A Review on Deep Learning and Intrusion Detection System Technologies to Secure IoT

Abstract

Today the usage of internet has increased and the emergence of new technologies has invaded computer systems and networks.  IoT  is a new technology which   opens up opportunities for new services and new innovations that is  enabled by the developments in RFID, smart sensors and  communication technologies. The fundamental aspect is to have smart sensors that communicate directly without human intervention to deliver a new application. All objects will be connected and are able to communicate with each other, while they operate in unprotected environments. This aspect leads to major security challenges.  Companies are increasingly investing in these areas of research to optimize the detection of these attacks. Intrusion Detection Systems (IDS) are a vital tool for the protection of networks and data. Insights derived from the raw IoT data is highly complex that goes beyond the competence of traditional data analytical paradigms. Deep Learning models are better than conventional machine learning paradigms in the following ways. First, they mitigate the requirement for supervised feature sets to be utilized for training so that the features that might not be recognizable to a human can be extracted smoothly by Deep Learning models.  This work  focuses  on the review  related to  IoT, IDS and Deep learning, traversing different  areas  related   to security issues in IoT domain.