Arman Farhang - New Waveform Candidates for 5G: Options and Opportunities

Transcript

1. New Waveform Candidates for 5G: Options and Opportunities Arman&Farhang& TRINITY

   COLLEGE D U B L I N June 16, 2015 Supélec campus de Rennes

2. 1" Impact of timing and frequency offsets on multicarrier waveform

   candidates for 5G

3. Waveform Candidates for 5G Waveform&Candidates& Linear&pulse&shaping& Circular&pulse&shaping& Universal&filtered&mul;carrier&(UFMC)& & Filter&bank&mul;carrier&(FBMC)&

   Generalized&frequency&division& mul;plexing&(GFDM)& Circular&filter&bank&mul;carrier&(CHFBMC)& 2"

4. Fig.&1.&UFMC&system&model&in&the&uplink&[1].& Linear Pulse Shaping •  Universal"filtered"mul2carrier"(UFMC)" [1]&V.&Vakilian,&T.&Wild,&F.&Schaich,&S.&Ten&Brink,&and&J.HF.&Frigon,&“UniversalHfiltered&mul;Hcarrier&technique&for&wireless&systems& beyond&lte,”&in&IEEE&Globecom&Workshops&(GC&Wkshps)&2013,&pp.&223–228.& 3"

5. Linear Pulse Shaping •  UFMC"signal"in"frequency"domain" Fig.&2.&UFMC&signal&representa;on&in&frequency&domain&and&its&comparison&with&OFDM&[1].& 4"

6. Linear Pulse Shaping •  Filter"bank"mul2carrier"(FBMC)" Fig.&3.&Filter&bank&mul;carrier&(FBMC)&system&model.& 5"

7. Circular Pulse Shaping Fig.&4.&Filter&bank&mul;carrier&with&circular&pulse&shaping.& 6"

8. Circular Pulse Shaping Fig.&5.&TimeHfrequency&overlapping.& 7"

9. Circular Pulse Shaping •  OFDM"vs."GFDM"or"CJFBMC"data"packet" OFDM" GFDM"or"CJFBMC" 8"

10. Linear vs. Circular Pulse Shaping Fig.&6.&Linear&FBMC&transmit&signal.& Fig.&7.&Circular&FBMC&transmit&signal.& 9"

11. Impact of Synchronization Errors on Waveform Candidates for 5G Receiver"window"

    User"1" User"2" User"3" User"4" Fig.&8.&Timing&misalignment&between&different&users.& Fig.&9.&Frequency&misalignment&between&different& users.& 10"

12. Timing and Frequency Misalignment Transmit"signal"of"user" Received"signal"at"the"base"sta2on" Signal"of"user"""""aTer"going"through" the"channel" TransmiUed"symbols" es2mated"at"the"base"

    sta2on" 11"

13. Sensitivity to Timing Offset Fig.&10.&Mul;ple&access&interference&(MAI)&as&a&func;on&of&;ming& offset&for&different&waveforms.& 12"

14. Sensitivity to Frequency Offset Fig.&11.&Mul;ple&access&interference&(MAI)&as&a&func;on&of& frequency&offset&for&different&waveforms.& 13"

15. Sensitivity to Frequency Offset Fig.&12.&Amplitude&spectrum&of&the&receiver&matched&filter&(MF)&in&CHFBMC&and&zeroHforcing&(ZF)&detector& in&GFDM.& 14"

16. Sensitivity to Timing and Frequency Offset Fig.&13.&Bit&error&rate&(BER)&performance&of&different&waveforms.&The& normalized&TOs&and&CFOs&are&selected&randomly&between&H0.5&and&+0.5.& 15"

17. Sensitivity to Timing and Frequency Offset Fig.&14.&BER&performance&of&different&waveforms.&The&users&are&quasiH synchronous&in&;me&and&the&CFO&errors&are&selected&randomly&between& H0.5&and&+0.5.& 16"

18. Conclusions •  To"reduce"sensi2vity"to"2ming"and"frequency"offsets,"windows" with"smooth"edges"should"be"applied"to"both"transmiUer"and" receiver." •  Among"all"the"waveforms,"FBMC"and"UFMC"par2ally"sa2sfy"this" condi2on."" •  OFDM,"GFDM,"and"CJFBMC"fail"our"tests"as"they"lack"windowing"

    in"their"conven2onal"form."However,"improvements"are" possible"in"these"waveforms." 17"

19. 18" Frequency Spreading Equalization in Multicarrier Massive MIMO

20. New Waveforms and Massive MIMO for 5G 19" Waveform&Candidates& Linear&pulse&shaping&

    Circular&pulse&shaping& Filter&bank&mul;carrier&(FBMC)& & Universal&filtered&mul;carrier&(UFMC)& Generalized&frequency&division& mul;plexing&(GFDM)& Circular&filter&bank&mul;carrier&(CHFBMC)&

21. •  Massive"MIMO:"a"mul2user"system"similar"to"code" division"mul2plexing"(CDMA)"systems" Base" sta2on" MT1" . " . "

    . MTK" . " . " . H11" HKM" 1" M" 20" New Waveforms and Massive MIMO for 5G

22. •  CMT"modula2on" (a)&Spectra&of&baseband&data&streams&(black)&and&ves;gial&side&band&(VSB)&por;on&of&each&(other&colors).&(b)&CMT& spectrum&consis;ng&of&modulated&versions&of&the&VSB&spectra&of&the&baseband&data&streams.&VSB&signals&are&modulated& to&the&subcarrier&frequencies&f 0 ,&f 1 ,&"#"#"#,&fN−1 .&

    [1]&B.&FarhangHBoroujeny&and&C.&(George)&Yuen,&“Cosine&modulated&and&offset&qam&filter&bank&mul;carrier&techniques:&a& con;nuousH;me&prospect,”&EURASIP#Journal#on#Applied#Signal#Processing,#2010,#special#issue#on#Filter#Banks#for#Next# GeneraEon#MulEcarrier#Wireless#CommunicaEons,&vol.&2010,&p.&16&pages,&2010.& 21" Cosine Modulated Multitone (CMT)

23. 22" Frequency Spreading implementation of CMT

24. 23" Frequency Spreading implementation of CMT

25. •  SelfJequaliza2on"property"of"FBMC,"[2],"makes"it"a"viable"candidate"for" MIMO"applica2on." " •  CMT"offers"the"following"advantages"over"OFDM:" –  Higher""bandwidth"efficiency" –  Lower"sensi2vity"to"CFO"

    –  Lower"PAPR" –  More"flexible"carrier"aggrega2on" –  Blind"channel"equaliza2on"capability"enabling"pilot"decontamina2on,"[3]" 24" Filter Bank Multicarrier for Massive MIMO ! [2]&A.&Farhang,&N.&Marchej,&L.&Doyle,&B.&FarhangHBoroujeny,&“Filter#Bank#MulEcarrier#for#Massive#MIMO”,&In&Proc.&Of&IEEE& VTCHFall&2014,&Vancouver.& [3]&A.&Farhang,&&A.&Aminjavaheri,N.&Marchej,&L.&Doyle,&B.&FarhangHBoroujeny,&“Pilot#decontaminaEon#in#CMTMbased#massive# MIMO#Networks”,&In&proc.&of&ISWCS&2014.&Barcelona.&

26. 25" Minimum Mean Square Error Frequency Spreading Equalization Number of

    receive antennas Number of users

27. 26" Minimum Mean Square Error Frequency Spreading Equalization MMSE estimates

    of MMSE filter tap weights Spreading matrix Phase adjustment matrix

28. •  Single"user"case" 27" Numerical Results (a)&and&(b)&compare&the&signal&to&interference&ra;o&(SIR)&performance&of&the&MF&linear&combining&technique&for&the& cases&of&8&and&16&subcarriers,&respec;vely,&for&different&number&of&receive&antennas,&Nr .&&

29. Numerical Results SIR&performance&&comparison&between&polyphase&implementa;on&(PPN)&and&frequency&spreading&FBMC&systems&having&16& subcarriers&&for&different&number&of&receive&antennas.&

30. •  Mul2user"case" Numerical Results Signal&to&noise&plus&interference&(SINR)&performance&of&MMSE&linear&combining&for&the&case&of&having&16&subcarriers&and&6&users& where&the&receiver&input&signal&to&noise&ra;o&is&H1dB.&

31. •  An"effec2ve"MMSE"equaliza2on"scheme"for"FBMCJbased" massive"MIMO"systems"was"derived." •  Frequency"spreading"equaliza2on"enables"us"to"widen"the" subcarrier"bands"further"than"what"was"proposed"in"[2]."" •  Further"widening"the"subcarrier"bands"in"frequency"brings" improvements"in"terms"of"bandwidth"efficiency,"robustness"to" carrier"frequency"offset,"peakJtoJaverage"power"ra2o"and"

    latency"compared"with"polyphase"based"FBMC"systems." Conclusions [2]&A.&Farhang,&N.&Marchej,&L.&Doyle,&B.&FarhangHBoroujeny,&“Filter#Bank#MulEcarrier#for#Massive#MIMO”,&In&Proc.&Of&IEEE& VTCHFall&2014,&Vancouver.&

32. Thank"you" Any"comments"or"ques2ons?"