Two-stage stochastic programming for resource allocation in mobile underwater acoustic networks

Abstract

Underwater Acoustic Networks (UANs) enable long-range communication for a range of applications. However, UANs are severely limited by scarce acoustic spectrum, which is shared with marine animals and can lead to disruptive interference, contributing to anthropogenic ocean noise. In addition, the underwater environment is highly dynamic, where: marine-animal presence is uncertain, channel conditions fluctuate, and Doppler effects degrade communication reliability. Prior UAN studies have addressed these challenges separately, despite their combined effect on resource allocation. Accordingly, this paper proposes a two-stage Scenario-based Stochastic Programming (SSP) framework with a hybrid risk approach that jointly treats these challenges within a single resource allocation model for uncertainty-aware decision-making. In this model, channel allocation is decided in the first stage, while adaptive power allocation is performed in the second stage after uncertainty is realised. Compared with its deterministic expected-value counterpart under identical scenarios, this framework improves objective value by 6% to 19%. Further, it raises mean network capacity by an indicative 14% over an existing deterministic study at average animal presence. These gains improve robustness in worst-case and mixed-application conditions, at the cost of a modest increase in transmit power. These results confirm that stochastic optimisation provides a vital foundation for marine-aware UAN design.