Assessing the influence of selfishness on the system performance of gossip based vehicular networks

Abstract

In delay tolerant vehicular networks, gossip is an efficient forwarding scheme, which significantly reduces the message transmission overhead while maintaining a relatively high transmission rate in the high mobility vehicular environment. This mechanism requires vehicles as the network nodes to forward messages according to the system-defined gossip probability in a cooperative and selfless way among all the vehicles in the system. However, in the real word vehicular networks, most of the vehicular nodes exhibit selfish and non-collaboration behaviors to reduce the gossip probability in order to save their own energy and other limited resources in the vehicular nodes. In this paper, we study how node selfishness influences the performance of energy-constrained gossip forwarding based vehicular networks. We consider two typical forms of selfishness in the realistic vehicular networks: individual selfishness and social selfishness, and study the networking performance by focusing on the average message transmission delay and mean transmission cost. First, we model the message transmission process with selfish behaviors in the gossip forwarding based delay tolerant vehicular networks using a continuous time Markov chain. Based on this useful model, we derive closed-form formulae for average message transmission delay and mean transmission cost. Then, we give extensive numerical results to analyze the impact of selfishness on system performance of the vehicular networks. The results show that gossip forwarding in delay tolerant vehicular networks is robust to selfish behaviors since even when they increase the message transmission delay, there is a gain on the message transmission cost.