ICMetric Technology for Embedded Devices Security

Abstract

The interesting in embedded systems is related to the matters of providing privacy, safety, and security. The security of any data kept on the system can be exposed by any unauthorised access. It was found that no security system is perfect for all applications; this keeps the research area open for suggesting a new security system or improving an old one, therefore, a security system is suggested in this thesis. The proposed system is presented to apply identification and improve the security of embedded devices. However, the proposed system is based on a new technology called an Integrated Circuit Metric (ICMetric), which relies on the special internal features of each device. By using the ICMetric, low level device features are used to create an identification for device. The proposed work is based on using the bias in accelerometer, gyroscope, and magnetometer to create a unique ICMetric number for every device. This number is utilised in dataset extracted from trace file that has been generated by Network Simulator Version two (ns-2) to perform identification and detection. The proposed system is composed through three main phases. The first phase is bias readings extracted from sensors. Whereas, in the second phase, ICMetric number is generated by using the bias readings that extracted from sensors in the first phase. In the third phase, the security system is tested and evaluated to measure its effectivity. In other words, it is tested with dataset that extracted from the trace file of network simulator. In this phase performance metrics are calculated, which are error rate, confused matrix and accuracy rate. An identification system is proposed in this thesis for providing identification of intelligent wheelchairs in two schemes. The first proposed scheme is based on the bias readings that generated by gyroscope and magnetometer sensors. Whereas, the second proposed scheme is depended on three ICMetric numbers provided by bias readings generated from three types of sensors. The proposed schemes have been simulated by using the dataset extracted from trace files. The simulation results have been compared and studied for high levels of detection capability and performance metrics.