Threshold Policy for Route Discovery Initiation in Mobile Ad hoc Networks

TL;DR

This paper proposes a threshold-based policy for initiating route discovery in mobile ad hoc networks, optimizing throughput by balancing discovery costs and network dynamics, with a method to set the threshold based on system parameters.

Contribution

It introduces a systematic rule for setting the threshold in route discovery, extending previous heuristic approaches to more dynamic and realistic network models.

Findings

Threshold policy effectively balances route discovery frequency and throughput.

Results are applicable to networks with node mobility and dynamic relay participation.

The approach generalizes to different mobility models.

Abstract

Achieving optimal transmission throughput in data networks in a multi-hop wireless networks is fundamental but hard problem. The situation is aggravated when nodes are mobile. Further, multi-rate system make the analysis of throughput more complicated. In mobile scenario, link may break or be created as nodes are moving within communication range. `Route Discovery' which is to find the optimal route and transmission schedule is an important issue. Route discovery entails some cost; so one would not like to initiate discovery too often. On the other hand, not discovering reasonably often entails the risk of being stuck with a suboptimal route and/or schedule, which hurts end-to-end throughput. The implementation of the routing decision problem in one dimensional mobile ad hoc network as Markov decision process problem is already is discussed in the paper [1]. A heuristic based on threshold policy is discussed in the same paper without giving a way to find the threshold. In this paper, we suggested a rule for setting the threshold, given the parameters of the system. We also point out that our results remain valid in a slightly different mobility model; this model is a first step towards an `open' network in which existing relay nodes can leave and/or new relay nodes can join the network.