A Congestion Avoidance Routing Protocol for Cognitive Scale-Free Networks

Network performance is strongly dependent on network topology, especially in scale-free network in which nodal-degree distribution is a power-law distribution. In this paper, a routing protocol for cognitive scale-free networks, called as CSRP (Cognitive Scale-free Routing Protocol), was proposed. In CSRP, each node predicts the numbers of queuing packets in its neighbor nodes when routing decisions are made, so that the optimal path is established with the tradeoff between the shortest path length and load bearing. Based on the comprehensive understanding of network traffic distribution and intelligent routing decision, CSRP can reduce network congestion dramatically. Further, we analyze the critical threshold of arrival rate where the network status changes from free to congestion. This value can reflect the maximum capacity of a system handling its traffic. Under the same network scenario, CSRP has the best critical value. Compared with other existing routing protocols, CSRP is more successful in keeping delay low and more traffic flow can be allowed into the network.