Research On Intrusion Detection Technique Based On Wavelet Neural Networks

As networks and computer technologies are widely applied in politics, economy and military, more and more networks security problems take great threats into networks application. In the face of increasingly serious problems of networks information security,it has been an urgent important problem to protect security of computer system,networks system and the whole information infrastructure.As a new dynamic security prevention technology ,it plays an important roles in protecting information system security and is also the second security line of defence behind firewalls.Wavelet Neural Networks(WNN) is a new branch of Artificial Neural Networks theory research,and is combined wavelet analysis with Artificical Neural Networks idea.The wavelet network imbibes advantages of both methods ,while has well function approximation capability and fault-tolerance capability at the same time. The innovation of this paper is to accomplish intrusion detection model based on wavelet networks;the convergent rate of neural networks is not only accelerated in this model , but also the problems of the conventional BP algorithm such as falling in to the local minimum point easily is solved. Intrusion detection methods based on the five different types of algorithm are conducted emulating experiments. The results of experiment show that compared with intrusion detection method based on conventional BP algorithm ,the first type of intrusion detection approach based on Morlet wavelet function has a detection rate increase of 4.96 percent, a false alarm rate decrease of 3.84 percent and a false negative rate decrease of 1.92 percent .As this algorithm is easily carried out , the intrusion detection technology based on the WN is cetain practically value in the environment of highspeed network in the future.