Font Size: a A A

Performance Analysis And Optimization Of Complex Field Network Coding (CFNC) System Under The Non-orthogonal Multi-access Relay Channels

Posted on:2014-08-29Degree:DoctorType:Dissertation
Country:ChinaCandidate:X CaiFull Text:PDF
GTID:1268330428475765Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
The multiple-access relay channel (MARC) represents a typical scenario of cooperative transmissions for uplink cellular networks wherein multiple sources deliver their messages to one common destination with the assistance of one or multiple relays. There are two main MARC schemes based on the source information transmission mode, i.e. orthogonal and non-orthogonal MARC, where the non-orthogonal MARC can attain higher throughput than that of orthogonal one. Complex field network coding (CFNC) uses operations over complex fields and distinguishes the signals by the constellation space of the "summed signal". At the receiver, the received signal is decoded by maximum likelihood detection algorithm. Moreover, the CFNC system with non-orthogonal MARC further improves the throughput. However, existing works have ignored the design of CFNC coding coefficients, relay forwarding strategies, power allocation and coding gain analysis, which are essential to the system design in uplink cellular networks, thus the main focus of this thesis.Firstly, by analyzing the instantaneous and statistical symbol error probability (SEP) upper bound for the CFNC system with non-orthogonal MARC, the optimal CFNC coding coefficients at the source and relay nodes are presented for arbitrary number of source nodes, together with some specific optimal CFNC coding coefficients. The results unveil that the optimal CFNC coding coefficients at the relay nodes greatly improve the SEP performance by eliminating the superposition of signals. Besides, adaptive and non-adaptive optimization algorithms for CFNC coding coefficients design at the source nodes are proposed. Adaptive optimized CFNC coding coefficients at the source nodes improve the instantaneous SEP performance and hence improve the average SEP performance. Non-adaptive optimized CFNC coding coefficients at the source nodes improve the SEP performance at higher correlated multiple-access channels.Secondly, for non-orthogonal MARC, the SEP upper bound of the CFNC system with K-th optimal relay selection (KBS-CFNC) is analyzed, which consists of N source nodes, M relay nodes and a destination, then the diversity order, coding gain and optimal power allocation algorithms are obtained. The theoretical results show that the diversity order of KBS-CFNC is M-K+2and full diversity M+1can be achieved when K=1. The optimal power allocation can greatly improve the SEP performance of KBS-CFNC. The optimal power allocation allocates more power to the link with worse quality. When the soruce-relay link is worse than the relay-destination link, more power is allocated to the sources such that at least one of the relays could be successfully chosen. As the quality of relay link is improved (improve M or the source-relay quality), more power should be allocated to the relays such as to improve the relay-destination link. Moreover, more power should be allocated to the source nodes as K increases, such as to improve the success choosing probability of the optimal relay node given the number of relays M.Thirdly, based on the adaptive decode-and forward (SDF) relay strategy, the SEP upper bound, diversity order and coding gain of three canonical CFNC systems with parallel relay (PR-CFNC) are analyzed. SDF relay strategy can avoid error propagation when the destination owns only single-hop channel side information (CSI). For the type-Ⅰ with PR-CFNC scheme, the throughput is1/2sym/S/CU since NM symbols are transmitted in2M time slots. For the type-Ⅱ with PR-CFNC scheme, the throughput is1/2sym/S/CU since N symbols are transmitted in2time slots. For the type-Ⅲ with PR-CFNC scheme, the throughput is1/(M+1) sym/S/CU since N symbols are transmitted in M+1time slots. Furthermore, type-Ⅰ and type-Ⅲ can achieve full diversity order M+1while type-Ⅱ achieves diversity order2. The implementation complexity of the three PR-CFNC schemes decreases from type-Ⅰ, type-Ⅱ to type-Ⅲ. The SEP performance of type-Ⅲ is the best among the three PR-CFNC schemes, and type-Ⅰ is worse than type-Ⅲ but better than type-Ⅱ.The optimal power allocation of type-Ⅲ with PR-CFNC scheme is analyzed by minimizing the SEP upper bound. Both the theoretical and simulated results show that the optimal power allocation can improve the SEP performance. In the optimal allocation, in order to improve the SEP performance of type-Ⅲ with PR-CFNC scheme, more power should be allocated to the relays when the quality source-relay channel is improved (M or the source-relay quality is increased, N is decreased, or the modulation level at the source nodes is decreased).Fourthly, the CFNC scheme with channel coding for the non-orthogonal MARC is studied. Channel coding with soft-in soft-out (SISO) decoding is equipped at the source and relay nodes, the iterative decoding structure and the corresponding Max-Log-MAP iterative decoding algorithm are developed. Note that the SISO iterative decoder is applicable for the whole network and multiple transmission error from different links can be corrected.Finally, pilot design for multiple-access relay channel is studied and the pilot symbol placement including the start point and the interval is provided. By analyzing the Cramer-Rao lower bound, we show that the selection of start point will not affect the estimation of frequency offset, and increasing the interval will increase the accuracy of frequency offset estimation. The gain over frequency offset estimation by increasing the interval is named as "placement gain". Then the effective SNR is obtained by uMMV algorithm and it is shown that the "placement gain" comes from the increase of the effective SNR, the higher the effective SNR, the smaller the estimation error. Both the theoretical analysis and simulations show that the faster the channel varying, the higher the mean square error (MSE); the higher the SNR and the smaller the number of pilot symbols, then the smaller the MSE. Moreover, the proposed equally space placement method can reduce the number of pilot symbols and hence save the energy for transmitting the pilot symbols.
Keywords/Search Tags:complex field network coding, non-orthogonal multiple access relay channel, relay communication system, symbol error probability, power allocation, multi-user detection, adaptive decode-and-forward relay
PDF Full Text Request
Related items