On The Capacity Bounds And Rateless Network Coding Design Of Cooperative Relay Networks

In the past decades, the rapid development of wireless communication has a profound effect on our daily life. Today’s wireless networks have to face the explosive growth of popularity of user equipments and their demands for data transmission. One potential solution is cooperative communication, which can effectively enhance the link reliability as well as increase the sys-tem throughput and coverage. Being a key role in the next generation communication standards (e.g., Long Term Evolution-Advanced), cooperative communication has received tremendous in-terests in the research community. However, there remains quite many theoretic and practical issues to be addressed for cooperative communication. The fundamental limits of cooperative relay networks, under the framework of information theory, are still open problems. Moreover, the practical coding schemes to achieve the promised gains of cooperative relaying are far from fully investigated. Network coding, originally proposed for noiseless networks, has shown its significant advantage when applied to the cooperative relay networks. Nevertheless, in order to combat the fading, interference and noise, which are inevitable in wireless cooperative relay networks, network coding should be incorporated with channel coding to work properly. As a highly compatible channel code, rateless code, has been found synergistic matching in coopera-tive relay networks. By merging the two, we propose a class of flexible and robust joint network and channel coding schemes with good performances.In this dissertation, resorting to multiuser information theory and coding theory, we investi-gate the capacity bounds as well as joint network and channel coding design in cooperative relay networks. The contents of this dissertation are listed as follows.In the first part of this dissertation, we revisit the non-restricted half-duplex Gaussian amplify-and-forward two-way relay channel (AF-TWRC). The inner and outer bounds of its capacity region are derived by evaluating Shannon’s inner and outer bounds for general two-way channels. We prove that, in contrast to the case of Gaussian two-way channel (GTWC) where the two bounds coincide thus establishing its capacity region, the two bounds of AF-TWRC do not coincide unless both terminals are noise-free, due to that the intermediate relay node introduces additional power constraint relating to the correlation of channel inputs from both terminals. This indicates that the capacity region of non-restricted Gaussian AF-TWRC, which was previously believed to be already established in a way similar to that of GTWC, still remains open. We further investigate a special case of non-restricted AF-TWRC where the relay is noise-free, and obtain its capacity region based on the tighter dependence balance bound.In the second part, we propose a three-stage rateless coded protocol for a half-duplex time-division two-way relay system, where two terminals send messages to each other through a relay between them. In the protocol, each terminal takes one of the first two stages respectively to encode its message using rateless code and broadcast the result until the relay acknowledges successful decoding. During the third stage, the relay combines and re-encodes both messages with a joint network-channel coding scheme based on rateless coding which provides incremen-tal redundancy. Together with the packets received directly in previous stages, each terminal then retrieves the desired message using an iterative decoder. The degree profiles of the specific rateless codes, i.e., Raptor codes, implemented at both terminals and the relay, are jointly opti-mized for both the AWGN channel and the Rayleigh block fading channel through solving a set of linear programming problems. Simulation results show that, the system throughput as well as the error rate achieved by the optimized degree profiles always outperforms those achieved by the conventional degree profile optimized for Binary Erasure Channel (BEC) and the previous network coding scheme with rateless codes.In the third part of this dissertation, we consider an asymmetric time-division multiple ac-cess relay system consisting of two sources, one relay and one destination, where the channel conditions and message lengths of the two sources are allowed to be different. To enhance the link robustness and the system throughput, joint network-channel coding (JNCC) is employed with a specially designed rateless code which conducts both the channel coding and network coding simultaneously. In particular, at the sources, messages are rateless coded and then broad-casted to the relay and destination. While at the relay, a novel two-dimensional (2-D) LT code is proposed, which jointly encodes the precoded message bits of the two sources using a2-D degree profile and at the meantime completes the network coding inherently. Interestingly, the proposed scheme can be degraded to several conventional JNCC schemes based on rateless cod-ing. To further approach the theoretical limit, the corresponding degree profiles implemented at both sources and the relay are jointly designed based on the extrinsic information transfer (EXIT) function analysis. Simulations show that our proposed JNCC scheme with the optimized degree profiles outperforms other JNCC schemes with the conventional profiles both on the BER and throughput performances.