Particle collision model based explanation of self-similar emergence in local wireless networks

Self-similar traffic and transmission have been discovered in Ethernet LAN and then spread to WAN and other type of networks by measurements of real traffic. Recent studies show the wireless ad-hoc network can also exhibit self-similar behaviour to some degree differing from the assumption of Poisson distribution in the past. However, the reason of its emergence is still unclear. It cannot be explained wonderfully by the achievements in wired networks that the superposition of large number of heavy-tail distributions can lead to self- similar traffic distribution. This paper introduces Particle Collision Model (PCM) to explain the spontaneous self- similar trends in wireless ad-hoc networks. The simulations show that the interaction and packets forwarding among nodes can lead to self-similar trends in traffic flow though it was obscured in real networks. This can explain the emergence of Self-similarity in MAC layer and wireless networks, and also the self-similarity in small networks.