Transmission-Cost Minimization for Packet-level Coding on Multi-path Wireless Networks.

Topological redundancies in modern wireless infrastructures provide opportunities to enhance reliability and latency performances through multiple data paths, however, existing standards for wireless access networks cannot fully utilize such an added degree of freedom. Packet-level coding, which proactively adds redundancy to the transmitted data at the packet level, can efficiently utilize the added bandwidth from all data paths and enhance reliability with low latency. Thus it is a good candidate to supplement single-link/physical layer techniques (e.g., channel coding) and provide a better chance of realizing new reliability- and delay-critical services, such as URLLC. In this paper we study the problem of how to optimally design the coding rate and packet distribution rule over multiple data paths to transmit a set of data packets under stringent reliability and delay constraints, so that radio resource cost is minimized. Since such problems are often NP-hard and obtaining optimal solutions usually requires very high-complexity exhaustive search, we propose novel heuristic algorithms that use the path resources in the order of their cost-effectiveness scores to approximately solve this problem. Numerical simulations show that our proposed algorithms can drastically reduce the computational complexity while still able to achieve near-optimal performance.