Analysis Of Traffic Capacity And Strategies For Improving Traffic Capacity Efficiency Based On Complex Network Theory

ABSTRACT:Since the pioneering discovery of small-world phenomena and scale-free feature in the end of20th century, the study of complex networks has been rapid development. It has been used to solve practical network problems. One of most important function of network is to transport objects running on it. The maximum carrying capacity of network under free-flow state can be measured by a parameter named as network traffic capacity. With the rapid development of the society, the rapid growth of network data, traffic congestion occurs on networks more frequently. Network traffic capacity shortage problem is becoming more and more obvious. Studies about how to reduce the network congestion and improve the traffic capacity are constantly emerging. After the research, researchers began to realize that traffic dynamics are highly relevant to the underlying network structure. In addition, traffic capacity is also significantly affected by routing strategies and network resource deployment and so on. Methods of how to improve traffic capacity are of three main types:designing efficient routing strategy, optimizing network topology and distributing finite network resources. The main contents of this thesis focus on improving network traffic capacity from these three aspects.(1) On the issue of optimizing routing strategy, this thesis presents a called improved routing strategy to enhance the traffic capacity of scale-free network. Under the shortest path routing strategy, the nodes of large betweenness cause congestion with high probability than the nodes with small betweenness. In our routing strategy, we consider the betweenness of every node as the factor of path cost directly by weighted the link, packets will choose the route which makes the path cost function minimum. So the heavy traffic load can be redistributed from nodes with large betweenness centrality to nodes with small betweenness centrality. It can improve the network traffic capacity and alleviate the congestion.(2) On the issue of optimizing network topology, this thesis reports the finding of a new weighted gradient network model on the unweighted substrate networks, where the edge weight depends on the degree of the node it connects and a tunable parameter. We study the different networks’congestion, such as scale-free networks and random networks. In addition, we also propose an efficient strategy to enhance the network transport efficiency by adding links to the existing networks. In our proposed strategy, we consider both the node betweenness and the shortest path length as two important factors. We separately evaluate the network capacity, network load, average shortest path length and robustness of different network models. Extensive simulation results show the effectiveness of the proposed strategy.(3) On the issue of resource allocation, in real communication systems, network resources are the link bandwidth, and node’s delivering capacity and node’s queue length and so on. In general, these network resources are finite, and their distribution is often not uniform. So reasonably allocating these network resources can significantly improve the performance of network system. In the network, each node has a finite queue length to store packets due to physical and economic constraints. Assuming the total queue length resource is limited, this thesis proposed a betweenness-based queue resource allocation strategy. The theoretical analysis and simulation results show that the queue resource allocation strategy can strongly alleviate the network congestion and enhance the network traffic capacity.(4) On the issue of optimizing two-layer network traffic capacity, in the previous studies of traffic dynamics on complex networks, the network model is assumed to be a single-layer structure. However, most real networks, such as peer-to-peer networks and wired-wireless networks have a two-layer structure, and these layers are interactive and mutually dependent. So for two-layer complex network research, we propose a heuristic global static routing algorithm to enhance traffic capacity of two-layer complex networks. Through the simulations found that the traffic capacity of our routing strategy is about10times more than the shortest path routing strategy. In our study, the network topology is composed of logical layer and physical layer subnetworks. Virtual links on the upper layer can be changed or constructed easily, and therefore the topological structure of the upper logical layer can be efficiently constructed by a link removal strategy. The structure of the two-layer network can be optimized freely based on this thesis proposed link removal algorithm. Extensive simulation confirmed that the traffic capacity of two-layer complex network can be significantly improved by removing a fraction of edges from the logical layer.This thesis carries out in-depth research and analysis on improving the network traffic capacity problems and the factors affecting the traffic capacity. Several strategies are proposed to enhance the traffic capacity of single-layer networks and two-layer networks with corresponding theoretical analysis and simulations. These strategies will provide feasible suggestions for network planning, improving the efficiency of network transmission and enhancing the performance of network.