Optimizing CNN Model for Anomaly Detection in IoT through Software-Defined Wireless Sensor Networks (SD-WSN)

Optimizing CNN Model for Anomaly Detection in IoT through Software-Defined Wireless Sensor Networks (SD-WSN)

연구 분야: Networking

학회: 2024 IEEE Globecom Workshops (GC Wkshps)

Software define networks (SDN) are used to provide centralised control to the network, and their integration to wireless sensor network (WSN) leads to the development of Software-Defined Wireless Sensor Networks (SD-WSN). Due to SD-WSN the management of the sensor devices improves, but they also become vulnerable to many cyber attacks. In this context many papers are published that are focused on the anomaly detection in SD-WSN. However, these papers are not includes the optimal optimizer selection issue in the detection process. Therefore, in this paper we first compare nine available optimizes (Adam, SGD, Adagrad, AdamW, NAdam. Adamax, ASGD, NAdam, and RAdam) and then use the most efficient optimizer for the anomaly detection.