Improved successive interference cancellation algorithms for mimo detection

Abstract

Error propagation in successive interference cancellation (SIC) based receivers degrades the bit error rate (BER) performance of MIMO systems. Ordering of the sequence of detection of transmit vector has been successful in mitigating its effects on BER performance to a certain extent. Recently, multi feedback successive interference cancellation (MF-SIC) has been proposed for further mitigation of error propagation in SIC based detection. In MF-SIC, reliability of each detected soft estimate of symbol in the transmit vector is found at every layer of transmit vector by shadow area constraint (SAC) criteria. For a reliable soft estimate, quantization is followed to give the detected symbol. For an unreliable soft estimate, multi feedback loop is initiated. Multi feedback loop computes M nearest constellation points to the soft estimate called as neighbouring symbols and using the partial transmit vector detected so far, M partial transmit vectors are computed. SIC is run on these M partial transmit vectors to give complete candidate transmit vectors. Optimal transmit vector out of these partial transmit vectors is found based on ML criteria and its corresponding neighbouring symbol is fedback as the detected symbol for the unreliable soft estimate. However, while deciding the best symbol from neighbouring symbols, the SAC is not checked in the sub levels of MF loop which may result in an unreliable decision. Our work, improved multi feedback successive interference cancellation (IMF-SIC), investigates on the reliability nature of soft estimates in those sub levels of MF loop. SIC in MF loop is replaced by MF-SIC to take the reliability nature of symbols at sub levels too into account. This modification results in recursive computation of MF loop, making the algorithm complex. Complexity is checked by limiting the number of recursions by nRecmax. Subsequently LLR based reliability ordering is employed to IMF-SIC to obtain ordered multi feedback successive interference cancellation (OIMF-SIC). Simulation results reveal that the proposed algorithms outperform the existing algorithms such as the conventional SIC, the ordered SIC and the MF-SIC in terms of bit error rate (BER), and approaches near optimal performance. For 4 4 and 8 8 MIMO system with 4-QAM scheme, when IMFSIC is employed near optimal performance is obtained. For 16 16 MIMO system with 4-QAM scheme, when OIMF-SIC is employed near optimal performance is obtained. At higher order modulation techniques like 16-QAM, for 4 4 MIMO system, with IMF-SIC, an SNR gain of 1 dBat BER of 1:5 10􀀀3 is obtained over MF-SIC. For 8 8 MIMO systems along with IMF-SIC, an SNR gain of 2 dB at BER of 10􀀀2 is obtained over MF-SIC. With OIMF-SIC, for 4 4 and 8 8 MIMO system, an SNR gain of nearly 1 dB at BER of 2 10􀀀3 and 2 dB at BER of 10􀀀3 is obtained respectively over IMF-SIC.