AIRP: A system-based selection of adaptive intrusion recognition and prevention

Abstract

Machine learning techniques became a common feature of intrusion detection systems, enabling real-time responses and learning and adaptation. A well-trained model can be developed to enhance intrusion detection efficiency by using a detailed data set of many attack styles. However, for machine learning methods, high dimensional data poses an essential challenge. The processing of related features offering redundancies improves device time, which is a crucial issue for consumers with minimal resources such as battery, electricity and more. In this article we propose the Trust Malware Prevention and Precision System and the trust-based Secure Network intruder classification and prediction system. Based on a new function discovery algorithm, this reduces the number of functions in the data entry. The functions were originally randomly clustered to maximise the possibility that they would be included in the production of separate collections and organised depending on their exactness. Only the high classification functions are then chosen for any packet obtained in a network that is saved as part of the past output of the node. AIRP recommends regular machine cleaning, which evaluates and regularly renews the confidence relationships of participant nodes. AIPR gives the exact time for nodes cleaning states and limits the exposure window of nodes by offering a complex algorithm. For the final rating judgement for both models, the past behaviour of the node with the learning algorithm is determined. Any attack observed decreases the efficiency of the involved nodes and contributes to complex cleaning of the device. An appraisal using the NSLKDD and UNSW data sets reveals that both models can identify malicious activities with higher precision, detective frequencies and false alerts that offer less than state-of-the-art approaches. Accordingly, with strong defectiveness and false alarm frequency, AIRP is even better than state-of-the-art technology. • Confidence Intrusion Detection and Accuracy System and the Trust based Intrusion Identification and Classification System for Secure Network. • Only the high classification functions are then chosen for any packet obtained in a network that is saved as part of the past output of the node. • AIRP recommends regular machine cleaning, which evaluates and regularly renews the confidence relationships of participant nodes. • AIPR gives the exact time for nodes cleaning states and limits the exposure window of nodes by offering a complex algorithm. • The consistency detection for AIRP, for example, is 81% for UnSW, 83.47% for AODE online, 88% for Cadf, 90% for EDM, 80% for TANN and 72.6% for NB.