Research On Efficient Relay Technology

In recent years, the fast growing applications of mobile internet and multimedia service have greatly boosted the industrial development of mobile wireless communication. To meet the ever increasing requirements of bandwidth and quality of service, a series of novel technologies need to be applied. Among them, distributed network scheduling based on relay techniques is deemed to be very promising for the next generation of mobile wireless networks. Relay techniques have recently been one of the hottest research spot. Various relay protocols have been designed and analyzed from the aspects of Shannon information theory, however very few of them focus on practical implementation to current wideband systems which typically consist of OFDM (Orthogonal Frequency Division Multiplexing), M-ary modulation and capacity approaching FEC (Forward Error Control) code. Therefore, in this work, we specifically researched efficient relay protocols which can be practically implemented to those key components, and analyzed their computation and storage complexities, signaling overhead, etc. Moreover, we proposed some optimized solutions based on DF (decode-and-forward) protocol.The main contribution and innovation are as follows:1) For classic three-node relay model based on OFDM, M-ary modulation and FEC code, we researched conventional DF and DQF (Decode-quantize-forward) protocols and proposed the code-level BER modified DF protocol (BMDF) and the symbol-level BMDF protocol. For code-level relay protocol, relay has to demodulate OFDM signal and perform full FEC decoding, while for symbol-level protocol, relay only need to demodulate OFDM signal and make hard decision in M-ary constellation. The proposed protocols need some side information at terminals, and it can be the channel coefficients of the link from sender to relay, or the BER which relay made, or clipping value of LLR (log-likelihood ratio). The implementation complexities of both proposed protocols are very low, and they perform better than the conventional DF, and also better than DQF when relay positioned closer to sender. Compared to the code-level BMDF, the symbol-level one reduce the implementation complexity even further with some sacrifice on performance.2) For linear multi-hop relay network based on OFDM, M-ary modulation and FEC code, we researched the symbol-level DF protocol and proposed a BMDF protocol and an improvement by employing subcarrier pairing(SP) and power allocation(PA). Because the demands of implementation complexity and latency are very strict for multi-hop network, symbol-level DF is more efficient for this model. The side information for proposed protocol can only be BER itself, as it should be calculated on each relay node iteratively. When signaling overhead is constrained in the system, we also proposed an effective method to save it. Even we only transmit a single side information for every hop, the performance gain is great. The employment of SA and PA further boosts the end-to-end performance.3) For two-way relay system based on OFDM, M-ary modulation and FEC code, we researched current DF, AF(amplify-and-forward) and DNF(denoise-and-forward) protocols, and proposed the BMDF and BMDNF strategies. Different from above one-way communication systems, relay performs network coding to save time and boost system’s throughput. Especially for 2-stage two-way relay DNF protocol, the network coding should be designed with the assist of pre-coding, and the BER calculation of relay is quite different and more complicated compared to one-way system. Side information for BMDF protocol could be channel coefficient, BER itself or clipping value of LLR, but for BMDNF with BPSK or phase aligned pre-coding QPSK, it can be BER or clipping value of LLR, and for BMDNF with phase-and-amplitude aligned pre-coding 16-QAM, no signaling is required. For 3-stage two-way relay, the proposed BMDF performs much better than DF, while for 2-stage, BMDNF performs much better than AF and DNF, and it gives the best throughput result among all analyzed protocols.