The Study Of Traffic Dynamical Behaviors For Complex Networks

The TCP/IP-based communication networks have developed enormously over the past two decades. Both the size of networks and the number of users have increased explosively. With the emergence of more and more in-depth network applications, the humankind has involved into a network-based information era. Meanwhile, congestion is a basic network phenomenon that cannot be completely avoided. Severe congestion can cause the deterioration of some performance indices, including the time delay, the transmission speed and the network throughput. How to improve the transport capacity of these networks and alleviate the traffic congestion has been an important topic in complex dynamical networks. In general, the capacity is influenced by three factors: routing algorithm, the node packet-delivering capacity and the topology of a network. Furthermore, when large number of information packets rush into the network, congestion will also occur due to the finite memory.The basic routing protocol in communication networks aims at the fastest transmission, which is based on the estimate of the shortest path from a given source node to the destination. Although the Internet and other communication networks work in a reliable way under the basic protocol, it will be less efficient when network enters into the severe congestion phase. Therefore, we firstly introduce a general strategy that incorporated local traffic information into the basic shortest path routing policy which is controlled by a tunable parameter. By simulation, combining with the knowledge of the topology and the local traffic information of the network, the new strategy behaves more efficiently compared with the basic routing protocol. But the new routing policy ignores the information between the neighbors and the destination. In the following, by estimating the information from neighbors to destination, we propose an improved routing strategy with a tunable parameter. We investigate the effectiveness of the proposed routing strategy for BA scale-free network with different tunable parameter. Compared to the new strategy incorporated local traffic information, the simulation shows that the improved routing strategy not only has higher transport capacity, but also behaves more efficiently in the tolerance against random attacks of links.Secondly, we propose a limited packet-delivering capacity model for traffic dynamics in BA scale-free networks. In this model, the total node's packet-delivering capacity is fixed, and the allocation of packet-delivering capacity on node i with degree k_i is proportional to k_i~β, whereβis a adjustable. We have applied this model on the shortest path routing strategy and found that there exists an optimal value of parameterβleading to the maximal network capacity.A summary of the full text and the directions for further studies are given at the end of this thesis.