Multi-Relay-Assisted Low-Latency High-Reliability Communications With Best Single Relay Selection

We study a multi-node Internet of Things system supporting low-latency high-reliability communication to a destination node. The rest of the nodes are potential relays in which the best single relay (BSR) is selected to assist the transmission to the destination. The system operates with finite blocklength (FBL) codes to satisfy the low-latency requirement. The scope of this work is to derive and improve the FBL performance of the considered BSR system. On the one hand, we extend Polyanskiy's FBL model of a single-hop scenario to the considered relaying system and derive the corresponding achievable reliability. On the other hand, by employing a practical FBL coding scheme, namely polar codes (PCs), an FBL performance bound attainable by a low-complexity coding scheme is presented. In particular, we provide a reliability bound of a dynamic-length PC scheme. Addressing a source-driven BSR strategy, as well as a relay-driven BSR strategy, we investigate two viable strategies for relay selection in the FBL regime, while the corresponding performance under an infinite blocklength (IBL) assumption serves as a reference. We prove that the two BSR strategies have the same performance in the IBL regime, while the relay-driven strategy is significantly more reliable than the source-driven one when considering the FBL regime. Furthermore, following the derived FBL performance model, we provide an optimal design to minimize the overall error probability via blocklength allocation. Through simulation and numerical investigations, we show the appropriateness of the proposed analytical model. Moreover, we evaluate both the achievable performance with FBLs and the performance of PCs in the considered scenarios while comparing the source-driven and relay-driven strategies.