Optimal Hybrid Frequency Hop Communication System Using ...

3426OPTIMAL HYBRID FREQUENCY HOP COMMUNICATIONSYSTEM USING NONLINEAR ADAPTIVE JAMMERCOUNTERMEASURES AND ACTIVE FADING MITIGATIONChad S. Bergstrom and J. Scott ChuprunMotorola Space and Systems Technology Group8201 E. McDoweU Road, MD H1175Scottsdale, Arizona, USA 85257-1417602-441-3278, 602-441-2584 (fax), ~21191 @email.mot.comABSTRACTDigital mobile battlefield environments imposechallenging obstacles to battle group communicationsand data management. As future communicationcapabilities cascade from tactical commandersto individual soldiers, increased spectral clutterwill complicate the task of preserving datathroughput. Applied within a programmable radioarchitecture, frequency hop (FH) and direct sequence(DS) waveforms have inherent strengthsthat may be exploited to overcome battlefieldcommunication obstacles. In this paper, we presenta methodology that combines FH waveforms withpulse concealment and transform domain signalprocessing in order to overcome interference, hostiledetection, and jamming. Rejection algorithmsare demonstrated which significantly improve systemperformance in the presence of high-powerpartial-band jammers, co-site communication signals,and fading.INTRODUCTIONRecent studies have shown that co-site and cochannelinterference pose especially formidableobstacles to effective battlefield communications[l].Even relatively low levels of interferencehave been found to interfere with mission criticalparameters, such as GPS.*Prepared through collaborative pdcipation in the Advanced Telecommunicatims& Information Distribution Research Program(ATIRP) Consortium sponsored by the U.S. h y Research Laboratoryunder the Federated Laboratory Rogram, Cooperative AgreementDAAU)1-96-2-0002. The US. Government is authorized to reproduceand distribute reptints for Government purposes notwithsmding anycopyright notation thereon.In order to combat ambient and hostile interference,time and frequency domain signal processingmethods may be applied within a spread spectrumcommunications architecture. In this manner, softwareconfigurahle radios use nonlinear signal processingmethods in concert with FH and DS waveformsto mitigate interference while preservinghigh data throughput, even in the absence of coding.The properties of DS waveforms inherently providefor interference rejection, low probability of intercept(LPI) and low probability of detection (LPD).However, practical power control limitations arewell known to restrict channel capacity [2,3]. Incontrast, FH systems are less dependent on powercontrol, yet can he easier to detect, locate, and jamunder certain conditions. Higher hop rates on theorder of 1-2 Khopskec enhance FH performancerelative to jammer and detection threats [4,5].However, DoD requirements for future communicationsystems include provisions for detection anddirection of arrival (DOA) capability [6]. Hence,any communication methodology used by U.S. andallied forces must be able to defeat similar detectionchallenges posed by enemy battle units. Inorder to meet the challenges posed by the futuredigital battlefield, we present a hybrid FWDS approachthat preserves LPI/LPD characteristics andtolerates severe interference environments.APPROACHFigure 1 illustrates an example FWDS hybridspectrum, including typical interference sources. Inthis paper, we focus on adaptive mitigation of partialband interference and jammers, in contrast to0-7803-4984-9/98/$10.00 0 1998 IEEE.

3428tics for each Wsn are similar to white noise. Assuch, perturbations in the channel serve to color ordistort the spectral characteristic from this ideal.Hence, by obtaining an estimate of the spectral envelope,and by using this estimate to inverse filterthe corresponding data, a whitened, or “restored”spectrum may be obtained in moderate interferenceconditions or severe fading conditions. In order toensure that the underlying phase information is undisturbed,an all-pole spectral envelope estimate isrequired. Equation 1 illustrates an all-pole spectralmodel, where ak represents the prediction coefficientsfor an Nth order model that is obtained usingclassic autocorrelation and durbin recursion [ 121.k=lOnce the ak prediction coefficients are computed,the inverse whitening filter is given by Equation 2,which provides the residual waveform. Hence, Eq.2 represents the countermeasure that is applied tothe received data x(n) by the adaptation processorin order to whiten the corrupted spectrum.Fig. 2 - QPSK Constellation in presence of partialband noise jammer, showing destruction of I,Qstate information, and restored QPSK constellationafter application of inverse filter countermeasure.In addition to the damaging effects of jammers andcosite signal interference, communication link integritycan suffer from shadowing and frequencyselective fading. Figure 3 illustrates the BER improvementthat is obtained using the inverse filtercountermeasure in the presence of frequency selectivefading having deep excursions, circa 40 dB.Note that the damaging effects of the fade arelargely removed by the inverse filter countermeasureover a range of channel noise levels. Resultingperformance is restored to near fade-free levels.Ny(n) =x(n)+Ca(k)x(n-k)k=lEq.2Using the inverse filter approach, a partial bandnoise jammer having bandwidth B = (0.25)Wsn isapplied to the hybrid FH/DS signal within the hopbandwidth. Using a reasonable number of predictioncoefficients, for example N = 24, the inversefilter is applied using Eq. 2. Without reduction indata throughput, the resulting spectral content iswhitened, restoring data that had been lost to jamming.To graphically illustrate the effect of the inversefilter countermeasure, Figure 2 shows theeffect of the partial band noise jammer on the dataQPSK constellation before and after the inversefilter mitigation. Note that the I,Q state informationis obliterated by the damaging effects of the partialband jammer. Following application of the inversefilter Countermeasure, the I,Q state information isrestored.Fig. 3 - BER improvement in frequency selectivefade environment after application of Inverse FilterCountermeasure.The inverse filter countermeasure provides improvementin BER performance at moderate tohigh Jammer/Signal (J/S) levels, while approaching

3429the asymptotic limit at low J/S levels. Hence, incontrast to other techniques which can induce significanterrors in relatively clean conditions, applicationof this method at low or negligible jammerpower does not induce undue BER performancedistortion. In comparison, the adaptive inverseweight countermeasure described below has similarasymptotic behavior, yet provides superior BERimprovement for high-power interference andjamming.Adaptive Inverse Weinht Countermeasure Thesecond countermeasure method considered for thispaper involves transform domain signal processingfor interference mitigation. Selective spectral limitinghas long proven effective for jammer and interferencemitigation 1131. We improve upon suchapproaches by computing modal spectral statisticestimates ,oi for each signal component i. Thesenon-arithmetic statistics may be used to selectivelytarget only those statistical outliers that exceedsome value a = F+ko of the fundamental transmissionmode.Figure 4 illustrates a typical multimode statistic forthe FWDS signal in the presence of a partial bandnoise jammer. Using the transform domain magnitude,an adaptive interference rejection function isapplied that measures the modal statistics, determinesthe fundamental mode, and applies an interferencerejection weight w that is inversely proportionalto the observed deviation 6 relative to thedominant mode. In this approach, the statisticaloutliers are reduced by the weight given by Eq. 3,where m is an integer. In addition to minimizingthe statistical extremes, application of Eq. 3 to + nproximity samples provides further performanceimprovement by enhancing the attenuation of theinterference envelope.corrupted by partial band noise jammer.Figure 5 shows the significant BER improvementgained by the adaptive weight countermeasure inthe presence of QPSK interference encompassing aquarter bandwidth, (0.25)Wsn, with a channel SNRof -15 dB. Given identical interference bandwidths,the BER performance shown in Figure 5 is similarfor QPSK cosite and partial band jammer interference.As with the inverse filter countermeasure, theadaptive inverse weight method approaches theasymptotic limit at low J/S ratios. Note that theadaptive inverse weight dramatically improvesBER at high jammer power levels, resulting in 3Eq. 3band QPSK interference using Adaptive InverseWeight countermeasure.

3430The final interferer considered for the adaptive inverseweighting countermeasure is a four-tone hostilejammer. As with the partial band interferers,the adaptive inverse weight is effective for tonejammers even at very high J/S ratios. This approachis especially beneficial against narrow band interference,which is readily canceled without distortionof the underlying hybrid FWDS data. Figure 6shows the BER performance for this method inhigh-power tone interference. Note that the fourtone jammer is mitigated using the adaptivemethod, resulting in a near-flat BER correspondingto the -15 dB S NR used for this example.CONCLUSIONWe have presented a hybrid Frequency Hop DSPNtransceiver approach that preserves LPI/LPD characteristicsand mitigates severe interference. Thisapproach is especially applicable to communicationsfor small unit operations (SUO), where bothstealth and link integrity are essential [SI. In thedigital mobile battlefield, the severity of interferenceand jamming necessitates countermeasuresthat supplement inherent spread spectrum processinggain without reducing data throughput. Wehave demonstrated BER improvement using receiver-basedpreprocessing functions, including anadaptive inverse weight countermeasure basedupon spectral modal statistic estimates, and an inversewhitening filter countermeasure based on anall-pole spectral model. In contrast to adaptiveprocessing gain methods, both countermeasuresimprove BER performance without sacrificingsystem bandwidth. Furthermore, each approachexhibits orders-of-magnitude performance improvementrelative to prior-art full-band normalizationmethods. Of the two countermeasures, theinverse weight function provides superior BER improvementin the presence of high-power partialband jammers and communication signal interference.The adaptive inverse filter function furtherprovides a mechanism for improving BER performancein the presence of frequency selectivefades. Hence, a composite jdfade scenario willsignificantly benefit from a cascaded mitigationapproach that applies both methods to alleviate severeinterference and frequency selective fading.tone jammers using Adaptive Inverse Weightcountermeasure.REFERENCES[l] Joe C. Capps, US Army Signal Corps CommunicatorMagazine, Tactical Internet (EPLRS,SINCGARS SIP), TXFFI Lessons Learned Report,July 1, 1997.[2] R. Cameron and B. Woemer, “PerformanceAnalysis of CDMA with Imperfect Power Control”,IEEE Transactions on Communications, Vol.44, July 1996.[3] D. Torrieri, “Frequency Hopping and FutureArmy Mobile Communications”, Proceedings,ATIRP Annual Conference, University of Maryland,1997.[4] D. Torrieri, “Fundamental Limitations on RepeaterJamming of Frequency-Hopping Communications”,IEEE Journal on Selected Areas in Communications,Vol. 7, No. 4, May, 1989.[51 U.S. Government, “Contractor’s Design Handbookfor Protecting RF Transmissions”, Revision1, July 1994.[6] Joseph Mitola III, “Spectrum Supremacy F’rogramConcept”, Mitre Technical Report, Jan. 7,1997.[7] Kumpumaki, Isohookana, Juntti, “Narrow-BandInterference Rejection Using Transform DomainSignal Processing in a Hybrid FWDS Spread

3431Spectrum System”, Proceedings, Milcom ‘97,Monterey CA, November 1997.[8] B. Kaspar, LtCol, USAF, “SUO SAS Phase IIIndustry Brief‘, January 29, 1998.[9] Analog Devices, “White Paper on the TexasInstruments TMS320C6x”[lo] Susan Gilfeather and Scott Chuprun, “ComplexArithmetic Processor Performance Metrics onLPI Waveforms”, MILCOM ‘95 Classified Proceedings,November, 1995.[ 111 Motorola, “SPEAKeasy I1 - An IPT Approachto Software Programmable Radio Development”,Proceedings, Milcom ‘97, Monterey CA, November1997.[12] Deller, Proakis and Hansen, Discrete TimeProcessing of Speech Signals, p. 297, MacmillanPublishing Co., 1993.1131 Simon, Omura, Scholtz, Levitt, Spread SpectrumCommunications Handbook, p. 45 1, McGraw-Hill. 1994.*The views and conclusions contained in this document are those ofthe authors and should not be interpreted as representing the officialpolicies, either expressed or implied of the Amy Research laboratalyarthe U.S. Government