Attack Detection and enhanced Data Security Using Quantile Regressive Extreme Learning Machine and Contextual Cryptosystem in Wireless Networks

Abstract

 Wireless networks enable devices to communicate and share data without the need for physical connections, such as cables or wires. The increasing usage of computer networks raises additional cybersecurity concerns, necessitating the implementation of preventive measures to protect valuable data. Intrusion Detection Systems (IDS) are essential components of network security, designed to monitor computer networks and systems for suspicious activities, unauthorized access, or potential attacks. The key functions of IDS include collecting, analyzing, and identifying abnormal behavior within the system, as well as responding to potential threats. Secure data transmission in wireless networks is a vital aspect for cryptographic techniques-based intrusion detection systems (IDS). To enhance attack detection accuracy, a novel method called the Quantile Regressive Extreme Learning Machine based Contextual Naccache–Stern   (QRELM-CNSC) has been developed for wireless networks. QRELM-CNSC includes two major processes such as classification and secure data transmission within the wireless network. First, the Quantile Regressive Sequential Extreme Learning Machine classifier is employed for efficient attack detection in wireless networks and achieving higher accuracy. In the Extreme Learning Machine classifier, a number of data samples and their features are considered as input at the input layer. In Hidden Layer 1, Camargo's Index Targeted Projection Pursuit model is applied to select significant features from the dataset. With the selected features, Quantile Regression is applied in Hidden Layer 2 to analyze the data samples. Finally, the data samples are classified as normal or attack nodes (i.e., Fuzzers, Analysis nodes, Backdoors, DoS nodes, and Exploit nodes) in the output layer. Subsequently, sensitive normal data samples are transmitted securely using the Pseudo Randomized Contextual Naccache–Stern Cryptosystem. The proposed cryptosystem consists of three processes namely contextual key generation, encryption, and decryption. In the key generation process, both contextual public and private keys are generated. After key generation, the sender encrypts the data using the receiver's public key and transmits it to the receiver. The authorized receiver then decrypts the ciphertext to obtain the original data. This process ensures secure data transmission, enhancing data confidentiality in wireless networks. Experimental evaluations are conducted on various factors, such as attack detection accuracy, precision, recall, F-measure, data confidentiality rate, and attack detection time, concerning different numbers of data samples. The performance analysis results indicate that the proposed QRELM-CNSC method achieves better attack detection accuracy, precision, recall, and data confidentiality while minimizing time consumption