System Implications in Designing a 60 GHz WLAN RF Front-End

System Implications in Designing a 60 GHz WLAN RF Front-EndAli Behravan, Florent Munier, Tommy Svensson, Maxime FlamentThomas Eriksson, Arne Svensson, and Herbert ZirathDept. of Signals and Systems (S2) and Microtechnology Centre at Chalmers (MC2)Chalmers University of Technology,S-41296 Gothenburg, Swedenhttp://www.s2.chalmers.se http://www.mc2.chalmers.se/AbstractIn this paper, we seek to evaluate the performance ofa 60 GHz WLAN system taking into account RF circuitryimperfections and hardware requirements for variousmodulation techniques. A model of an RF frontendis developed, including physical imperfections of thecircuitry such as power amplifier (PA) nonlinearity andvoltage controlled oscillator VCO phase noise. Given theRF front-end model, several modulation techniques suchas Orthogonal Frequency Division Multiplexing (OFDM)and Continuous Phase Modulation (CPM) are considered.The evaluation of the system performance in termsof bit error rate allows a better understanding of physicalcircuitry limitations, and optimal modulation parametersas well as circuit design recommendations can be derived.KeywordsOFDM, CPM, 60 GHz, Power Amplifier, Phase noise.1. IntroductionThe need for high data rates communication systems hasbeen dramatically increasing in recent years [1]. The 60GHz band (59–64 GHz), an unlicensed frequency band,has been investigated as a potential band for wireless highdata rate transmission. One of its main properties is theexistence of strong attenuation due to the oxygen absorptionand obstacles, resulting in a good frequency reusefactor, but also limiting the coverage of the cells in thecellular communication network.Radio frequency (RF) hardware with ideal characteristicsis difficult to design for the 60 GHz band. Problemssuch as power amplifier (PA) non-linearity and oscillatorphase noise are more prominent for these circuits thanfor circuits designed for lower frequencies. Therefore,we should take these effects into account in the overallcommunication channel.This report is a first effort in an ongoing work aimingat evaluating the performance of various communicationtechniques in presence of hardware imperfections. Westudy how the performance of a system is affected by thenon-linearity of the amplifier, and the phase noise of theoscillators and phase-locked loops at the transmitter andreceiver. Even for a noiseless channel, the degradationdue to non-linearities and phase noise can become severe,leading to poor performance. We have chosen to studyan orthogonal frequency division multiplex (OFDM) system,due to its good performance for multipath channels.We also study a CPM system, which is more resistant tothe non-linearities of the power amplifier.In section 2, we give a general background on nonidealitiesarising from the RF front-end of a wireless tranceiveras well as a brief insight on modulation methodsemployed in the baseband back-end. Section 3 describesthe models of the RF hardware, including phase noise andpower amplifier non-linearity. In section 4, some experimentsare performed in order to characterize the effectsof the hardware imperfections for an OFDM and a CPMsystem.2.1. System description2. BackgroundFigure 1 shows the block diagram of the communicationsystem. In the baseband modulator block, input symbolsare modulated (using CPM or OFDM modulation)and fed into the analog RF front end. The signal is upconvertedto the desired frequency band, and then amplifiedto an adequate power level, to be transmitted throughthe channel. The receiver amplifies the input signal, anddown-converts it to baseband, where it can be demodulatedand further processed. Due to nonlinearities in thepower amplifier and phase noise of the VCO, the demodulatedbits may contain errors even on an noiseless channel.2.2. Front-end imperfectionsPower Amplifier (PA) linearity and Voltage ControlledOscillator (VCO) stability are critically important inpractical systems. If not considered, they may severelydegrade the performance of a digital communication system.These imperfections, depicted in Figure 1 as solidline boxes, are studied in this paper.

Baseband ModulationIF AmpMixerRF PALeeson’s noise spectrum model described in [6]. Moreon this can be found in section 3.2.Baseband Demodulation4X2X"Black Box" 60GHz VCO modelIF AmpSplitterMixerLNA60GHz ChannelFigure 1: 60 GHz transceiver front-end. Solid lines boxesinclude non-ideal properties studied in this paper. Thebaseband modulation investigated are both QPSK/OFDMand CPM schemes.2.2.1. Amplifier nonlinearitiesMany transmission systems and particularly multicarriersystems show a notable sensitivity to the non-linear effectscaused by the use of non-linear amplifiers. Thenon-linear distortions cause signal compression creatinginterference between subcarriers [2] called Inter-CarrierInterference (ICI). In order to maintain acceptable performancein presence of nonlinearities, large input back-offmay be required resulting in low power efficiency. Therefore,a study of the system is required to find a reasonabletrade-off between transmitted power and link performanceof the system.2.2.2. VCO phase noiseAnother important matter when it comes to design a wirelesstransceiver is the quality of the VCO. The amount ofphase noise present in the VCO is a measure of this quality,usually given as a ratio of power in one phase modulationsideband to the total power per unit bandwidth, ordBc/Hz at a certain offset from the carrier frequency.1One can see phase noise as a random change in the phaseterm of the oscillator.The existence of this perturbation leads to a mismatchedup/down-conversion of the signal. In multicarrierschemes like OFDM, phase noise is a particularlysensitive factor that causes subcarriers to interfere betweeneach other and hence increases the error probabilityof the communication link. Another effect of phasenoise is introducing some uncertainty in the modulatedsignal. Phase noise and more generally ½ « noise generationfor simulation has been extensively studied in [3]and [4]. Some results are reported regarding effects onOFDM systems in [5]. There are several ways of modelingthe phase noise process such as an AR filter basedmodel [3], or by shaping the frequency response of phasenoise according to some practical measurements using1 dBc denotes the decibels relative to the carrier power.2.3. Modulation methodsAmong the existing modulation methods, we focused onOFDM and CPM. OFDM issues in broadband wirelesssystems have been widely discussed [1] and considerationsfor 60 GHz channel have been reported in [7].CPM is seen as an attractive modulation technique due toits constant envelope that makes it less sensitive to nonlinearities.2.3.1. OFDMOrthogonal Frequency Division Multiplexing (OFDM) isa multicarrier transmission technique, which divides theavailable spectrum into a number of subchannels, eachone being modulated by a low rate data stream. OFDMuses the spectrum efficiently by overlapping the subchannelsin such a way that they are kept orthogonal [8].OFDM transmission schemes are particularly interestingin the presence of time dispersion of the channel due tomultiple paths arriving at different time. Another advantageis the possibility to adjust the data rate of each subcarrieraccording to their Signal-to-Noise ratio. On theother hand, OFDM presents several drawbacks of whichtwo of them are analyzed in this paper. First, the OFDMsignal is more sensitive to frequency mismatch and phasenoise compared to single carrier schemes. Second, thehigh peak-to-average power ratio of an OFDM signal requireslinear amplifiers, which tend to be costly and toreduce the power efficiency. Both effects deteriorate theorthogonality between subcarriers and introduce ICI.2.3.2. CPMIn Continuous Phase Modulation (CPM) the informationsymbols are transmitted by changing the phase of a carriersignal. The transmitted phase function is continuousover time for all symbol sequences, which makes the envelopeof the transmitted signal to be constant. Therefore,non-linear amplifiers are not a problem. In addition, thecontinuous phase function in CPM creates small spectralside lobes as compared to e.g. the discontinuous phasefunction in constant amplitude PSK. Each CPM symbolis transmitted by a phase function called the phase responsefunction, having an amplitude depending on theCPM symbol, and the transmitted phase is a superpositionof these phase functions. The total phase change foreach symbol depends on a parameter called the modulationindex, . A common notation for CPM with a linearphase response function of length one is ContinuousPhase Frequency Shift Keying (CPFSK). The continuousphase restriction and phase response functions longerthan one CPM symbol introduces finite memory in themodulation. Hence, the Viterbi algorithm is an optimum

eceiver for CPM on the AWGN channel. A standard referenceof CPM is [9].3. Modeling the front-end imperfectionsIn this section, we describe the models we have employedin order to simulate the front-end. We focus our attentionon the power amplifier in the transmitter, and on theoscillators in the transmitter and the receiver. We havederived equivalent baseband models of the amplifier nonlinearitiesand the phase noise.3.1. Baseband equivalent model of the non-linear amplifierThe non-linear characteristic of the amplifier can befound using single tone measurements. The nonlinearitiesare usually represented by a power series. The inputoutput relationship of a nonlinear amplifier can thus bewritten asݴص ´Ü´Øµµ ÆÒ½ Ò Ü Ò´Øµ (1)where ܴص is the modulated signal at the input of the amplifier,ܴص ´Øµ Ó×´¾ ¼ Ø · ´Øµµ (2)Both ´Øµ and ´Øµ are narrow band baseband signal.Writing ܴص in terms of its equivalent low-pass, andsubstituting in (1), it emerges that only odd terms in theseries contribute in the output around the desired frequencyband, and the equivalent input-output relation inthe baseband can be written asݴص ¾Æ ½Ñ¼ ¾Ñ·½¾ ¾Ñ ¾Ñ ·½Ñ ·½Ü´Øµ ¾Ñ·½ ´Øµ (3)where ܴص and ݴص are equivalent baseband input andoutput of the nonlinear amplifier.A similar procedure can be followed to find the effectof the intermodulation terms in the baseband [10] but thisis not considered in this study. The input-output characteristicof the nonlinear amplifier is shown in figure 2.The model was derived by fitting a power series to theamplitude and phase curves of a 60 GHz power amplifier.3.2. Baseband Representation of Phase NoisePhase noise originates from several sources; one partstems from instabilities in the transmitter oscillator, anotherpart from the receiver oscillator, but also Dopplereffects and channel noise passing through the phaselockedloop contributes to the total phase noise. In thisdocument, we generate the phase noise process ´Øµ asa Gaussian distributed random process with a variancePSD [dBc/Hz]0−20−40−60−80−100Pout (dBm)22201816141210864−4 −2 0 2 4 6 8 10 12 14Pin (dBm)Figure 2: AM-AM effect in nonlinear amplifier−12010 −1 10 0 10 1 10 2 10 3 10 4 10 5Frequency [Hz]Figure 3: Phase noise power spectral density.Ò ¾ . Its spectrum is defined by an AR filter whose transferfunction is given asÀ´Þµ (4)½ «Þ ½where and « are parameters used to control the cutofffrequency and the power of the phase noise. The spectrumof phase noise generated by this model is shown infigure (3).We also define a phase SNR asËÆÊ ½¼ÐÓ ½¼´ ¾ ¾ Òµ (5)where ¾ Ò is the phase noise power. Here it is important tonotice that a VCO with phase noise will have a resultingbandwidth which is not only varying with the phase noisepower but also with the phase noise process bandwidththat is dictated by the AR filter.Ideally, a complex baseband signal ܴص ´Øµ ´Øµis upconverted by a phase noise-free VCO with center

frequency ¼ and the output of the mixer is given byܴص ʴص ¾¼Ø·´Øµ (6)However, due to physical imperfections present in theVCO, the phase of the carrier is not fixed but rather affectedby a random phase noise process ´Øµ so that ܴصbecomesQ3210−1ܴص ʴص ¾¼Ø·´Øµ·´Øµ (7)The equivalent complex baseband modulation affected byphase noise isܴص ´Øµ ´Øµ·´Øµ (8)This process can be used to simulate the total phase noiseexperienced by the modulation. We may look at ´Øµas the combination of phase disturbances occuring inthe transmitter and receiver oscillators as well as disturbancesarising from the channel (e.g Doppler shift).4. Experiments4.1. Simulation setup4.1.1. OFDMThe system in figure 1 is used as the setup for this simulation.An OFDM signal with 256 subcarriers, each modulatedby 16QAM or DQPSK, is generated in the basebandmodulator. During upconversion phase noise is introducedby the oscillator, and the power amplifier willfurther distort the signal.4.1.2. CPMA CPM system consisting of a transmitter and a Viterbidetector performing coherent detection, combined witha simple phase estimator, has been simulated with thephase noise process. The CPM system is: binary CPFSKwith ½¾ (MSK) or 4-ary CPFSK with ½ andGray mapping, which gives 2 bits per CPM symbol.4.2. Results4.2.1. OFDMThe effect of amplifier non-linearity on the OFDM signalcan be seen by looking at the received signal constellationafter the FFT block. As shown in figure 4, due to therandomness of the nonconstant amplitude signal in eachsubcarrier interval, a distribution of signal points is received,instead of a single point. Also because of the amplifiergain compression, the center of these clouds getscloser to the origin. Another effect is the rotation of thesignal space due to AM-PM effect. The variance of thedistribution is a function of the number of subcarriers and−2−3−3 −2 −1 0 1 2 3IFigure 4: Received signal constellation for N=256 andIBO=3.27dB0.080.060.040.020−3 −2 −1 0 1 2 3AmplitudeFigure 5: PDF of the amplitude of OFDM signal with 256subcarriersthe center depends on the power of the transmitted signal.When the number of subcarriers is increased, a Gaussiandistribution will be a valid approximation for the distributionof the received points in the signal space (figure5). Therefore, the effect of the nonlinearity on the performanceof the system can be completely described by themean and the variance of the signal points.Using a Gaussian approximation, the bit error rate ofan OFDM system with 256 subcarriers in the presence ofa non-linearity has been computed. Note that because ofthe rotation of the signal space and the power compression,the received signal constellation is not symmetricaland the distance between adjacent points in the signalconstellation are not the same. Therefore we can not usethe standard methods of calculating bit error rate for 16QAM. The conditional symbol error probability must befound first and then the total symbol error rate can bederived by averaging the conditional probabilities. Figure6 shows simulation results for the bit error rate versusamplifier input back-off in a noise-free system. TheBER versus SNR for a system including AWGN channelisshowninfigure8.Figure 7 present BER results of an OFDM system usingDQPSK, and of a standard DQPSK system, for differentphase SNR as defined in equation 5. In this simulation,we inject a phase noise in the transmitted basebandsignal. We assume that phase noise is introduced by the

transmitter and receiver VCOs without contribution fromthe channel.the OFDM system is highly sensitive to amplifier nonlinearitiesand oscillator phase noise. The CPM system isimmune towards amplifier nonlinearities, and less sensitivethan OFDM towards phase noise.10 −2 IBO (dB)BER10 −410 −610 −8IBO=3.27IBO=3.9IBO=4.6IBO=5.36IBO=6.19Without NL10 −2 Eb/No (dB)10 −100 2 4 610 −4Figure 6: Bit error rate versus amplifier input backoff fora noise-free systemBER10 −610 0 Phase SNR (dB)10 −110 −810 −2Bit Error Probability10 −310 −410 −510 −610 −105 10 15 20 25 30QPSKOFDM−QPSK10 −7−10 −5 0 5 10 15 20 25 30Figure 8: Bit error rate versus SNR for different IBOsFigure 7: BER vs phase SNR for QPSK andQPSK/OFDM10 0 Phase SNR [dB]4.2.2. CPMIn the CPM system simulations, the phase noise processhas the same bandwidth as in the OFDM system simulations.Furthermore, the bandwidths of the CPM systemand the OFDM system are the same. This means thatwhen QPSK or DQPSK is used in the OFDM system and4-ary CPM is used in the CPM system, the bit rates arethe same.In figure 9 the bit error rate for CPM caused by thephase noise is shown. It is clearly seen that the 4-aryCPFSK is more sensitive to the phase noise, which is nosurprise since the phase trajectories are much more densethan in MSK. However, the bit rate is twice as high withthe 4-ary CPFSK as compared to MSK.Bit Error Probability10 −210 −410 −6MSK4−ary CPFSK h=1/410 −8−25 −20 −15 −10 −5 0 5Figure 9: BER vs. phase SNR for MSK and 4-ary CPFSK ½.5. ConclusionsWe have performed system simulations of OFDM, CPM,and single-carrier QPSK, including RF circuitry imperfections.The results show that with the assumed models,

6. References[1] Luis M. Correia and Ramjee Prasad, “An overviewof wireless broadband communication,” IEEE CommunicationsMagazine, pp. 28–33, Jan. 1997.[2] A. Vaccardi D. Dardari, V. Tralli, “A theoreticalcharacterization of nonlinear distortion effects inofdm systems,” in IEICE Transactions Communications.IEEE, Oct. 2000, vol. 48, pp. 1755–1764.[3] N.J Kasdin, “Discrete simulation of colored noiseand stochastic processes and 1/ « power law noisegeneration,” in Proceedings of the IEEE. IEEE,May 1995, vol. 83, pp. 802–827.[4] P.Balaban M.C Jeruchim and K.S Shanmugan, Simulationof Communication Systems, Plenum Press,1992.[5] M. Van Bladel T. Pollet and M. Moeneclaey, “Bersensitivity of ofdm systems to carrier frequency offsetand wiener phase noise,” in IEICE TransactionsCommunications. IEEE, feb,mar,apr 1995, vol. 43,pp. 192–193.[6] D.B Leeson, “A simple model of feedback oscillatornoise spectrum,” in Proceedings of the IEEE. IEEE,1966, vol. 54, pp. 329–330.[7] Maxime Flament, On 60 GHz Wireless CommunicationSystems, Licenciate thesis, Chalmers Universityof Technology, Gothenburg, Sweden, Dec.2000.[8] R. van Nee and R. Prasad, OFDM for Wireless MultimediaCommunications, Artech House, 2001.[9] J.B. Andersson, T. Aulin, and C.-E. Sundberg, DigitalPhase Modulation, Plenum Press, 1986.[10] K.G. Gard H. Gutierrez and M.B Steer, “Characterizationof spectral regrowth in microwave amplifiersbased on the non-linear transformation of a complexgaussian process,” in IEEE Transactions on MicrowaveTheory and Techniques. IEEE, July 1999,vol. 47, pp. 1059–1069.