Exploit Network Coding Over GF(2^q) for Multi-user Cooperative Wireless Networks

Abstract

In this paper, we present a decode-and-forward network coded (DFNC) scheme over GF(2^q) for the multi-user cooperative communication systems. In particular, we consider a cooperative network with m users transmitting independent packets to the same destination. These users form a cooperation set to help each other by using linear network coding. We propose a coding coefficients construction method which can efficiently reduce the transmission overhead from m(q + log₂ m) to m bits compared with conventional random network coding. Furthermore, we propose a novel decoding algorithm—credit-based updating algorithm in order to improve the solvability of decoding set of equations at the destination. The proposed decoding algorithm is combined with channel decoding and is applied on symbol-level. It can fully make use of the error recovery property of network coding while conventional decoding algorithms (e.g., Gaussian elimination) overlook it. We theoretically analyze the diversity performance in terms of information outage probability, and the results show that diversity order of m + 1 can be achieved for a m-user cooperation system. Moreover, we conduct extensive simulations to show that DFNC outperforms other transmission schemes in terms of symbol error rate and achieves higher diversity order. We also demonstrate that the proposed decoding algorithm provides significant performance gain over conventional decoding algorithm.