Low-complexity iterative soft detection for LDPC coded multi-relay channels

Official URL: http://192.168.1.20/record=b1301353 (Table of Contents)

Next generation wireless communication applications require reliable transmission of data at high data rates and a guarantee of quality-of-service over wireless links. However, degradations inherent in wireless channels, such as multipath fading, shadowing, path loss, and noise lead to reduction in the communication capacity and range significantly. One way to combat these adverse limitations is to employ spatial diversity, which can be achieved, for example, by transmitting independent copies of the signal over relay nodes, resulting in improvements in the transmission rates, reliability, and the capacity of the channel under pre-mentioned detrimental effects. In addition to exploiting diversity, the capacity of the channel can be further increased by employing an error correction code such as low-density parity check (LDPC) codes and turbo codes, etc. Throughout this thesis, we consider LDPC coded full-duplex multi-relay channels using Estimate and Forward (EF) and Decode and Forward (DF) protocol. We focus on designing optimal and sub-optimal iterative soft detectors. Although the use of multirelaying improves the channel reliability, the performance of the system is degraded because of the interference caused by multiple received signals coming from all relay nodes. To reduce the effect of the interference, maximum a posteriori (MAP) detector can be employed. Unfortunately, the complexity of the MAP detector grows exponentially as the number of relays increases. In the literature, two computationally efficient sub-optimal detectors have been proposed based on Taylor expansion or Central Limit Theorem (CLT) assumption to alleviate this problem. However, we find out that the correlation between intrinsic and extrinsic information stemming from these suboptimal detectors is very high, and this correlation degrades the detector performance. To remedy that, in this thesis, we developed two new detectors: Soft Decorrelating Detection-Taylor (SODED-Taylor) and Soft Decorrelating Detection-CLT (SODEDCLT), which improves the performance of sub-optimal detectors about 0.8 dB - 1 dB.