Camera Authentication Using Physical Features based on Handle System ID

Surveillance cameras are now heavily employed in IoT systems, which is crucial for security and other smart services. However, its current authentication system is vulnerable to the impersonation attack from fake devices, whose MAC/IP address and product serial number are identical to legitimate devices. Thus, we present a novel authentication mechanism based on the fingerprints extracted from physical sensor features, which can distinguish the fake devices with exactly the same configurations. Specifically, we propose: 1) a mechanism to extract the sensor fingerprint of a camera applying Photo Response Non-Uniformity and using it in Handle system 2) a mechanism Handle System ID to identify the IoT device with physical features, device state information 3) a simple, yet practical and efficient technique to match a camera device with a given Handle ID. Experimental results illustrate that our mechanism can work more than 10 times faster than conventional methods. The accuracy of rogue connection detection is up to 95.6% and false rejection is below 2.8%.