Minimum-energy packet forwarding over lossy networks under deadline and reliability constraints

This paper studies minimum-energy packet forwarding over multi-hop lossy networks under deadline and reliability constraints. We assume a routing topology in the form of a directed graph with packet loss processes on links described by finite-state Markov chains, and formulate the forwarding problem as a finite-horizon constrained Markov decision process. We show that the minimum energy forwarding policy under hard deadline and reliability constraint can be computed using dynamic programming, and that the optimal forwarding policy is a randomized policy over two history-independent and deterministic policies. Closed-form optimal policies are derived for some particular scenarios. Numerical examples show that the transmission energy cost of achieving reliabilities close to the maximum can be significant when links are bursty. In addition, transmission power adjustments can further reduce energy cost. Finally, we develop simple heuristic policies with a good balance between transmission energy cost and reliability.