Enhancing Intrusion Detection Using Binary Arithmetic Optimization with Sparse Auto Encoder for Fog-Assisted Wireless Sensor Networks

Abstract

Abstract Intrusion detection in Fog-Assisted Wireless Sensor Networks (Fog-WSNs) is a critical security measure designed to protect the functionality and integrity of these networks. Fog computing extends the abilities of traditional WSNs by introducing edge devices or fog nodes that can process data and provide computational resources. Intrusion detection in Fog-WSNs is vital to maintain the privacy and security of data generated and transmitted by sensors and processed at the fog node. Machine learning (ML) algorithms are widely employed to detect intrusion. These models can be trained on historical data to detect known attack patterns and can adapt to emerging threats. By implementing effective intrusion detection mechanisms, these networks can alleviate risks and ensure the reliable operation of critical applications in different fields namely IoT, smart cities, and industrial automation. This study designs an Intrusion Detection using Binary Arithmetic Optimization with Sparse Auto encoder (BAOA-SAE) technique for Fog Assisted WSN. The major aim of BAOA-SAE method is to recognize and classify the presence of the intrusions in the network. Primarily, the BAOA-SAE method focuses on the election of prominent features in the network data. Next, the SAE model is implemented to categorize the presence of intrusions. To improve the intrusion rate, the BAOA-SAE technique employs bacterial foraging optimization algorithm (BFOA). A broad range of simulated outcomes were performed on benchmark IDS dataset. The extensive outcomes highlighted the high efficiency of the BAOA-SAE method over other existing techniques.

Keywords

• Fog computing
• Wireless sensor networks
• Machine learning
• Intrusion detection
• Feature selection