Distributed underwater systems, consisting of multiple sensing platforms can provide critical data to better understand complex and inter-related ocean processes. However, to correctly interpret data collected by such systems, we need to know when and where samples were obtained. Since GPS is not available underwater, the position of devices should be estimated with respect to certain known references that may reside on the surface or on the sea-bed. The main challenge is that devices are energy-constrained. Further, low-cost solutions to both the sensing platforms and the overall system design would be critical in making these systems more prevalent and available to scientists. Existing underwater tracking techniques are not well-suited to distributed systems because they are built around stand-alone platforms. They ignore vital relative information between devices and require long range communication which is both expensive and has high energy consumption. As a result, existing techniques do not scale well in multi-vehicle systems. To address these challenges, we take a systems perspective to underwater positioning. This means that instead of viewing each node as a separate entity, as in traditional systems, we consider the network as a whole. Therefore, we shift our focus to tracking a collective, rather than independently positioning a number of devices. This paradigm-shift allows us to use collaboration between devices to improve the performance and energy-scalability of mobile distributed systems. However, it makes the problem of tracking and localization more complex. Our proposed solutions draw on the framework of factor graphs to optimally and jointly estimate the trajectories of multiple nodes by combining information in 4 dimensions. This framework allows us to leverage both from network density and accurate motion information, if available. In addition, we have identified node mobility as a key factor that can both impede and improve performance. We show how mobility in combination with delays in medium access is an impairment to time-synchronization and localization and propose cross-layer and collaborative approaches to counter its effect. Our proposed strategies are essentially aimed at more efficient localization and tracking in underwater networks where resources are constrained. We believe that our techniques in combination with existing systems would address the localization requirements of a wide range of underwater applications
