Medium Access Control in Distributed Networks with Large Propagation Delay

Abstract

Most of the Earth is covered by water, so underwater acoustic networks (UWANs) are becoming increasingly popular in a variety of undersea applications. The needs to understand the underwater environment and exploit rich undersea resources have motivated a further development of UWANs. Underwater acoustic signals suffer from more difficult physical channel phenomena than terrestrial radio signals due to the harsh underwater environment, such as sound absorption, time-varying multipath spread, man-made and ambient noise, temperature and pressure dependent refraction, scattering and Doppler shift. Among all the challenges, the large ratio of propagation delay to packet duration (relative propagation delay (a)) is arguably the most difficult one to address in the Medium Access Control (MAC) layer. In this dissertation we focus on the examination and improvement of the MAC layer function in UWANs, based on a critical examination of existing techniques. Many MAC techniques have been proposed in recent years, however most of them assume the ratio of the propagation delay to the packet duration is negligibly small (a<<1). When the relative propagation delay a increases (a>>1), these protocols perform poorly. This is because the large a leads to both a large negotiation delay in handshaking based protocols and the space-time uncertainty problem as the packets do not arrive at each node contemporarily. Some underwater-oriented protocols have been proposed which attempt to address these issues but the more successful rely on master nodes or a common understanding of geometry or time. We show by analysis and simulation that it is possible to eliminate collisions in ad-hoc networks with large relative propagation delay (a>>1) as well as improving the channel utilisation, without a common understanding of geometry or time. This technique is generally applicable, even for truly ad-hoc homogeneous peer-to-peer networks with no reliance on master nodes or other heterogeneous features. The mechanism is based on future scheduling with the inclusion of overhearing of RTS messages and allowing third-party objections to proposed transmissions. This MAC mechanism is immediately applicable in underwater acoustic networks (UWANs), and may find other uses, such as in space or very high rate terrestrial wireless networks. In summary, the key contributions of this study are: investigating the causes of the poor performance of existing MAC protocols in ad-hoc UWANs with large relative propagation delay, fully detailing the problem in order to propose analytic solutions, demonstrating how the MAC layer of an ad-hoc UWAN can eliminate packet collisions as well as improve channel utilisation without time synchronization or a network’s self-configuring phase to gain knowledge of the geometry, and verifying the utility of the proposals via both theoretical analysis and simulations.