Detection of stable mobile communities in vehicular Ad Hoc Networks

A vehicular Ad Hoc Network (VANET) is a type of wireless network where nodes represent moving vehicles. Some VANET-based applications might require organising vehicles into groups (communities). However, the high mobility of the vehicles and their tendency to travel in clustered flows makes them a challenging subject for community detection. Communities should be formed on the fly, using only local knowledge about their neighbourhood, and evolve as vehicles move over time. The main goal of this paper is to evaluate if stable community detection is possible in highly dynamic VANETs. To do so, the SandSHARC algorithm is extended with two different stability metrics: link duration and mobility similarity. In order to evaluate the stability of the evolving mobile communities, new quantitative and qualitative metrics are proposed, and run simulations on two different VANET scenarios: highway and urban. We analyse how the two stability metrics improve the stability of the detected communities. The results demonstrate that stability depends on the mobility patterns of the underlying traffic network. The proposed method can be applied in a distributed manner and produces communities with robust communication links between members what makes it viable for VANET applications.