Journal of Discrete Mathematical Sciences and Cryptography · 2023 · 47 citations · 0 references
ComputingComputer ScienceCybersecurityComputer SecurityInternet Of ThingsMachine LearningDeep LearningIntrusion DetectionIot SecurityThreat DetectionThreat ModelingInternet Of Things Device SecurityIot Intrusion DetectionMachine Learning For Intrusion DetectionMachine Learning For IotMitigation
The IoT environment is highly vulnerable to attacks, yet existing detection methods are limited, and deep learning approaches require large datasets and long training times, leading to overfitting and poor performance. The study aims to develop a security system capable of detecting and countering both known and unknown threats across the growing number of IoT devices. The authors propose a machine‑learning based self‑protection system to safeguard the IoT environment.
The Internet of Things (IoT) environment must prioritise security because of the IoT’s significant attack susceptibility for a variety of reasons. The IoT attack detection technique or mitigation procedure is the extent of the currently available solutions. However, there are fewer autonomous security provider approaches available, and they are inappropriate for the IoT environment’s evolving threats. Because of this, there has to be a security system in place that can detect and counteract both known and unknown threats for the increasing number of IoT devices. Deep Learning (DL) based intrusion detection systems does not consider attack signatures and normal behavior to obtain detection rules, it requires large data sets for training and takes a longer time to train the data. Many a time, insufficient dataset configuration prompts the minimization of a learning calculation, bringing about over fitting and helpless grouping rates. The goal of the proposed study is to create and implement a machine learning-based self-protection system to safeguard the IoT environment.