Fuzzy Similarity Fusion based Feature Selection and Optimization Enabled Deep Learning for Intrusion Detection in WSN

An essential component of data protection is intrusion detection. With the widespread use of wireless sensor networks in military and environmental monitoring, security concerns have grown. The lack of physical defence devices makes data transferred via wireless sensor networks susceptible to attackers. To protect against such threats, appropriate intrusion detection techniques are essential. In this paper, we will construct and explore the suggested fuzzy similarity fusion approach, as well as student psychology Earthworm Optimization Algorithm (SPEWA)-based Deep Q network for intrusion detection. The WSN network will be simulated first, followed by data packet transmission in WSN. The intrusion detection process will then be carried out at the base station. In the base station, the input data will be transmitted to the feature selection module, which will use a newly invented fuzzy similarity fusion approach that combines wrapper, Canberra distance, and fuzzy logic. Once the relevant features have been chosen, intrusion detection will be performed by the Deep Q network, which will be trained using the Student Psychology Earthworm Optimization Algorithm (SPEWA).The SPEWA will be redesigned to incorporate the Student Psychology Based Optimization Algorithm (SPBO) and the Earthworm Optimization Algorithm (EWA). The experiment will be conducted out using the Python programmed BoT-IoT and the MQTT-IoT-IDS2020: MQTT Internet of Things Intrusion Detection Dataset. Furthermore, the performance analysis of the created fuzzy similarity fusion approach will be assessed using four metrics: energy, accuracy, recall, and f-measure.