IoT Device Malware Detection Using Soft Computing Learning and Wide Madaline (WML-IOT)
A. Punidha, Easwaramoorthy Arul, E. Yuvarani · Advances in computer science research · 2023
IoT device manufacturers use backdoors, which are covert control techniques, to make their products supportable.But the front window is really for the hackers.Nevertheless, a firmware is installed to lock the back door once the back door has been located.For hackers, these backdoors serve as either a user ID or a password.These malware operate by wiping out the memory of an IoT device, wiping out firewall rules, wiping out network configuration, and stopping the device.It's as damaging as it can be without frying the circuits of the IoT device.For recovery, victims must manually reinstall the system firmware, which is too challenging for most device owners to complete.Many owners of IoT devices should probably discard them because they think they have experienced a hardware failure, not realising that malware has infected them.A firmware attack like this on IoT devices is classified using Wide (Deep) Madaline Learning (WML).A single output unit is labelled malicious or benign by training a Wide Madaline with numerous input clusters that have a malicious or benign API.Then, using broad Madaline learning, this was trained to find a malicious pattern in unidentified IoT firmware.The results show that various IoT device firmware attacks were classified with 97.24% True Positives and 0.07% False Positives.