MCIDS-Multi Classifier Intrusion Detection system for IoT Cyber Attack using Deep Learning algorithm

Shruti Singh, Swedel Viola Fernandes, Vaibhav Padmanabha, PE Rubini

20212021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV)24 citationsDOI

Abstract

The massive increase in development, deployment and usage of IoT has given rise to smart cities. These smart city devices have the ability to perform communications on their own which require free flow of data. The interconnectivity has also resulted in exponential growth of data being processed, making it susceptible to intrusion attacks. Traditional IDS systems are not designed to work efficiently in an IoT network as these devices have restricted resources and sparse functionality. To tackle the cyber security threats in IoT, MCIDS (Multi Classifier Intrusion Detection system) has been proposed which is based on deep learning algorithm. The UNSW-NB15 dataset is utilised to train and test the model. Proposed Solution can effectively detect and alert Reconnaissance, Backdoors, Analysis, DoS, Fuzzers, Generic, Worms and Shellcodes and achieve high accuracy with low false positives.

Topics & Concepts

Computer scienceInterconnectivityIntrusion detection systemSoftware deploymentClassifier (UML)Internet of ThingsMachine learningAttack modelArtificial intelligenceData miningAlgorithmComputer securityOperating systemNetwork Security and Intrusion DetectionAdvanced Malware Detection TechniquesInternet Traffic Analysis and Secure E-voting