Research On Algorithm For Unstructured Peer-to-Peer Overlay Network Topology Automatic-restoration

Distributed, self-organizing P2 P network as a carrier has been widely used in the network. However, P2 P overlay network topology has become one of the key factors which affect the application of the quality of service of Internet. With the expanding network of information, covering an increasingly complex network topology, how to effectively optimize the network topology is the key P2 P technology research.As the scale of P2 P networks growing larger and larger, the strong dynamic characteristic of nodes have challenged the stability of overlay topology networks. Apart from dynamically logging in or logging out of the system, the nodes in P2 P network may show some invalid behaviors caused by various abnormal conditions, which can make the network topology structure abnormal and affect the efficiency and the robust degree of the network seriously. This will destroy the stability of overlay topology networks and affect P2 P network system robustness and availability. So many researchers put a lot of effort on how to detect P2 P overlay network topology abnormal situations correctly and apace. They have opposed many P2 P overlay network topology detection algorithms according their research. But formal algorithms have many deficiency facing with largescale P2 P networks.This article studies the technology of topology anomaly detection and quick automatic-restoration in depth under unstructured P2 P environment, its object is to improve the survivability of the network topology and guarantee the transmission efficiency. Because traditional detection algorithms have such defects as large communication overhead, low transmission utilization-rate and etc., we propose a algorithm based on gossip protocol for network topology anomaly detection under the environment of unstructured P2 P overlay network. This algorithm can enable each node in P2 P overlay network to detect the abnormal connections in the adjacency list autonomously only depending on local information. Through theoretical analysis and simulation experiments, it is proved that this algorithm has the advantages of small communication overhead, high adaptability, and etc. Under the environment of dynamic network transmission, this algorithm can reduce the communication overhead further by its passive detecting function.We also propose a quick automatic-restoration algorithm for network topology anomaly. This algorithm is applicable to any distributed network universally, including unstructured P2 P overlay network. This algorithm can enable the node to maintain the backup connecting-points which come from each neighbor by means of designing and maintaining the backup connection mechanism, and make a response as soon as the adjacency-list anomaly is found; moreover, since this mechanism has sustainable repairing capability, it will guarantee the network connectivity even when large amount of topology anomalies occur many times continuously in the network, thus it can improve the reliability and the survivability of the network. While receiving the date packet, the node can perform the path compression optimization simultaneously so as to improve the network transmission efficiency. Through theoretical analysis and simulation experiments, it is proved that this algorithm has the advantages of small communication overhead and strong network survivability, and it can improve the network transmission efficiency substantially.