| Summary: | 碩士 === 華梵大學 === 資訊管理學系碩士班 === 97 === On wireless networks hidden to security problem, various the development and research, The phenomenon wireless networks security problem is very important. In this study, Research on Neural-based risk management model with Spectrum analysis for wireless networks intrusion system is proposed. We use the neural network with risk management to complement the abnormal detection and the misuse detection.
This research integrated the physical characteristic of the AP signals, the data flow rate of the AP, and the user’s activities through the neural network model to determine the intrusion or not. At first, we collected the characteristic AP signals as the identification pattern and trained with the neural network to classify the output as the low, middle, or high risk value as output. Then, the data flow rate of the AP is also classified into the low, middle, or high risk value by neural network model. Based on the history records of security event, the user’s activity is also classified into the low, middle, or high risk value. The final simulation results: When we use only AP characteristic model, it has 85% ~ 90% accuracy rates. Added the data flow detection part, it can improve about 5%. The total accuracy can rise to 96% in the best mode. Base on the data of research, we know that Neural Network model with risk management for wireless networks, can be judgment the risk levels of wireless network, and the wireless network risk behaviors, the verification and the data of test result will be approved by finally chapters.
|
|---|
