A simulation-informed robust optimization framework for the design of energy efficient underwater sensor networks

Given that data generation rates of sensors might deviate from what is anticipated during the configuration phase due to several reasons such as event-driven data spikes, dynamic environmental conditions, propagation delay and data buffering, etc., designing robust transmission schemes is pivotal for Underwater Wireless Sensor Networks (UWSNs). Despite advances in underwater technologies, UWSN optimization under traffic uncertainty remains underexplored. This paper presents a novel simulation-informed robust optimization framework for designing energy-efficient UWSNs. We begin with a comprehensive review of the literature that addresses uncertainty in system parameters for wireless network design, followed by an analysis of research focused on modeling the motion of underwater objects. Then, we propose simulating an intrusion detection environment that includes moving targets, such as autonomous underwater vehicles and submarines navigating along 3D routes. To improve simulation accuracy, real bathymetric data is used to define the interactions between system elements including vehicles, sensors, and ocean topography. Then, the expected data generation rates of sensors and the corresponding admissible intervals are determined using the results from multiple simulation runs. The resulting data set is used to determine and conduct comprehensive analyses on optimal deterministic and robust configurations, where the maximum battery allocated to a sensor is minimized. The scenario-based comparison of network functional time between deterministic and robust configurations indicates that the robust design substantially outperforms the deterministic configuration across all data rate realizations, even at the lowest level of deviation from the expectations.