Maritime vehicle tracking using underwater gliders and particle filters

A major difficulty in tracking unidentified vessels at sea is due to the need to observe large areas for long periods of time for sparsely appearing targets. A possible solution is to use a network of energy efficient underwater gliders that use buoyancy engines and are equipped with sensor array, processing, and communication hardware. Gliders, however, have a limited size and payload which limits the length of their sensor array and the capabilities of their power, processing, and communication modules. This reduces the resolution of measurements that can be collected and processed on board of gliders and be communicated to a fusion center for accurate target tracking. In this work, we propose the use of adaptive compressive sensing and processing to enable accurate target tracking while using low signal acquisition, processing, and communication rates, short sensor array lengths, and simplified sensing and processing hardware hosted on gliders. Moreover, we propose a particle filtering method to fuse data from multiple sensing nodes and process sensor array measurements with non-linear, non-Gaussian relationship to the target state. A simulation based scenario of tracking a single target by four underwater gliders is used to evaluate the effectiveness of the method.