Securing IoT using layer characterstics

IoT has gained significant popularity and securing such mission critical network remains a big concern. There are already reports of misusing such system. Recently there were reports of sending spam from a refrigerator. Bigger risk of security in IoT is that unrelated systems are connected and if any attacker gets into one object, it could result in harming much bigger segment. Also such connected networks will really be disconnected in terms of management and applied policies. So applying secured policy at each network segment will only help at certain level but it would not be feasible to secure whole ecosystem of IoT devices/objects. Security has been identified as one of the biggest challenge in IoT segment and there are no solid specs or techniques to come over these issue end to end. This paper outlines use of machine learning combining with interesting dimensions like layer characteristics of IoT device's network stack. Paper also provides example of layer characteristics, which are specific to a device, and creating a baseline of such characteristics using machine learning algorithm will help in establishing the identity of IoT devices without the need of sophisticated PKI infrastructure.