AMIND: An Asynchronous Middleware for Improving Neighborhood Discovery in Crowded Networks

Abstract

Neighbor discovery (ND) is crucial in deploying wireless sensor networks. Despite the considerable research on ND, most studies focus on networks composed of only two nodes. However, for cases in which more than two nodes are involved, the probability of collision increases, potentially preventing a node from discovering all its neighbors. This problem is especially aggravated for ND protocols that exhibit periodic behaviors that can lead to recurring collisions. This behavior is typical of deterministic ND protocols, in which its inherent operation forces regular on/off patterns and cyclical collisions. To address this problem, this paper introduces AMIND, a new middleware capable of disrupting the cyclic behavior of deterministic ND protocols. It achieves this goal by introducing random shifts in their operation, significantly increasing the number of nodes discovered compared to the existing ND protocols. As a result, AMIND allows a node to discover unknown neighbors after each shift. Simulation results show that the proposed middleware enhances node discoveries that previous ND protocols can achieve, even in scenarios with high collision rates.