On an Alternative Approach to Congestion Detection in Ad Hoc Networks

Ad hoc wireless networks without pre-existing infrastructure depend on mutual collaboration among nodes. Congestion in such networks presents more of a challenge than in other types of networks. An essential task in this context is to determine when a node is congested. This paper considers the use of T-entropy to detect network congestion, and in particular how to define an appropriate entropy threshold for congestion detection. The paper first discusses how to recognise a congested node from the perspective of (one of) its neighbours. We then simulate ad hoc network scenarios, which we analyse for goodput and entropy to determine the relationship between entropy and the network congestion situation. As its main result, this paper proposes an entropy threshold algorithm for congestion detection in these scenarios for use in an AODV (Ad hoc On-Demand Distance Vector Routing) derivate, RAODV (Relieving AODV), that uses uncongested neighbours to detect and relieve congestion.