performance evaluation of multi layered space frequency
play

Performance Evaluation of Multi-Layered Space Frequency Time Codes - PowerPoint PPT Presentation

Performance Evaluation of Multi-Layered Space Frequency Time Codes for MIMO-OFDM Systems Dr. Samir Al-Ghadhban Assistant Prof. EE Dept, KFUPM, Saudi Arabia http://faculty.kfupm.edu.sa/ee/samir NCTT-MCP08 Aug, 2008 Outline Background


  1. Performance Evaluation of Multi-Layered Space Frequency Time Codes for MIMO-OFDM Systems Dr. Samir Al-Ghadhban Assistant Prof. EE Dept, KFUPM, Saudi Arabia http://faculty.kfupm.edu.sa/ee/samir NCTT-MCP08 Aug, 2008

  2. Outline • Background and motivation • IQ-Space Frequency Time codes • Multi-Layered STBC vs VBLAST • Multi-Layered SFT Codes NCTT-MCP08 Alghadhban 2

  3. Introduction: Multiple Input Multiple Output (MIMO) Channels 1 1 • A MIMO channel is a wireless link between M T 2 2 transmit and M R receive antennas. Scatterers Tx Rx • MIMO channels boost the information capacity M M T R of wireless systems by order of magnitude [Telater95][Foschini98].   h t ( ) … h ( ) t 11 1 M   T =  H ( ) t � � �    h ( ) t � h ( ) t   M 1 M M R R T NCTT-MCP08 Alghadhban 3

  4. Introduction: Open Loop MIMO Communication Systems Open Loop MIMO Communication Systems Transmit Spatial Diversity Multiplexing Differential Block Coding Block Coding Trellis Coding [Tar98] D-BLAST [Fos96] V-BLAST [Wal99] [Tar00] [Ala98][Tar99a] B 1 � B 1 B 2 � STBC B STBC � 2 B B 1 Combiner and 1 G V-BLAST 1 Detector Detector � B − K 1 B K-1 � B MIMO Fading K Channel NCTT-MCP08 Alghadhban 4 MIMO Fading B Channel K

  5. Multi-layered STBC is a single user system that consists of K parallel STBC • It combines spatial STBC B 1 G 1 multiplexing with � B 1 transmit diversity. SGINC • It is a V-BLAST Detector system with STBC on � B K each layer. STBC B K G K MIMO Fading Channel NCTT-MCP08 Alghadhban 5

  6. How does MLSTBC compare to V-BLAST and STBC? STBC V-BLAST B 1 � B B 1 2 � B 2 STBC STBC � B B 1 Combiner and 1 V-BLAST G 1 Detector Detector � B − K 1 MIMO Fading B K-1 � MLSTBC B Channel K MIMO Fading B Channel K STBC B 1 G 1 � B 1 SGINC Detector � B K STBC B K G K MIMO Fading Channel NCTT-MCP08 Alghadhban 6

  7. Comparison of MLSTBC and V-BLAST over 4x4 MIMO-OFDM, N c =64 and L =4 0 0 0 10 10 10 V-BLAST-OFDM BPSK V-BLAST-OFDM QPSK V-BLAST-OFDM 16QAM 1/2 rate STBC-OFDM 256QAM MLSTBC-OFDM 16QAM MLSTBC-OFDM 256QAM MLSTBC-OFDM QPSK -1 10 -1 -1 10 10 -2 10 -2 -2 10 10 OFDM SER OFDM SER OFDM SER -3 10 -3 -3 10 10 -4 10 -4 -4 10 10 -5 10 -5 -5 -6 10 10 10 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 E s /N 0 E s /N 0 E s /N 0 NCTT-MCP08 Alghadhban 7

  8. Motivation • Pervious work on MLSTBC over MIMO- OFDM systems didn’t take advantage of the available frequency diversity. • Our Goal is to design MLSTBC system that takes full frequency diversity advantage over MIMO-OFDM channels. • The solution is to add space frequency time (SFT) codes at each layer. NCTT-MCP08 Alghadhban 8

  9. Design criteria of SFT codes • The maximum diversity available in MIMO-OFDM systems is M T LM R [Ben Lu 2000] . • The design criterion is to maximize the minimum effective length and break up channel correlation in frequency domain by interleaving. • To achieve this diversity, the minimum effective length of the SFT code should be equal to at least M T L , which needs large number of states for practical values. • For example, at M T =2 and L=3, we need 1024 states. And at L=4, we need 16384 states NCTT-MCP08 Alghadhban 9

  10. Design criteria of SFT codes • Our goal is to simplify the design and reduce the number of states required to achieve the full spatial and frequency diversity. • Our approach is based on concatenating trellis coded modulation (TCM) and space time block codes (STBC). [Lateif 2003] • Spatial diversity is guaranteed by STBC and frequency diversity is provided by TCM. • We further reduce the number of states of TCM by using IQ-TCM [AlSemari 97]. NCTT-MCP08 Alghadhban 10

  11. IQ-TCM [AlSemari97] • The minimum effective length of • Thus, when k is reduced by TCM is upper bounded by: a half, l min at most doubles and this is the reason behind the diversity increase of IQ- ≤ +   l v k / 1   min TCM. Where v is the number of memory elements and k is the number of inputs. 8-states 4AM-TCM 3 -1 -1 3 In: Interleaver -3 -1 1 3 De: De-interleaver [ ] s ∈ − -3 1 1 3 1 -3 1 bps/Hz I I-TCM rate h* 1/2 � -3 1 s 4 -A M S ig n a l S e t { } h s ∈ I 1 6 - Q A M I-Dec -1 3 + In × De � s Q 3 -1 1 bps/Hz Q-Dec Q-TCM rate j 1/2 -3 1 [ ] s ∈ − -3 1 1 3 2 bps/Hz IQ-16QAM-TCM Q 1 -3 NCTT-MCP08 Alghadhban 11

  12. 2 bps/Hz Comparison • 8-states 8PSK-TCM: v =3, k =2 � l min =2 ≤ +   l v k / 1   min • 8-states IQ-16QAM-TCM: v =3, k =1 � l min =4 8-states 4AM-TCM 3 -1 -1 3 In: Interleaver -3 -1 1 3 De: De-interleaver [ ] s ∈ − -3 1 1 3 1 -3 1 bps/Hz I I-TCM rate h* 1/2 � -3 1 s 4 -A M S ig n a l S e t { } h s ∈ I 1 6 - Q A M I-Dec -1 3 + In × De � s Q 3 -1 1 bps/Hz Q-Dec Q-TCM rate j 1/2 -3 1 [ ] s ∈ − -3 1 1 3 2 bps/Hz IQ-16QAM-TCM Q 1 -3 NCTT-MCP08 Alghadhban 12

  13. IQ-SFT Alamouti STBC Code In= Interleaver P/S= Parallel to Time Time CP= Cyclic Prefix Serial converter b 2 1 s 1, I I-TCM 1, I 1   − * s   s s   2 2,1 1,1   −  s *  1 s + In   − = 1,2 s * = IFFT P/S 2,2  s CP s  b 1 � 2 �   1, Q   1, Q Q-TCM j   − s s *         1, N N c c 2, Nc Encoder STBC b s 1   s * 2, I I-TCM  s  2, I 2   2,1 1,1   s *  s  s   = 2,2 IFFT P/S s * = s CP 2 1,2 + In   2 � 1 �   b s     s *   2, Q 2, Q s   Q-TCM j     2, N N c 1, Nc c I-Dec (R e) Tim e 2 Tim e 1 1 s � 1 Q -Dec   (Im ) 2 t STBC y   1 t y Decoder 2 1,1   1,1 Rem   Com biner at t y S/P FFT t 1 De y   2   t = 1,2 I-Dec Y CP each = 1,2 1 (R e) t Y 2  1  � 1 � subcarrier   �   s t  y  2 N t  1  Q -Dec y   2 (Im )   1, L c 1, L NCTT-MCP08 Alghadhban 13

  14. Advantages of concatenated IQ-TCM-STBC at 2bps/Hz FCS Minimum number of states to achieve full diversity Length ( M T LM R ) L Tarokh STTC 8PSK-STBC IQ-16QAM-STBC QPSK 2 64 4 2 3 1024 16 4 4 16384 64 8 5 262144 256 16 6 4194304 1024 32 7 67108864 4096 64 NCTT-MCP08 Alghadhban 14

  15. The discrete received signal over T time slots at the i th subcarrier is M R : total number of receive antennas N G : number of transmit antennas per group M T : total number of transmit antennas = + Y H S V i i i i  S  1, i   S IQ -SFT   2, i =   + H H � H V O FD M    G  1, i 2, i K i , i D em odulator � 1   M ulti-Layered S   K i , IQ -SFT D ecoder S k,i is the k th STBC at the i th IQ -SFT O FD M D em odulator layer. G K H k,i is the M R × N G MIMO matrix from group k to the receiver at the i th subcarrier. NCTT-MCP08 Alghadhban 15

  16. Due to the short code length of STBC, the received signals over T slots are rearranged into a vector ˆ = + STBC y Hx η B 1 G 1   x � B 1 1   x    ˆ ˆ ˆ  = 2 + H H � H η    SGINC 1 2 K  � Detector   x   K � B K STBC x k is the symbols of the k th layer. B K G K MIMO Fading Channel ˆ H ⋅ × is the MIMO M T N k G matrix from group k to the receiver. NCTT-MCP08 Alghadhban 16

  17. Serial Group Interference Nulling and Cancellation (SGINC) • Group interference nulling : Based on an ordering criterion, assume that the first detected group is the k th group. Then, the algorithm calculates the orthonormal bases of the null space of: ˆ ˆ ˆ ˆ   =  � � H H H H H  − + k 1 k 1 k 1 K H N • Denote the orthonormal bases of the null space of by , k k then the received signal for the i th group after nulling is: � = = + � � y N y H x η k k k k k � H Where is the post-processing channel matrix. k NCTT-MCP08 Alghadhban 17

  18. SGINC � = � H � x H y • STBC Combiner: k k k • IQ-SFT Decoder • Group interference cancellation : After Decoding the k th Layer, its contribution is subtracted from the received signal and the processing is repeated serially for each group. • Ordering: – MaxMin FN – MaxAverage FN – Blind power allocation • Number of receive antennas should be greater than or equal to number of layers. NCTT-MCP08 Alghadhban 18

Recommend


More recommend