Underwater networks are envisioned to enable several applications for oceanographic data collection, environmental monitoring, navigation and tactical surveillance. Many applications make use of Autonomous Underwater Vehicles (AUVs) equipped with underwater sensors. Underwater communication links are based on acoustic wireless technology, which poses challenges due to the unique underwater environment such as high propagation delays, high bit error rates, and temporary losses of connectivity caused by multipath and fading phenomena. For data collection and monitoring tasks, underwater vehicles can either periodically send the measured data to the surface station (sink) or the sink can initiate a query to the sensors asking for the information of interest. The former case is reduced to unicasting, where the data is sent periodically by nodes to a specific destination, i.e., the surface station. In the later case, query dissemination can involve either broadcasting or geocasting technique, depending on whether the query is sent to all the nodes, or a subset of nodes based on location respectively. As broadcasting can be viewed as a special case of geocasting, geocast protocols provide a general routing scheme for query dissemination. In either of the cases, reliability is a crucial factor for underwater communications.

Reliability, especially in a mobile environment, is a major concern due to network dynamics. Due to the high propagation delays involved in underwater communications, we do not consider transport solutions for reliable communications. Rather, we consider the lower layers for ensuring reliability. In this work, three versions of unicasting and geocasting protocols have been proposed, which integrate Medium Access Control (MAC) and routing functionalities and leverage different levels of neighbor knowledge for making optimum routing decisions. Performance evaluation has been done for unicast protocols in terms of different end-to-end metrics, for static and mobile scenarios, with an aim of finding an optimal level of neighbor knowledge required in either of these scenarios. It is observed that based on different end-to-end metrics considered, one version of unicast protocol outperforms the other. Thus, based on the application requirements and scenarios considered, an optimum level of neighbor knowledge can be utilized for periodic data delivery from nodes to the surface station.