Underwater acoustic sensor networks: Medium access control, routing and reliable transfer

Date of Completion

January 2008


Computer Science|Physics, Acoustics




Recently there have been growing interests in monitoring aquatic environments for scientific exploration, commercial exploitation and coastline protection. The ideal vehicle for this type of extensive monitoring is a mobile underwater sensor network (M-UWSN), consisting of a large number of low cost underwater sensors that can move with water currents and dispersion. M-UWSNs are significantly different from terrestrial sensor networks: (1) Radio channels do not work well under water. They must be replaced by acoustic channels, which feature long propagation delays, low communication bandwidth and high channel error rates; (2) While most ground sensors are static, underwater sensor nodes may move with water currents (and other underwater activities), as introduces passive sensor mobility. Due to the very different environment properties and the unique characteristics of acoustic channels, the protocols developed for terrestrial sensor networks are not applicable to M-UWSNs, and new research at every level of the protocol suite is demanded. ^ In this dissertation work, we investigate three fundamental networking problems in M-UWSN design: medium access control, multi-hop routing and reliable data transfer. (1) Medium access control (MAC): the long propagation delays and narrow communication bandwidth of acoustic channels pose the major challenges to the energy-efficient MAC design in M-UWSNs. For the first time, we formally investigate the random access and RTS/CTS techniques in networks with long propagation delays and low communication bandwidth (as in M-UWSNs). Based on this study, we propose a novel reservation-based MAC approach, called R-MAC, for dense underwater sensor networks with unevenly distributed (spatially and temporally) traffic. Simulation results show that R-MAC is not only energy efficient but also supports fairness. (2) Multi-hop routing: In M-UWSNs, energy efficiency and mobility handling are the two major concerns for multi-hop routing, which have been rarely investigated simultaneously in the same network context. We design the first routing protocol, called Vector-Based Forwarding (VBF), for M-UWSNs. VBF is shown to be energy efficient, and at the same time can handle node mobility effectively. We improve the performance of VBF by proposing a hop-by-hop approach, called Hop-by-Hop Vector-Based Forwarding (HH-VBF). Our simulation results show that HH-VBF is more robust in sparse networks. Further, to solve the challenging routing void problem in M-UWSNs, we design a smart void avoidance algorithm, called Vector-Based Void Avoidance (VBVA). VBVA can effectively bypass different types of voids such as convex voids, concave voids and mobile voids. VBVA is the first algorithm to address 3-dimensional voids and mobile voids in mobile sensor networks. Our simulation results show that VBVA can achieve almost the same success delivery ratio as flooding, while it saves much more energy. (3) Reliable data transfer: In M-UWSNs, the long propagation delays, the low communication bandwidth and the high channel error rates all pose grand challenges to the reliable data transfer. We first tackle this problem by proposing a novel hop-by-hop erasure coding approach. Our results indicate its significant performance improvement upon the most advanced approaches explored in underwater acoustic networks. We also develop a mathematical model for the erasure coding scheme, providing useful guidelines to handle node mobility in the network. The second approach we explore is a network coding scheme, in which we carefully couple network coding with multi-path routing for efficient error recovery. We evaluate the performance of this scheme using simulations. And the results show that our network coding scheme is efficient; in both error recovery and energy consumption. ^ In this dissertation, we present these three strands of research work for M-UWSNs in detail, and at the end we point out some potential directions worth future investigation. ^