The Research Of Ad Hoc Network Traffic

Ad Hoc self-organizing networks, whose shortened form is Ad Hoc, is a temporary, multi-hop and self-governing distributed control network without basic network equipments and a center node, which is composed of mobile nodes with wireless network interface. Network topology change optionally and nodes in Ad Hoc also can move optionally.None center nodes and low computing ability of nodes make traditional mechanism of encryption and authentication not implement. The changes of topology, node number, trust relations and data engendered and transmitted in network, let network safe scheme, which is configured statically, including database and file system, not be applicable again. Traditional firewall can't dispose because of blurry network boundary. Therefore, Ad Hoc security has become a hot research point.Network Intrusion detection, including misapplication detection and anomaly detection, can recognize the attacks which attempt to happen, are happening or have happened, and it is a kind of active network security protection measure. The most existing intrusion detection technologies have the deficiency of high false positive rate and negative rate, and poor real-time performance. Especially, high detection accuracy is usually based on abundant or self-contained training data.In this dissertation, Ad Hoc network key technologies and security requirements are firstly expounded. Then, PSO algorithm and QPSO algorithm are systematically introduced.Ch.4 introduced the application of K-Means clustering algorithm and QPSO algorithm in Ad Hoc anomaly detection. Particle swarm is firstly initialized by the result of K-Means clustering algorithm, and clustering process is based on Euclidean distance among data vectors. Then clustering center is found by QPSO algorithm. Simulation experiment is on NS-2 with use of KDD 99 dataset, and experiment result shows the method is effective.In Ch.5, wavelet neural network, trained by QPSO algorithm, is used in Ad Hoc anomaly detection. To test performance, gradient descent algorithm, PSO algorithm and QPSO algorithm are used to train the same wavelet neural network's parameters for Ad Hoc anomaly detection. Simulation experiment result on NS-2 with use of KDD 99 dataset shows the method is effective.In Ch.6, based on Ch.5, B-QPSO algorithm is introduced. To test performance, it's compared with gradient descent algorithm, PSO algorithm and QPSO algorithm. Simulation experiment result on NS-2 with use of KDD 99 dataset shows the method is effective.The research is indicated that QPSO algorithm, combined with other intelligence algorithms, can be used in Ad Hoc anomaly detection. Simulation experiment shows that the accuracy of anomaly detection is enhanced and the false positive rate for normal state in the network anomaly detection was declined.