A Model-Based System for Intrusion Detection Using Novel Technique-Hidden Markov Bayesian in Wireless Sensor Network

Intrusion detection is one of the major challenges faced in wireless communications. The wireless network, where tiny sensors are comprised with deployment in remote areas and in few critical applications, is called wireless sensor network (WSN). Since WSN is computationally intensive, the framework is designed using the novel technique of machine learning known as hidden Markov Bayesian model. The model is also termed as naïve Bayesian hidden (NBH) as a decision function and knowledge-based Bayesian hidden system (KBHS) for detecting an intruder. Our approach mainly aims to detect the intruder based on the trained data using a novel algorithm applied for intrusion detection system framework. In phase 1, the dataset is preprocessed and trained. Then, the trained data is forwarded for detection where the decision function is applied by using hidden Markov Bayesian approach. The detection is performed in this stage and then updated for testing the detected data in the second phase. The simulation results are tabulated, and experiments obtained determine better performance with high detection rate and accuracy. The results also prove high throughput with minimum delay in transmission. Also, the obtained results are compared with that of weighted support vector machine (WSVM). The comparison shows better performance rate of WSN with high security.