A Model-Based Monitoring Scheme for Disruption-tolerant Underwater Sensor Networks

Underwater sensor networks have the potential to enhance our ability of observing and monitoring underwater physical phenomena, warning against natural disasters, and protecting critical areas, such as ports, from malicious attacks. However, the adverse underwater conditions and the limits of underwater acoustic communication, such as volatile link quality and temporal node isolation, remarkably affect the quality of real-time monitoring. In this paper we analyze the problem of designing a disruption/delay tolerant data collection scheme for real-time underwater sensor monitoring. More precisely, we point out the pros and cons of current data collection approaches, and propose a model-based monitoring scheme that provides probabilistic user guarantees in the presence of temporal communication disruptions. We outline the benefits of our model-based approach, such as resource-efficiency, high adaptability and anomaly detection, and discuss some future work.