Internet of Things device detection

Forbes puts its Internet of Things (IoT) market estimate to $457B by 2020. Yeah that is super exciting, but it is also upending the cybersecurity as we know it! As we connect thousands of new Internet connected devices to our networks, we create potential gaping holes to our network security. The IoT security problem is complex and multifaceted but one essential element of the solution is an accurate classification of IoT devices.  The current State of the Art for IoT classification is using data rates and burstiness, activity cycles, and signaling patterns.  Our intuition and observation was that IoT devices use a highly structured language (TCP/IP) to communicate. First, our project preprocesses this network data into a “sentence” structure.  Then we use Natural Language Processing (NLP) on that “sentence” to classify IoT devices which works with amazing results.  These results are the foundation for anomaly detection to pinpoint a device that misbehaves beyond their intended functionality.