Intrusion Detection Method For WLAN With Home Automation Using Deep Learning Approach

The join of Home Automation makes Wireless Local Network(WLAN)in the home scenario more complicated,no longer stopping with providing Internet access service,and gradually evolving into WLAN with Home Automation(WLAN-HA).However,unlike enterprise WLAN,WLAN-HA is a kind of domestic product which is lack of professionals to carry out safety maintenance.Worse still,its potential risk from WLAN is not neglectable.In order to prevent family privacy leakage and reduce cybercrime,it is necessary to study the intrusion detection technology for WLAN-HA to supplement the original security mechanism.At present,there are still some shortcomings in the field of intrusion detection for WLAN-HA,which specifically go as follows: 1)Some work proposes models with low volume like Naive Bayes Classifier,which cannot fit complex network data to achieve enough accuracy.2)Some researches use simple detection features,which produces heavy overfitting and reduces accuracy in the real situation.3)Most studies verify their methods using simulation without actual measurement data.4)No existing research has proposed a solusion of intrusion detection for WLAN-HA.Therefore,this thesis selects appropriate features and models for empirical research on WLAN-HA,and the main work is as follows:1)By fully studying 802.11 standard and its implementation,the network structure of WLAN-HA and various related intrusion means,an empirical analysis is conducted to explore communication characteristics and network topology stability in WLAN-HA.With the concern of WLAN-HA traits and related intrusion means,a feature extraction method is designed to solve simple feature problem,and a related dataset for learning intrusion detection models is constructed via an experimental network.2)To better convey state information of WLAN,a traffic sequence model of WLAN is designed.Using a high-volume RNN(Recurrent Neural Network)approach,an RNN WLAN intrusion detection method based on traffic sequence classification is proposed.According to the experiment result,the proposed method surpasses the latest Stacked Auto-Encoder detection method by about 11.15% on detection performance.3)Using the above intrusion detection method,a prototype system of intrusion detection for WLAN-HA is designed and implemented under the reference of deployment form of Home Automation systems called as Cloud-Terminal-Console architecture.In summary,this thesis proposes a high-accuracy WLAN intrusion detection method and designs a prototype system to make up for existing theoretical researches,which is conducive to reducing sercurity risk of Home Automation,dispelling the concerns of using Home Automation systems and therefore promoting development and popularization of Home Automation and IoT.