Throughput of Networks With Large Propagation Delays

Abstract

Propagation delays in underwater acoustic networks can be large as compared to the packet size. Conventional medium-access control (MAC) protocol design for such networks focuses on mitigation of the impact of propagation delay. Most proposed protocols to date achieve, at best, a throughput similar to that of the zero propagation delay scenario. In this paper, we systematically explore the possibility that propagation delays can be exploited to make throughput far exceed that of networks without propagation delay. Under the assumptions of the protocol model in a single collision domain for a half-duplex unicast network, we show that the upper bound of throughput in an <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</i> -node wireless network with propagation delay is <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</i> /2. We illustrate network geometries where this bound can be achieved and study transmission schedules that help achieve it. We show that for any network, the optimal schedule is periodic and present a computationally efficient algorithm to find good schedules. Finally, we show that <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</i> -node network geometries that achieve throughput close to the <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</i> /2 bound exist for any <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</i> and present a lower bound on achievable maximum throughput for bounded geometries. This paper chiefly endeavors to explore the impact and potential of nonzero propagation delays on network throughput. We believe that the novel observations in this paper could motivate further research into this area, especially random access networks with large propagation delay, with a fundamentally changed outlook on maximum achievable throughput. This could lead to novel scheduling and network configuration approaches with applications in underwater and satellite networks.