On the Optimal ARQ Distribution for Low-Latency Communication over Line-of-Sight Dominated Multi-Hop Networks

Multi-hop networks that are dominated by line-of-sight (LOS) wireless channels have gained traction in the recent past owing to the emergence of wireless networks based on unmanned aerial vehicles. One of the challenges in such vehicular networks is to design communication strategies to provide both ultra-reliability and low-latency features. Towards providing ultra-reliability against channel impairments, it is well known that automatic repeat request (ARQ) based decode-and-forward relaying is an effective strategy wherein each transmitter can be allotted an appropriate number of re-transmissions based on the LOS component of its forward link. However, in order to provide low-latency features, it is also known that multiple re-transmissions may not be a favorable choice as the total number of re-transmissions across the relays incurs significant delay in communicating the packet in the end-to-end network. Identifying this conflict introduced by the ARQ protocol, we investigate the optimal allocation of the number of ARQs at each link so as to minimize the packet-drop-probability at the destination subject to a sum constraint on the total number of ARQs allotted to all the nodes in the network. First, we prove a set of necessary and sufficient conditions on the optimal ARQ distribution, and then use these conditions to propose a low-complexity algorithm to solve the problem statement. Through extensive simulation results, we show that the proposed algorithm significantly reduces the computational complexity when compared to exhaustive search and yet recovers the optimal ARQ distribution.