We propose development of a network layer protocol necessary to increase the utility of acoustic communications in the shallow water environment. Increasing attention has been given to collecting data from difficult to access coastal waters for diverse activities, to include scientific research, industrial and commercial concerns, and military applications. The preponderance of activity has focused on developing reliable methods for transmitting the information collected through the difficult time-varying shallow water medium. However, current network layer protocols, which are responsible for determining traffic routing, do not provide for guaranteed quality of services. Current protocols also may cause unnecessary message delays. The delivery of traffic in an acoustic network is complicated by the excessive propagation delays resulting from the speed of sound in water. The effect of this propagation speed is to cause an acoustic network connected by short hops to perform similar to a wire-based network with links between neighboring nodes of over 100,000 kilometers. However, the potential benefits and increased potential for application developments that stem from the implementation of wireless underwater networks make it worthwhile to explore means of mitigating the effects of the propagation delays. Central to the problem of route determination for network traffic is the discovery of the networks topology from which network nodes extract next-hop information upon which to base traffic forwarding decisions. Two principal methods are used for discovering the route information. Proactive routing methods pre-compute route data before network traffic is generated, thus when traffic is submitted by network applications the appropriate routes are already known. Reactive routing determines the route information in response to traffic submissions. This method, referred to as on-demand routing, seeks to minimize route discovery traffic by only determining routes necessary to support actual traffic patterns. This routing information is cached to increase responsiveness to submitted traffic and adaptability to topology changes. Both methods have their merits; however, neither adequately supports resource allocations necessary to assure guaranteed levels of service quality. We proposed a novel network protocol that provides many of the benefits of proactive routing yet retains the adaptability of reactive protocols. Our protocol is based upon a central master node which periodically probes the network for active participant nodes. In responding to the probes, the nodes provide the master node with sufficient information for the master node to determine all possible data paths through the network. From this information the master node makes all routing decisions for the network and provides irnext-hoply information to each non-master node, thus reducing the workload on non-master nodes. This information also enables the master node to optimize the allocation of traffic to network paths providing for active management of delays insuring delay variance and data capacity are within the established quality of service commitments. Fundamental to this approach is the separation of control traffic from data delivery. This separation allows data to be transmitted without first having to wait for traditional handshake mechanisms to provide access to the channel. This separation significantly reduces the expected delay to which traffic is subjected resulting in potentially higher data throughput. Progress in network layer protocols will expand the usefulness of underwater acoustic networks beyond applications that are limited to very low data rate traffic and non-time sensitive data types. Specific beneficiaries of this research activity include the Deployable Autonomous Distributed System, the SeaWeb Technology Demonstrations, the Autonomous Oceanographic Sampling Network, and the National Oceanographic partnership Programs Front Resolving Observational Network with Telemetry (FRONT) project.
