1. MIMO-OFDM with antenna selection
July 2003
MIMO-OFDM with antenna selection
Heejung Yu, Taehyun Jeon, and Sok-kyu Lee
ETRI
Wireless LAN Modem Research Team
Heejung Yu, ETRI

2. Contents System and signal model MIMO channel model and capacity
July 2003
Contents
System and signal model
MIMO channel model and capacity
General MIMO systems
MIMO systems with antenna selection
Simulation results
Conclusion
Heejung Yu, ETRI

3. System model July 2003 S/P IFFT Add CP Mapping Rayleigh Fading Channel
Remove
FFT
Det.
P/S
Demap
Tx ant 1
Tx ant 2
Tx ant N
Rx ant 1
Rx ant 2
Rx ant M
Heejung Yu, ETRI

4. Signal model Received Signal for one subcarrier
July 2003
Signal model
Received Signal for one subcarrier
H : Frequency domain channel matrix
(hi,j : channel from j-th Tx antenna to i-th Rx antenna)
P : Total Tx power
N : Total # of Tx antenna
x : Tx signal vector (1 x N)
y : Rx signal vector (1 x M)
n : AWGN noise vector (1 x M)
Heejung Yu, ETRI

5. July 2003
MIMO channel model
From IEEE /161r0-Indoor MIMO WLAN channel models
Rtx and Rrx are the receive and transmit correlation matrices
Hiid is a matrix of independent zero mean, unit variance, complex Gaussian random variables
Heejung Yu, ETRI

6. Channel capacity MIMO channel capacity H : (M  N) MIMO channel matrix
July 2003
Channel capacity
MIMO channel capacity
H : (M  N) MIMO channel matrix
M : # of rx antennas, N : # of tx antenna
I : identity matrix
Heejung Yu, ETRI

7. General MIMO-OFDM Channel correlation induces performance loss
July 2003
General MIMO-OFDM
Channel correlation induces performance loss
Geometrical characteristics
Antenna spacing
In PC-card, implement at most 4(?) antennas
Can achieve 4 x data rate
Cannot guarantee BER performance
Induce more retransmission
Loss in the throughput of WLAN
Heejung Yu, ETRI

8. July 2003
Solution
We use only some antennas out of all antennas in transmitter part.
Rx terminal : channel estimation and determine antenna selection method
Selection information transfer
Tx terminal : data transmission using selected antennas
Heejung Yu, ETRI

9. Selection methods Each sub-carrier base selection
July 2003
Selection methods
Each sub-carrier base selection
We select antennas in terms of sub-carrier
Selection pattern depends on a sub-carrier
Get high performance gain, but need more feedback information and hard to implement
Total sub-carrier base selection(Here, don’t consider)
We select antennas in terms of sum of all capacity corresponding to total sub-carriers
Get small gain, but need small feedback information and easy to implement
Heejung Yu, ETRI

10. Selection procudure Consider all combinations of Tx antennas
July 2003
Selection procudure
Consider all combinations of Tx antennas
Make a MIMO channel submatrix for each combination
Calculate Channel Capacity
Decide the antenna selection that maximize the capacity
Heejung Yu, ETRI

11. Simulation environments
July 2003
Simulation environments
Channel : Complex Gaussian flat fading channel (corresponding to one sub-carrier) with correlation
Capacity is calculated with more than 1000 channel generation
Same number of Tx and Rx antennas
Total Tx power(sum of all used antennas) is the same
Heejung Yu, ETRI

12. Simulation results (I)
July 2003
Simulation results (I)
Correlation
Heejung Yu, ETRI

13. Simulation results (II)
July 2003
Simulation results (II)
Correlation
Heejung Yu, ETRI

14. Conclusions Why use selection?
July 2003
Conclusions
Why use selection?
Small capacity gap
Detection complexity : matrix inversion( O(M^3) )
MIMO systems with antenna selection can achieve relatively high capacity with small detection complexity
Heejung Yu, ETRI

15. Further Work Initial channel estimation to decide antenna selection
July 2003
Further Work
Initial channel estimation to decide antenna selection
Antenna selection information transfer
Simplify antenna selection method
Heejung Yu, ETRI