Analyzing the Performance of Multi-hop Underwater Acoustic Sensor Networks

John H. Gibson, Geoffrey G. Xie, Yang Xiao, and Hui Chen

Abstract — Multi-hop underwater acoustic sensor networks constrain the performance of medium access control protocols. The efficiency of the well-known RTS-CTS scheme is degraded due to long propagation delays of such networks. Recently, interest in Aloha variants has surfaced; however, the performance of such protocols within the context of multi-hop networks is not well studied. In this paper, we identify the challenges of modeling contention-based medium access control protocols and present a model for analyzing Aloha variants for a simple string topology as a first step toward analyzing the performance of contention-based proposals in multi-hop underwater acoustic sensor networks. An application of the model suggests that Aloha variants are very sensitive to traffic loads and network size.

Index Terms—Aloha Protocol, Medium Access Control, Underwater Acoustic Network

Introduction

Underwater acoustic sensor networks (UASNs) are constrained by both link capacity and propagation delays. For such networks, traffic generally flows from the individual sensor nodes to a single gateway node that serves to interface the acoustic network with the external world. Each node is responsible for sending its own traffic to the gateway as well as forwarding all traffic from upstream nodes to the gateway. A medium access control (MAC) protocol for such networks must be tailored for the particular traffic pattern as well as the pertinent capacity constraints and propagation delays. In [1], it was found that the traditional RTS-CTS mechanism is inefficient in networks composed of more than just a few hops. Contention-based protocols that implement carrier sense mechanisms are also less effective for networks with extreme propagation delays unless large frames are used [1].

Contention-based protocols based on the simple Aloha protocol may be effective for such networks [2]; however, their performance in multi-hop environments subject to the specific traffic characteristics of an UASN is not as thoroughly understood. Before implementing such protocols in an operational network a more rigorous analysis of their performance expectation should be performed.

Theoretical analyses performed regarding MAC methods for underwater acoustic networks have so far focused on single hop topologies.

Fig. 1: String Topology

In particular, an analysis of the performance of an Aloha variant was reported in [3]. The topology studied consisted of a single receiving node surrounded by multiple contending sources. No consideration was given to the impact of having to relay traffic across more than one hop. The unique characteristic of UASNs, as noted, is that traffic in these networks tends to have a particular flow pattern. As all sensor- generated traffic flows to the gateway (GW), the offered load within a particular single neighborhood is inversely related to the number of hops that neighborhood is from the gateway, increasing the vulnerability of traffic to congestion as the traffic approaches the gateway.

Other factors beyond the traffic characteristics of the network complicate the analysis of the performance of MAC protocols within the context of a multi-hop topology. These include the half-duplex nature of the communication, time- varying and space-varying signal propagation losses, which make it a challenge to model the transmission error rates and link connectivity, application constraints, such as reliable service, and complex topology implementations. While a comprehensive theory to address all of these factors is desirable, such complexity is beyond the scope of this paper. Rather, this paper establishes an initial step towards such a theory by defining the problem space and developing a model of the performance of Aloha within the context of a simple, multi-hop topology as depicted in Fig. I. The model can be extended to more complex topologies, such as a tree composed of multiple string topologies.

The model provides a method for computing the expected network utilization and the probability of frame delivery to the gateway from an arbitrary sensor. The results offer insights useful in determining the appropriateness of an Aloha variant for such topologies. An application of the model indicates that Aloha variants may have applicability for simple UASNs with small loads.