The IoT has become an indispensable part of our lives at work and in our home applications. Due to the need for many IoT devices, IoT manufacturers are least concerned about security vulnerabilities during designing and developing these devices. Because of this, it becomes easier for adversaries to manipulate the hardware and insert Trojans or Remote File Inclusion to control remotely. This thesis aims to build a model to identify hardware Trojans in IoT devices using multiple machine learning models. We used different machine learning models to evaluate the performance and accuracy. In addition, we chose a distinctive feature that can detect the presence of Trojan in these devices. The proposed model is estimated using a smart city testbed and existing and real-time datasets generated. The testbed used was designed to simulate and assess the Hardware trojan attacks such as the DOS attack and covert channel attack. We could measure the power profile and network traffic on the IoT gateway device to analyze the performance and accuracy using the real-time dataset to detect the attacks.