Heuristic-Driven Anomaly-Based Intrusion Detection for Predicting and Mitigating Security Threats in Iot Network Transactions

The paper describes an intrusion detection system that works by detecting unusual activities and is created to resist new threats in IoT networks. Identifying the different and scattered attacks that IoT devices encounter is a challenge for regular detection techniques. To fix these limitations, the suggested method presents two major heuristic measures: the Intrusion Scope Heuristic (ISH) and the Benign Scope Heuristic (BSH) which are made through sequential bi-gram analysis and improved by using a floating sequential forward search algorithm. Feature weights projected from these heuristics allow for precise categorization of network records into either benign or intrusive types. Empirical probabilities and projection weight ratios are used in the system to make predictions more precise. Based on experiments using the IoT Sentinel dataset, the method has a 94% accuracy, detects nearly all threats with a sensitivity of 98.8% and misses very few threats. Using this approach guarantees better intrusion detection, introducing promising methods for handling realtime threats and their prevention in IoT networks.