I am interesting in IOT security and machine learning/deep learning. I am looking for good project ideas to work on. example: http://projectabstracts.com/20138/a-malicious-pattern-detection-engine-for-embedded-security-systems-in-the-internet-of-things.html
closed as too broad by Steffen Ullrich, tlng05, Anders, Polynomial, Teun Vink Dec 26 '17 at 11:14
Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.
IoT security is currently so poor that there are endless opportunities for improving the state of the art using machine learning.
First, pick an area of IoT that you're interested in. I'd recommend you pick one where you already have existing systems you can leverage, and where monitoring/sniffing tools are widely available. Do you have a home automation system? Do you have a car with a CANBUS network? Do you wear a personal health tracker? Do you have web cameras monitoring your property? It's easiest to start when you already know something about the domain you're trying to protect.
Next, develop a simple threat model. Diagram the system's inputs and outputs. Identify network protocols. How is each trust boundary crossing step authenticated? How are the messages protected? Make some assumptions and do some reading: if you think the devices are simple, and they talk to each other, they might be less secure than if they authenticate to a cloud server, therefore they might benefit from your tool's protection.
Now, see if you can insert some kind of sensor in between the devices and their controllers or clients. For example, if they're connected via WiFi and use IP, you could attach a bro sensor to the network switch that connects them. Commercial tools could help here; for example the Packet Squirrel might be a cheap way to sniff packets from the middle of an Ethernet cable. If the devices talk via RF, can you use an SDR to monitor the airwaves?
The important thing to develop is the capability to collect traffic. Also, take into account the ways you might analyze the traffic. Is timing important? Make sure your packet capture tool has precision timing capability. Are electrical or RF characteristics important? You might need specialized equipment; or at least more sensitive than a $20 RTL-SDR dongle.
Once you have traffic in hand, start applying your analytic skills. What kinds of patterns are normal? What kinds of attacks show patterns that are abnormal? Are there regular time-based packets, where fast or slow responses stand out? Are there particular sequences of packets that you can consider normal or abnormal? Can you train your neural network to be silent in the presence of normal traffic? Can you come up with a generic solution that doesn't generate false positives on several systems, not just yours? Most importantly, when you attack the system, does your neural network detect it?
Again, you'll probably have the most success if you start with something you already know and enjoy. Good luck!