticipants!

• Digital audio Broadcasting (DAB)has been operating for years in thegeneral market and should replaceVHF FM broadcasts. The qualityloss compared to VHF FM is notgreat but it is doubtful if thepopulation will accept 100 millionworthless radios in Germany aloneeven after a long period of use andlow priced receivers. It is not clearwhat the thoughts of DAB are inother countries. (In the UK it iseven more troublesome because theBBC standard for DAB is not thesame as the European standard andis due to be harmonised with evenmore useless receivers – Andy)

• Digital Video Broadcasting (DVB)was publicised to have better imagequality and in addition still moreprograms. These are (usually) of thelowest technical quality (data rateunder 2MBit/sec) and squeezed intothe transmission paths. Withmodern large screen televisions onethe bad image quality can be seenshowing “dancing pixels”.

• DVB-T has the advertising sloganof television for all, many programsusing a Yagi antenns..

• DVB-H (television on the mobilephone), it is doubtful if this willbecome a market success.

• Now DRM (Digital RadioMondiale) is about to digitise themedium and short wave broadcastbands. Is a market success to beexpected here? No mass market, butan interesting addition: a clear jumpin audio quality (almost VHF FMquality) compared to AMbroadcasts and at the same timeworldwide reception with onlytelescopic antenna an no additionalinfrastructure.


98

3.1. DRM has the followingcharacteristics

• The 9 or 10kHz channels used onthe long, medium, and shortwavebands are sufficient

• Further use of existing, moderntransmitting plants

• Good reception with lowsignal/signal-to-noise ratio

• Stability against multi-pathpropagation in the ionosphere

• Convenient operation as withmodern VHF FM receivers withRadio Data System stem (RDS)

In addition the following benefits are: • A new, efficient audio data

compression technique calledAdaptive Audio Coding with HighQuality Spectral Band Replication(AAC with HQ SBR) supplies VHFFM quality with a data rate around25kBit/sec.

• The transmission uses digitalfrequency multiplexing (OrthogonalFrequency Division Multiplex) andsplits up the 9kHz channel into 40to 110Hz wide sub channels that aremodulated with a low data rate.Depending upon the system the subchannels are modulated with four-phase quadrature amplitudemodulation (2 bits/symbol) up to64-phase (6 bits/symbol). The longsymbol duration makes the systeminsensitively to multi-pathpropagation; the multi-carrierprocedure has sufficient redundancythat the loss of a carrier by Fadingor disturbers can be corrected.

• Interleaving of the data over several100 milliseconds improves the errorcorrection in the case of bundleerrors.

The disadvantage of the system is thatdespite special synchronisation channels,synchronisation takes a very long time.

Fig 11: Spectrum of a DRM signal around the carrier centre (high resolution).


99

The following points affect this: • Search of the pilot channels, carrier

synchronisation and determinationof the mode

• Waiting for the Interleave period • Error correction and data expansion

The entire procedure takes some seconds,so that it is not possible to carry out atransmitter search in the usual sense.Because modern world receivers have astored frequency lists and transmissiontimetables stored in EEPROMs, anautomatic search of the data channels inthe DRM system could be the solution.Digital Radio Mondiale cannot replacethe normal AM broadcasts. In additionthe existence of a billion AM receivers istoo large. It will be an interestingalternative for holidaymakers to heartheir home stations on a telescopicantenna without systems such as theInternet or a satellite antenna. Onlycurrent saving DSP demodulators andAAC audio decoder are missing. So farthere are unfortunately only expensivereceivers. Phillips have announced aDRM demodulator IC for car radios!

3.2. The DRM modulation format

Channel width: Standard 9kHz (LW, MW) and 1 kHz(SW), in addition there are halfchannels with 4.5 or 5kHz spacing anddouble channels with 18 or 20kHzspacing.

Stages of the error protection(Robustness):

There are 4 stages A to D (Table 1); Ais only for channels with Gaussiandistributed noise to D for channelswith time-variable multi-pathpropagation with high transit timedifferences and Doppler shift.

Logical channels:The Fast Access Channel (FAC) is thesynchronisation channel; the ServiceDescription Channel (SDC) is fordecoding the multiplexer sub datastreams, the Main Service Channel(MSC) contains the data.

Symbol duration:16 2/3mS, 20mS, 26 2/3mS

Transmission capacity for data:Depending upon channel width anderror protection, 6 to 55kBit/sec(theoretical range 4.8 to 72kBit/sec),15 to 25kBit/sec is usually used.

Pilot channels:750Hz, 2250Hz and 3000Hz above thechannel cutoff frequency

Signal-to-noise ratio required:15 to 23dB depending upon channelmodel (only noise up to multi-pathpropagation with 6mS transit timedifference and 4Hz of Doppler shift),under favourable conditions 10 to11dB

Table 1: DRM Modes of error protection.

Mode Duration Tu Carrier Duration Duration Tg/Tu Numberspacing guard of symbol of symbols 1/Tu Tg interval frame Ns per

Ts=Tu+TgA 29mS 41 2/3Hz 2.66mS 26.66mS 1/9 15B 21.33mS 46 7/8Hz 5.33mS 26.66mS 1/4 15C 14.66mS 68 2/11 5.33mS 20mS 4/11 20D 9.3mS 107 1/7Hz 7.33mS 16.66mS 11/14 24


100

Data compression audio signalAAC (Adaptive Audio Coding),optional SBR (Spectral BandReplication) production “syntheticlevel”, alternatively mono or Stereo

4.Literature:

[1] Analogue and digital modulationprocedures, Mäusl/Göbel, Hüthig-Verlag2002

[2] DVB digital TV, U Freyer, VerlagTechnik 1997[3] The basics of the televisionengineering, G. Meals , Springer Verlag2005 [4] ETSI Technical Specification TS 101980. ver. 1.1.1 (September 2001)[5] BBC R&D White paper WHP 064,Digitally Radio Mondiale - revitalisingthe bands below 30 MHz, July 2003

With over 700 members world-wide, the UK Six Metre Groupis the world's largest organisation devoted to 50MHz. Theambition of the group, through the medium of its 56-pagequarterly newsletter 'Six News' and through its web sitewww.uksmg.com, is to provide the best information availableon all aspects of the band: including DX news and reports,beacon news, propagation & technical articles, six-metreequipment reviews, DXpedition news and technical articles.Why not join the UKSMG and give us a try? For moreinformation contact the secretary: Dave Toombs, G8FXM, 1Chalgrove, Halifax Way, Welwyn Garden City AL7 2QJ, UK orvisit the website.

The UK Six Metre Group

www.uksmg.com


101

In the first part of this article thegeneral conditions of digital radiostandards were described, in a similarway this article describes the DECTstandard. A comparison is made to thesimilar FM voice transmission stand-ards. The achievable range as well theability for trouble free reception andreception of data signals at low signallevel is considered under comparableconditions.

2.7. DECTDECT (Digital Enhanced Cordless Tel-ecommunications) [19], [24], [4] wasspecified as an ETSI standard in 1992. Itis a micro-cellular digital portable radionetwork for high user densities, to beused predominant in company buildingsfor data and language transfer. A combi-nation of TDMA and FDMA are used forthe DECT standard. The available fre-quency spectrum from 1880MHz to1900MHz is divided into 10 channelswith 1728kHz spacing. The centre fre-quency fc is calculated as follows: fc = f0 - c • 1728kHz with C = 0, 1,…, 9 and f0 = 1897.344MHz {1} The centre frequency for an active statecan vary by a maximum of ±50kHz. Themodulation is Gaussian minimum ShiftKeying (GMSK) with a range time prod-uct, B · T = 0.5. GMSK is a special typeof the MSK (Minimum Shift Keying).

Instead of switching violently betweenthe two frequencies, the switching edgesare flattened using a Gauss filter leadingto the fact that the signal range isreduced. The transmission capacity of each chan-nel is divided into 10ms long framescontaining 11520 bits. This frameworkgives a data transmission rate of1152kbit/s. Each 10ms long frame isdivided into 24 time slots. The first 12time slots are intended for downlink, thenext 12 for the uplink (Fig 6). For aduplex connection pairs of time slots areformed following each other within 5ms.Each of the 24 time slots has a length of480 bits. The first 32 bits are reserved forthe preamble and synchronisation. For aconnection 64 bits of control informationfollow:C (signalling for higher layers)P (Paging information) Q (system information, downlink only)M (MAC Layer information) N (handshake, identity information)as well as 320 bits of useful informationand a 4 bit parity check. A protection gapof 60 bits forms the remainder.DECT uses a building block principle,the system component parts are: the airinterface (Physical Layer), the accesslayer (Medium Access Control layer,

Current digital radio standardssimilar to FM voicetransmission, Part 2Continued from 2/2008

Michael Gabis, Ralf Rudersdorfer


231

MAC Layer), the link layer (Data LinkControl Layer), the network or switchinglayer (Network Layer) and the adminis-tration of the lower layers (Lower LayerManagement Entity).Changing the radio cell without anyinterruption passes on a conversation fora moving mobile user. The maximumdistance between the base station andmobile equipment in a free range is 300mand in buildings is up to 50m. This rangecan be increased using relay stations ormobile relay stations. For cost reasonsDECT was developed for user speeds upto 20km/h.The bit duration is 0.868µs. Withouterror correction time differences of under10% of the bit duration are permissible,DECT can work with a delay spread upto 200 – 300nS, which corresponds to atransmission distance difference of ap-proximately 100m being suitable for dif-ficult environments Note: Stations withbig reflections or problems with reflect-ing metal walls suitable cell planning andsector antennas can be used if necessary. DECT offers the following service possi-bilities: • Quality speech transmission in fixed

network

• Duplex data communication rateswith a total bit rate of 38.8kBit/s. Inthe half duplex mode the total bit rateof up to 931.2kBit/s is possible if all24 time slots used. Still higher datarates are possible by multiplexingseveral carriers. Thus many cordlessdata applications are possible such asInternet, ISDN, Fax or TV.

• Applications for DECT covers PBX(Private Branch Exchange, privatePABX), Telepoint and wireless localnetworks as well as speech tel-ephones.

The DECT system can handle up to 1000users in a Local Area (LA) [7]. For largernumbers of users, different cell areas canbe planned that can be administeredinternally by DECT. A Dynamic ChannelSelection procedure (DCS) is used tomanage the high speech and data trafficloads and their uneven load on a cell withtemporary and very variable load peaks.Thus in principle the entire frequencyspectrum with all 120 channels are avail-able to the mobile station and a suitablechannel is selected in each cell. The 120channels result from the 10 carrier fre-quencies and the 12 time slots for thedownlink or uplink.In cellular portable radio systems (GSM,

Fig 6: Frame structure for a DECT system.


232

UMTS,…) channel assignment takesplace via a Fixed Channel Allocation(FCA). This is a very precise cell plan-ning system but short load changes on anindividual cell are not dealt with verywell. With DECT there is no frequencyplanning, just one planning for the loca-tions is necessary called DCS. The sys-tem can lower the blocking probabilityindependently of changing loads adjustitself better than FCA networks. Thedynamic assignment takes place accord-ing to certain rules so that there are nolarge disturbances. One rule is the inter-ference adaptive channel assignment; dis-patching decisions are made based ononline measurements of the interferencelevel. There are two types central anddecentralised systems, DECT uses decen-tralised systems. The highest capacitygains can be obtained with centralisedsystems because all cell values are avail-able to the control system and all dis-patching decisions can be optimisedbased on the disturbances in adjoiningcells. However the signalling expenditurebetween cells and control unit is higher.With decentralised systems, like withDECT, each permanently cell installeddecides independently of other cells, sothat the danger exists that channel dis-

patching will clash with another cell. Ahigh threshold value for the C/I (Carrierto Interference ratio) reduces this danger,however this affects the capacity gains.The advantage is that there is no fre-quency planning and the system can copewith changes in network loads. Usinginterference adaptive channel assignmentprocedures it is possible to extend aDECT system by simply adding furtherpermanently installed stations and in-crease the capacity.The actual maximum number of channelsof a DECT cell depends on the existingnumber of transceivers. If it has only onetransceiver, then only one mobile stationcan be served per time slot, since atransceiver cannot work with two differ-ent frequencies at the same time. Thusonly 12 channels can be used. If achannel on a terminal is occupied by amobile, then the remaining channels inthe same time slot are marked as blindslots (Fig 7). The mobile station looks forthe possible channels within each frame-work in order to attain information abouttheir quality. If the terminal occupies achannel for transmission then it cannotsupervise the channels of the other fre-quencies in this time slot. Because ofeconomies in the design it is not usual tobe able to examine the time slots forquality before and after use. The blockedchannels are no longer available (Fig 8). To make the data transmission rate ofindividual connections variable severaltime slots can be assigned. This dispatch-ing can take place for the uplink anddownlink asymmetrically. The digital coding used for DECT systemis Adaptive Differential Pulse CodeModulation (ADPCM) using 32kBit/s.In micro-cellular systems with small cellsizes and moving mobile stations thenumber of cell changes are counted.Because of the substantial signallingoverhead it is best to keep the number ofhandovers as low as possible. The controlsystem uses a decentralised handoveralgorithm steered by the mobile station

Fig 7: Blind slots for a transceiverwith two connections and apermanently installed station.


233

that decide whether and when a handoveris necessary. A seamless handover ispossible because the old channel is onlyleft if the new channel is already avail-able. The user does not notice anythingunlike the non-seamless handoverschemes. In order to protect DECT against unau-thorised interference and abuse ruleswere formulated that ensure a high levelof system security: • Identification of a participant • Avoidance of unauthorised use of the

mobile station • Identification of a permanently in-

stalled station • Avoidance of unauthorised use of

permanently installed stations • Illegal listening to user or signalling

dataThe following safety precautions, similarto those used in public portable radiosystems, are part of the standard system:A pin enquiry can be implemented whenthe mobile station is switched on. Amobile station has an international sub-scriber identification that is clearly de-fined in a certified range. Each mobilestation is examined before the connection

is establishment in order to prevent unau-thorised network access. In addition anauthorising key is stored in the mobilestation and in the base station that iscompared by an identification procedure.Thus a mobile station is protected againstbeing manipulated by a permanently in-stalled station to access the network. Toprotect other users from illegal listeningto the user or signalling data the radiointerface uses coded transmission [24].

3.Comparison and evaluation

A direct comparison of the wirelesstransmission systems from chapter 2 incomparison to the similar FM voicetransmission can made on the basis of theperformance of standard receiver sensi-tivities. This is a measure of the qualityof the demodulated signal from thesmallest possible RF signal; it also deter-mines the best transmission range. Forspeech transmission, based on similartypes of modulation, the signal quality isexpressed as signal/signal-to-noise ratio.For digital transmission the Bit ErrorRate (BER) is used to determine thequality. The Bit Error Rate is the rela-tionship of the number of bits that wereincorrectly demodulated to the totalnumber of bits sent. A bit error rate of 6* 10-6 means that on the average 6 bitscan be wrong if 1 million bits were sent.A detailed search of the receiver sensi-tivities in the literature showed that oftenthey couldn’t be compared because theBit Error Rates are defined differently.To be able to compare these systems onthe basis of receiver sensitivity, the re-ceiver sensitivities were converted for aBit Error Rate of 10-3. For DECT noconversions were necessary since thestandard Bit Error Rate is referred to 10-3.

3.1. Receiver sensitivities The BER for an AWGN channel has the

Fig 8: Un-usable channels because ofthe switching duration of the mobilestation channel measurement.


234

Where constants a, b and c constantsdepend on the modulation system and thesignal to noise ratio

with the signal energy of a bit Eb and thenoise performance density N0. Q () repre-sents the Q-function.

3.1.1. Q-function The Q-function {4} [15] is often used incommunication systems instead of thecomplementary error function. The prob-ability for an incorrect (BER) is definedwith the help of the Q-function.

The following are calculated values ofthe Q-function:

3.1.2. TetrapolModulation: GMSK BT = 0.25 Sensitivity is indicated for Tetrapol for aBER of 1.5% [2]. The error probability iscalculated according to [16]:

a ≈ 0.68 for BT = 0.25BER = 10-3:

BER = 1.5%:

3.1.3. GSM/GSM-R/GSM-BOSModulation: GMSK BT = 0.3 With GSM sensitivity for a BER of2.4%, based on a data sheet of the SonyEricsson K610 mobile telephone. In ac-cordance with [26] the error probabilityis computed:

d2min = 1.7874 for BT = 0.3

BER = 10-3:

BER = 2.4%:

3.1.4. TETRA Modulation: p/4-DQPSK (coherent).Sensitivity is defined for TETRA as aBER of 1.2% [1]. The error probabilityamounts to [16]:

BER = 10-3

( )bb cbQaP γ⋅⋅⋅=form {2}

0NEb

b =γ{3}

⎟⎠

⎞⎜⎝

⎛=22

1)( xerfcxQ {4}

10-5 4.26489 10-3 3.09023 2 x 10-3 2.87816 10-2 2.32635 1.2% = 0.012 2.25713 1.5% = 0.015 2.17009 2.4% = 0.024 1.97737 2 x 0.024 1.66456

Q(x) = BER x

⎟⎟⎠

⎞⎜⎜⎝

⎛ ⋅⋅=

0

2N

EQP bb

α{5}

dBbidBbb 46.8log1002.768.0223.09.3 12 =→=⇒⋅⋅= γγγ

dBdBbbb 39.5log1046.368.0217009.2 222 =→=⇒⋅⋅= γγγ

Correction factor = dBdBbdBb 07.321 =−γγ

⎟⎟⎠

⎞⎜⎜⎝

⎛⋅⋅=

0

min25.0

NEdQP b

b {6}

⎟⎟⎠

⎞⎜⎜⎝

⎛⋅=⋅→⎟

⎟⎠

⎞⎜⎜⎝

⎛⋅⋅=

0

min2

0

min2 25.0

NEdQP

NEdQP b

bb

b

dBdBbbb 66.6log1063.47874.187816.2 111 =→=⇒⋅= γγγ

dBdBbbb 90.1log1055.17874.166456.1 222 =→=⇒⋅= γγλ


( )bb QP γ⋅= 2 {7}

dBdBbbb 79.6log1077.4209023.3 111 =→=⇒⋅= γλγ


235

BER = 1.2%:

3.1.5. Interpretation In Table 2 all systems described inchapter 2 are compared based on receiversensitivity as described in chapter 3.1.The systems are arranged according totheir BER. It can be seen that none of thedigital systems approaches the good re-ceiver sensitivity of a similar FM voicetransmission. This shows that FM voicetransmission has receiver sensitivity bet-ter than the state of the art problem freesystems. It can be seen that the narrowband systems Tetrapol, TETRA and DIISexhibit the best receiver sensitivities ofthe digital systems. This is because thesensitivity of a receiver is directly relatedto the noise of the received spectrum.Receiver sensitivity of similar transmis-sion modes is referred to by a certainsignal/signal-to-noise ratio of the de-modulated signal and often defined asSignal, Noise and Distortion to Noise andDistortion (SINAD). SINAD describesthe logarithmic relationship of signal tonoise performance in dB, where distor-

tions and all other defects are considered.A further comparison assumes that for asimilar voice transmission with a BER of10-3 corresponds to 20dB SINAD.

3.2. Maximum range The maximum possible attenuation be-tween transmitters and receivers is shownin Table 3. It shows the result of thedifference in transmitter power with re-ceiver sensitivity corrected for a BER of10-3.

amax = PSmax - PEmin {8}where amax is the maximum attenuation indB, PSmax is the maximum transmitterpower in dBm and PEmin is the receiversensitivity in dBm. For the case forobstacle-free propagation between thetransmitter and receiver

amax = a0 {9}with the free apace attenuation a0dB andthus the maximum range rmax

in metres when using isotropic antennas.

3.2.1. Interpretation Table 3 shows what is evident from

Table 2: System types and receiver sensitivity.

Receiver -94dBm -115dBm -104dBm -111dBm -110dBm -102dBmsensitivity BER= BER= BER> 1.2% 1.5% 2.4%Modulation GMSK FM p/4 GMSK CP- GMSK DQPSK 4GFSKCorrection +2.7dB +3.1dB approx. +4.8dBfactor +2.5dB for BER=10-3

Corrected -94 -115 -101.3 -107.9 -107.5 -97.1sensitivity BER=10-3bBm

DECT FM TETRA Terapol DIIS GSM, GSM-R speech GSM-BOS

Reference BER= 20dB ----------------- BER = 10-2 ---------------------standard 10-3

dBdBbbb 06.4log10225713.2 222 =→⇒⋅= γγγ


20max

max

104

a

r ⋅=πλ

{10}


236

Table 2 that none of the digital systemsreaches the range of the similar system.The theoretical range for FM voice trans-mission is high because on the one handthe transmitting power is higher in thecomparison to for instance DECT and onthe other hand due to the different freespace attenuation of the frequency bandsused. For an objective comparison of the sys-tems a more exact analysis is required.

3.3. Comparison for the samefrequency and same transmittingpower For this comparison the maximum rangewas compared using the equation {10}that uses the free space attenuation underideal conditions. The frequency rangeused for this comparison was not therange used by the systems but the 4m and2m bands that are used by BOS. Theresults are shown in Table 4. Table 5 shows the systems after standard-ising them all to a transmit power of 5W.

3.3.1. Interpretation The translation of the frequency rangesused by BOS shows the DECT standardcannot be compared because its band-width requirements are too large. GSMwould be possible, however the numberof channels is very small and one GSMchannel would use the worst frequency. In systems like TETRA, Tetrapol, DIISand GSM it can be seen in Table 3 thatcompared with the values in Table 2 thata substantially higher range is to beexpected with these systems on the badfrequency ranges. The systems are easierto compare if the transmitting power isstandardised. For this standardisation anoutput of 5W was selected because handheld radios are specified with a maxi-mum transmitting power of 5W. Thiscomparison of the systems on the samefrequency and same transmitting powerare shown in Table 5. In this comparisonthe FM voice transmission is clearly thebest for range. The conclusion from this analysis is thatsystems that have many decades of tech-

Table 3: System comparison on the basis maximum possible range.

DECT FM speech TETRA Tetra DIIS GSM or BOS pol GSM-R GSM-BOSCorrected -94 -115 -101.3 -107.9 -107.5 -97.2sensitivity BER=10-3dBm

Permitted transmitter powerBase station W 0.25 25 25 25 25 25Base station dBm 24.0 43.98 44 44 44 44Mobile W 0.25 5 1 1 5 2Mobile dBm 24 336.99 30 30 37 33Maximum attenuation between transmitter and receiverBase station dB 118 158.98 145.3 151.9 151.5 141.2Mobile dB 118 151.99 131.3 137.9 144.5 130.3Maximum range due to free space attenuationUsed freq. MHz 1890 150 75 390 390 390 900Wavelength m 0.16 2 4 0.77 0.77 0.77 0.33Range km 10 6328.6 12657.1 224.1 482.3 1026.4 86.4


237

nological development are to be pre-ferred. In addition similar radio connec-tions do not fail abruptly like digital

transmissions at a given Bit Error Ratethreshold.

Table 4: System comparison through frequency bands used by BOS.

DECT FM TETRA Tetrapol DIIS GSM speech GSM-R GSM-BOS2m band Not 25 & 92 Possible Possible Possible Possible(150MHz) max. possible channel 19 & 72 38 & 145 38 &145 2 & 9bandwidths bandwith channel channel channel channel0.48MHz too large & 1.82MHz

Range in free 126.12 6328.56 582.62 1254.02 2668.73 518.23space km

4m band May be 164 Possible Possible Possible Possible(75MHz) possible channel 130 260 260 max 16bandwidth no more channel channel channel channel3.26MHz than 1

channel


4m band May be 164 Possible Possible Possible Possible(75MHz) possible channel 130 260 260 max 16bandwidth no more channel channel channel channel3.26MHz than 1

channel


2m band Not 25 & 92 Possible Possible Possible Possible(150MHz) max. possible channel 19 & 72 38 & 145 38 &145 2 & 9bandwidths bandwith channel channel channel channel0.48MHz too large & 1.82MHz


DECT FM TETRA Tetrapol DIIS GSM speech GSM-R GSM-BOS

Table 5: System comparison for the same frequency and same transmittingpower.


238

4.Literature

[1] "TETRA fact sheet"; BAKOM -Federal Office for communication Swit-zerland; Version 1.4; 18.04.2001;http://www.ai.ch/ dl.php/de/20040719095714/ Fak tenbla t t -TETRA.pdf[2] “Tetrapol fact sheet"; BAKOM -Federal Office for communication Swit-zerland; Version 1.3; 26.03.2001;http://www.polyverlag.ch/dok/tetrapol.pdf[3] Benkner T., Stepping C.; “UMTS -Universal Mobile TelecommunicationsSystem”; J Sclembach specialised pub-lishing house; Edition 1; 2002; ISBN 3-935340-07-9 [4] David K., Benkner T.; “Digital port-able radio systems”; B.G. Teubner pub-lishing house Stuttgart; Edition 1; 1996;ISBN 3-519-06181-3 [5] “Operational and financial evaluationof TETRA, Tetrapol and GSM-900 Plat-form for a digital BOS portable radionet”; DIALOGUE CONSULT GmbH;Duisburg; Version 1.3; 15.03.2004;http:// www.iwi.uni-hannover.de/lv/seminar_ ss04/www/Martin_Bretschneider/bibliography/GerpottWalter04.pdf[6] “discus - which? Why? When? “;DIIS; 03.08.2006;h t t p : / / w w w . d i i s . o r g / g e r m a n /Article_01_ger.htm[7] Dijkstra S., Owen F.; “Everythingspeaks for DECT”; Phillips Telecommu-nication Review; Volume 51; No. 2; pp.41-45; August 1993 [8] Fonfara M.; “GSM Rail“; 28.07.2006;http://www.senderlisteffm.de/ gsm-r. HTML[9] Gabis M.; “Radio communication foremployment support - feasibility study

for the employment of new radio tech-nologies with fire-brigade employments”J.K. University of Linz; Thesis(diploma); 2007 [10] Greneche D., Wiederspahn F.;“Costs save”; Funkschau 1/2004; Side32f; WEKA technical periodical publish-ing house GmbH[11] Ketterling H.P.; “Digital operatingradio with DIIS “; WEKA technicalperiodical publishing house GmbH;Funkschau: 8/2000; Poing; 2000;h t t p : / / w w w . f u n k s c h a u . d e /heftarchiv/pdf/2000/fs08/fs0008046.pdf[12] Kreher R. Rüdebusch T.; „UMTSSignalling: UMTS interface, Protocols,Message flow and Procedures Analysedand Explained “; Wiley & Sons; Edition1; 2005; ISBN 0-470-01351-6 [13] Mouly M., Pautet M. - B.; “TheGSM system for mobile Communications“; Cell & Sys self-publishing house;Palaiseau France; 1992; ISBN: 2-9507190-0-7 [14] “Push to talk“; Nokia AustriaGmbH; 03.08.2006;http://www.nokia.at/german/phones/technologies/push_to_ talc/index HTML[15] Proakis J.G.; “Digitally Communi-cation”; Mcgraw Hill college; Edition 3;1995; ISBN 0-07-051726-6 [16] Rappaport T.S.; “Wireless commu-nication”; Prentice Hall PTR; Edition 1;1996; ISBN 0-13-375536-3 [17] Rauscher C.; “Basis of spectrumanalysis”; Rohde&Schwarz GmbH; Edi-tion 2; 2004; PW 0002.6629.00 [18] “Siemens Communications Lexi-con”; Siemens Enterprise Communica-tions GmbH & Co KG;h t t p : / / n e t w o r k s . s i e m e n s .de/solutionprovider/_online_lexikon/ [19] Springer A.; “Portable radio technol-ogy”; Lecture script; Institut for commu-nications technology/information tech-nology; University of Linz; Edition 5;WG 2003/04; March 2003


239

from the bit to multi-carrier modulation - basics of ...the-eye.eu/public/books/electronic...

Documents