On channel aware routing policies in shallow water acoustic networks

In this paper we consider shallow-water acoustic networks, and propose a routing policy that exploits qualitative information about the behavior of the channel, given some key parameters such as the position and depth of the source, the location of the receiver and the sea bottom profile. Our policy is based on a set of several synthetic channel realizations obtained using the Bellhop ray tracing software, where the channel variability is obtained via random perturbations of the sound speed profile and of the sea surface shape. The channel realizations are translated into Signal-to-Noise Ratio (SNR) statistics: the relay sought must comply with the constraint that the SNR exceeds a threshold with a given probability. We show that these SNR statistics allow the routing policies to identify geographic areas where a high SNR is more likely to occur. Our policy is compared to shortest-path routing (obtained via a centralized algorithm and oblivious of channel statistics), and to an optimal, genie-aided policy that always picks the best relay which complies to the SNR constraint. Results show that channel-aware policies consistently outperform the shortest path policy, and that our heuristic policy performs very close to the optimal one in several scenarios.