MIMO OFDM 2006. 11 충주대학교문철
발표자소개 연구분야 다중안테나시스템을위한채널모델링및다중안테나전송기술연구 무선자원할당기술연구 수행과제 ( 최근 3 년간 ) 2003~ 현재 : 차세대이동통신을위한다중안테나기술연구 ( 삼성전자 ) 2005 년 : 휴대인터넷을위한다중안테나기술 ( 삼성전자 ) 2006 년 : WiBro 시스템의 Hybrid SA/SDM/SDMA 운용방안설계 ( 삼성전자 ) 연구실적 ( 최근 3 년간 ) 특허 : 국내특허 6 건, 미국특허 4 건출원 논문 : SCI 논문 5 편게재, 국제학술대회논문 6 편발표. 3GPP2, WiBro MIMO 기술표준화 2/50
Wireless Channel Characteristics 3/50
Wireless Channel Frequency Selective Fading Time Selective Fading Space Selective Fading 4/50
Wireless Channel Frequency Selective Fading (Delay Spread) Power Delay Profile T s τ 4 T : ISI (Inter-Symbol Interference) s >> τ 4 : Negligible ISI ISI gets severe as the symbol duration decreases (data rate increases) Power Power Transfer Fucntion τ 1 τ τ τ 2 3 4 time Frequency Signal Symbol Duration T s 5/50
Wireless Channel Time Selective Fading (Doppler Spread) Doppler Effect Doppler Frequency v f λ cosθ d = θ 6/50
Wireless Channel Time and Frequency Selective Fading Channel 7/50
Wireless Channel Characteristics Space Selective Fading (Angular Spread) Spatial correlation effect I. Specular Channel - Urban with Line Of Sight - Higher antenna than surrounding building - BS height : 40m - Distance between Tx and Rx : 305.5m - Direct path angle : 50º 8/50
II. Scattering Channel - Urban with nonlos - Dominant Scatterer in front direction - Distance between Tx and Rx : 334.8m - Direct path angle : 60º 9/50
Space and Time Selective Fading Channel 5 5 signal level [db] 0-5 -10-15 signal level [db] 0-5 -10-15 -20 0.03 0.02 time [sec] 0.01 0 c 2 4 distance [d/λ] 6-20 0.03 0.02 time [sec] 0.01 0 0 2 4 distance [d/λ] 6 envelope fading of specular channel DOA = 0 o, AS = 10 o, f d = 100Hz envelope fading of scattering channel DOA = 0 o, AS = 90 o, f d = 100Hz 10/50
Multi-Antenna Technologies 11/50
Why multi-antenna technologies? Time-, frequency-, code-domain Time, frequency, and code domain processing are at limits Space Domain the final frontier Space processing is interesting because it does not increase ba ndwidth Multi-Antenna Technologies Specular channels Beamforming interference cancellation Scattering channels MIMO Systems Transmit diversity Receive Diversity 12/50
Multi-antenna 기술분류 송신기술 : Transmit diversity, MIMO 기술 수신기술 : Beamforming, receive diversity 기술, MIMO 수신기기술 Multi-antenna 기술동향 3G 이동통신기술의다중안테나기술표준화 3GPP LTE, 3GPP2 revision C: MIMO 기술표준화중. WiBro: smart antenna, transmit diversity, MIMO 기술표준화중. 4G 무선통신기술의핵심기술로서의 MIMO 기술부상 현재 cdma2000 1xEVDO 의 spectral efficiency 는최대 1.92 bps/hz, 100Mbps 전송을위해최소 50MHz 대역폭이요구됨. 따라서, 현재보다 5 배이상의 spectral efficiency 증가가요구됨. 이를위해서 MIMO 기술이요구됨. DVB-T 용다중안테나기술개발 BBC: 수신다이버시티, MIMO 기술개발 13/50
Classical Receive Diversity Maximal Ratio Combining at BTS 평균수신 SNR 증가 : 수신안테나수에비례하여증가. 공간다이버시티이득 : 페이딩을완화시켜에러성능개선. 현재사용기술. h 1 r= [h 1 h 2 ] T s+n y=[h * 1 h* 2 ][h 1 h 2 ]T s+ [h * 1 h* 2 ]n h 2 γmrc = (SNR) ( h 1 2 + h 2 2 ) CMRC = log 2 (1 + γ MRC ) [bps/hz] Capacity increases logarithmically with number of receive antennas... 14/50
Mobile Receive Diversity (1) Advantages Diversity gain improves received SNR: 2 branch diversity: 3~7dB gain Increasing spectral efficiency without upgrade of existing systems. 무선패킷시스템에서일반단말기보다좋은채널상태를가지므로더많은대역폭을할당받을수있음. 송신기가사용하는다중안테나기술에상관없이적용가능. ( 비표준화기술 ) 간단하면서도효과적으로용량개선. 3GPP, 3GPP2, WiBro, 지상파 DMB, 무선랜수신기에적용가능. MIMO 기술의 starting-line. h 1 h 2 15/50
Mobile Receive Diversity (2) Challenges Power penalty due to doubled RF chain and need for high computing power. RF chain 의증가로인한가격증가. 안테나소형화기술요구됨. 최대다이버시티이득을위해서는두 branch 가동일한이득을가지도록설계되어야함. 3GPP2 MSM5500 (Qualcomm) Mobile receive diversity for 1xEVDO 16/50
Transmit Diversity/Beamforming h 1 h 2 γ transmit diversity = (SNR/2) ( h 1 2 + h 2 2 ) γ beamforming = (SNR) ( h 1 2 + h 2 2 ) Beamforming (Transmit Adaptive Array) 3 db SNR increase if transmitter knows H=[h 1 h 2 ] T 단말이순방향채널정보를추정후기지국에 feedback 해야함. Transmit diversity (Space-Time Block Coding) 수신 SNR 이득없고다이버시티이득만존재. 순방향채널정보에대한 feedback 필요없음. 17/50
MIMO Antenna Technologies h 12 h 21 h 11 H h h 11 12 = h21 h 22 h 22 C MIMO = log 2 det[i +(SNR/2) HH ]= Where the λ i are the eigenvalues to HH SNR SNR = log 2 1+ λ 1 + log 2 1+ λ 2 2 2 Interpretation: Transmitter λ 1 λ 2 Receiver min(n r, n t ) parallel 공간 channel 형성, min(n r, n t ) 개의공간 channel에서로다른데이터전송 spatial multiplexing 18/50
Spatial Multiplexing MIMO 19/50
Optimal Spatial Multiplexing MIMO If transmitter knows H, - singular value decomposition of H H=UDV - precoding x by V y=hv H x+n - decoding y by U U H y=u H HV H x+u H n =Dx+n D=diag{λ 1, λ 2, λ 3, λ 4 } Transmitter p 1 p 2 p 3 p 4 - transmit power allocation to each pipe by Waterfilling principle - rate adaptation per pipe. λ 1 λ 2 λ 3 λ 4 Receiver 20/50
채널환경에따른 MIMO 용량변화 Specular Channel: small angular spread or LOS 환경 Only one pipe is used C beamforming = log 2 (1+SNR λ 1 ) [bit/(hz s)] Scattering Channel: large angular spread or NLOS 환경 C m = log 1 + MIMO 2 i = 1 SNR λi nt Capacity increases linearly with min(n r,n t ). 채널상태에따라 multiplexing order 조절필요. 21/50
MIMO 송신기술 (1) PARC (Per-Antenna Rate Control) 3GPP 표준화기술. 송신안테나별로독립적으로변조및코딩, 다중데이터 stream 동시전송. 채널상태에따라 multiplexing order 조절 수신기는 n t 송신안테나중에서실제데이터를전송할송신안테나결정. 2 nt -1 경우에대한용량계산요구됨. 송신안테나수증가할수록연산량증가. Spreading code 1 Substream 1 Data stream Coding, Mapping, substream 1 Substream 1 Coding, Mapping, substream 2 substream 2 Scrambling code Substream 2 Spreading code N 22/50
MIMO 송신기술 (2) PSRC (Per-Stream Rate Control) 송신 stream 별로독립적으로변조및코딩. 수신기 : 미리정해진제한된수의 precoding 행렬중에서최대용량을제공하는 precoder 선택후, precoder index feedback. 송신기 : 수신기로부터 feedback된 precoder로빔성형, 전송 2 jφ 1 A Ae W = [ w1 w 2 ] = jφ 2 Ae 1 A 채널환경에따라 multiplexing order 조절 23/50
MIMO 송신기술 (3) Precoder codebook 기술 (IEEE802.16e, WiBro) V(n T,n S,B)=V(4,1,B), V(4,2,B), V(4,3,B), V(4,4,B), B=3 or 6 codebook 정의. n T : no. of Tx ant. n S : no. of Tx stream. 2 B : no. of precoders per codebook Uncorrelated Rayleigh MIMO 채널행렬을 eigen decomposition 후, n S 개의 principal eigenvector 들로구성된 precoder 를 quantization. Grassmannian subspace packing 수신기는 2 B n T 개의 precoder 들중에서최대용량을제공하는 precoder 를선택하여, 송신기로해당 precoder 의 index 를송신기로 feedback. 송신안테나수증가할수록연산량증가. Feedback 정보량을줄이기위해수신기의연산량을증가시킴. 24/50
MIMO 송신기술 (4) Unitary precoded MIMO (3GPP, 3GPP2) 25/50
Scheduling 에따른 MIMO 기술 Single-user MIMO vs. multi-user MIMO User selection Single-user MIMO 기술 Stream selection Multi-user MIMO 기술 26/50
MIMO 수신기술 (1) Optimum Detector: Maximum Likelihood Detector bˆ ˆ b 1 2 = arg 2 { } r1 E h1,1 h2,1 b s 1 min constellation r2 2 h1,2 h2,2 2 b1, b2 b 16 QAM : 256 times computations per modulation symbol interval before making a decision 송수신안테나수가증가하거나높은 order 의변조방식이사용될수록복잡도가기하급수적으로증가, 구현불가능. Sphere decoding 27/50
MIMO 수신기술 (2) Vertical Bell-Labs Layered Space Time architecture (V- BLAST) n r n t required. Symbol by symbol detection. Using n ulling and symbol cancellation. Detecting the strongest data stream, subtracting that stream from the remaining data streams. Algorithm complexity is linear with the number of transmit antennas. Error propagation 현상 한 symbol 추정시발생한에러가다음 symbol 추정에영향을줌. Antenna Time s1 s1 s1 s1 s1 s1 s2 s2 s2 s2 s2 s2 s3 s3 s3 s3 s3 s3 V-BLAST 성능열화. 28/50
MIMO 수신기술 (3) Linear 수신알고리즘 y=hx+n x =w H y=w H Hx+w H n 수신알고리즘이간단. 연산량이적음. Nonlinear receiver (ML decoding, V-BLAST) 에비해상당한성능열화. Minimum mean square error (MMSE) W MMSE H = H nt HH + I SNR Zero forcing (ZF) W ZF = H + 29/50
MIMO 기술분석 Challenges of MIMO technology Cost vs. capacity gain Antenna 수에따라비용은선형적으로증가. 그러나, 실제시스템용량은 antenna 수에선형적으로증가하지않음. 현재, MIMO 시스템의 capacity gain 에대한 evaluation 단계. 상용화여부에대해서는신중. Hardware implementation 수신기의중요성및복잡도가기존 SISO 에비해상당히증가. 수신기가채널환경에최적인송신구조결정. 복조알고리즘. MIMO system 용량이수신기의복조알고리즘및송신구조결정알고리즘에상당히좌우됨. 수신기 operation 은표준화대상이아니기때문에, 수신기제조사에따른성능차별화대두. OFDMA + MIMO, MIMO + scheduling OFDMA 및무선자원할당기법과의 MIMO 기술의최적결합에대한연구선행되어야함. 30/50
Orthogonal Frequency Division Multiplexing 31/50
OFDM: Overview OFDM Concepts Parallel transmission using multiple carriers Transmits a block of N source symbols in parallel by using N carriers The effect of channel time dispersion (ISI) is greatly reduced Minimum inter-carrier spacing Chosen so that waveforms on different carriers N subcarriers are orthogonal in time, but overlap in frequency Spacing Δf =W/N... f 32/50
OFDM: Overview OFDM Properties Robustness to frequency selectivity, One-tap equalizer Bandwidth efficiency due to the overlapping orthogonal subcarriers FFT implementation Simultaneous elimination of ISI and ICI (by Cyclic Prefix) Multiple users by using subcarrier division multiple access Adaptability to channel conditions Sensitive to frequency offset Large PAR 33/50
OFDM: Overview Spectrum of normal time domain sequence Spectrum of classical frequency division multiplexing (FDM) Spectrum of Orthogonal Frequency Division Multiplexing 34/50
OFDM: Guard band Guard band is the last Tg seconds of the active symbol period prefixed to the waveform, making it a cyclic prefix Must be kept short, a fraction of T, yet longer than the channel impulse response Completely eliminates ISI, ICI Maintains subcarrier orthogonality Time waveform appears periodic to the receiver 35/50
Multipath Fading 36/50
Spectrum is multiplied by the frequency response of the channel Some frequency bins are attenuated, others are amplified 37/50
OFDM System using DFT/FFT In the Reciever, the FFT acts as a bank of filters, where the values of the resulting frequency bins become the signal constellation points. 38/50
OFDM Challenges: Nonlinear Amplifier Nonlinear distortion of carrier signal due to AM/AM and AM/PM characteristic of HPA (High Power Amplifier) Causing ISI (Inter-Symbol Interference) and ACI (Adjacent Channel Interference) Input Back Off (IBO) in order to operate in the linear region OFDM has high PAPR (Peak to Average Power Ratio) compared to Single Carrier System Output Power Back-off(dB) Output Phase (degree) 39/50
OFDM Challenges: Nonlinear Amplifier Constellation QPSK, SSPA (E b /N o = 16dB) 16QAM, TWTA (E b /N o = 16dB) 40/50
OFDM Challenges: Timing Offset Timing Offset Case 1) No Timing Offset Case 2) Timing Offset within GI QPSK,16QAM Constellation (E b /N o = 20dB) QPSK,16QAM Constellation before Channel Estimation (E b /N o = 20dB) 41/50
OFDM Challenges: Timing Offset Case 3) Timing Offset out of GI ISI (Inter Symbol Interference) Constellation (AWGN, E b /N o = 20dB) No Timing Offset Timing Offset out of GI (after Channel Estimation) 42/50
OFDM Challenges: Frequency Offset Due to Oscillator mismatch or Doppler Shift Breaking orthogonality performance degradation QPSK Constellation (E b /N o = 20dB) Frequency Spectrum of OFDM signal ε ε 43/50
OFDM based Broadcasting Technologies OFDM based Broadcasting Technologies Digital Video Broadcasting-T Digital Audio Broadcasting Eureka-147 Korean T-Digital Mobile Broadcasting Digital Video Broadcasting-H 44/50
OFDMA (Orthogonal Frequency Division Multiple Access) OFDMA 다중접속방식. WiBro, 3G 이동통신 evolution 에사용. 45/50
Multi Antenna Technologies for Broadcasting 46/50
Diversity Reception of DVB-T (BBC) Receive diversity MRC with two antennas Mobile test Mode Portable test Percentage of seconds with errors No diversity 45% Diversity 20% Measurement Building1 Building2 Mean signal gain 3.9dB 3.3dB END improvement 0.9dB 1.8dB *END=0.21*amplitude response ripple, db, pk-pk 47/50
MIMO for Broadcasting (BBC) MIMO for terrestrial broadcasting Initial over-air transmission tests have confirmed the basic principles. To be feasible within a broadcast network, any major changes to a broadcast transmission standard would have to be agreed through standardisation. MIMO for high capacity studio links Initial simulations deliver a 72 Mbit/s in an 8 MHz bandwidth (4 transmitting and receiving antennas). three times the capacity currently available and suitable for high definition cameras. 48/50
MEA Technologies for Broadcasting Receive Diversity 추가적인표준변경없이구현가능. MRC combining 사용할경우, 평균수신전력이수신안테나수에비례하여증가. Open-loop MIMO 각송신안테나별로 pilot 채널필요. 표준변경이요구됨. BLAST 와같이동시에전송되는데이터스트림들을분리하는수신부신호처리부개발필요. Closed-loop MIMO 나 BF 기술은 broadcasting 에적용이불가능함. 49/50
The End