Study On Decode-and-forward Relaying In Ultra-reliable And Low-latency Communications

As one of the three major application scenarios of 5G,Ultra-Reliable and LowLatency Communications(URLLC)brought severe challenges to the current communication systems.End-to-end ultra-low latency compresses the transmission time,then finite blocklength transmission has to be utilized.Thus,infinite blocklength-based channel capacity will not be able to achieve accurate evaluation of system performance in URLLC.At the same time,under severe time-frequency resource constraints,it is particularly important to consider the overhead of each part in the communication process and to optimize the allocation of transmission resources.On the other hand,relaying systems widely exist in existing communication networks,among which Decode-andForward(DF)relaying is indeed suitable for the applications in URLLC scenarios since it can avoid error propagation.Using finite blocklength coding theory and adopting the global error probability as the URLLC performance indicator,this paper discusses the performance and the corresponding optimizations of DF one-way relay and DF two-way relay in URLLC,respectively.Firstly,under the perfect CSI assumption,the one-way relaying realizes the theoretical minimum global error probability by optimizing the adopted transmit rates and blocklength of two hops.This problem is simplified and transformed,and we derive the closed expressions of the optimal transmission configuration and the corresponding optimal performance.Then,this paper discusses the performance optimization of DF oneway relaying under estimated CSI,as well as the optimization of training sequence.The derived closed expressions are utilized to solve the above two problems.Finally,the paper discusses the optimization of one-way relaying under statistical CSI and proves the convexity of the problem.For DF two-way relaying,it is divided into the multi-access phase in uplink,and the superposition-network code-based broadcast phase in downlink.The convexity of the uplink(downlink)error probability with respect to its code length(code length,power allocation ratio)is proved.Meanwhile,a simple approximation is proposed for the error probability in the downlink phase.Then,the blocklength and power allocation optimization problem under perfect CSI assumption is solved,and the theoretical best performance is thus achieved.Then,the optimization of blocklength-and-power allocation under estimated CSI for DF two-way relay is discussed,and the optimal design of the training sequence is also carried out.Finally,the performance optimization problem of DF two-way relaying under statistical CSI is solved.Based on the above results,for different CSI strategies,the best performance under static channels and the best long-term performance under block-fading channels are obtained,respectively.The numerical results verify the effectiveness of the optimizations in this paper and the accuracy of the proposed closed expression and approximation.At the same time,it is revealed that the ideal performance and actual performance of DF relaying in URLLC.In addition,the impact of online optimization and decoding overheads on the system performance are included in the discussions.The main conclusion is that,the statistical CSI strategy is a generally better choice for DF one-way relaying and DF two-way relaying.