الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 109 |  |  |
| | | from | 109 | | |
Abstract
The world has recently witnessed a trend toward embedded systems and the Internet of Things (IoT). These two have pervaded our surroundings; most technological appli- cations contain one or more embedded systems. Mobile phones, autonomous vehicles, Automatic Teller Machines (ATMs), wearable health and fitness monitoring devices, and many other items around us, have embedded systems. If you consider the preceding examples closely, you will notice that the erroneous performance of these devices may expose the lives of their users to significant harm in the event of unavoidance or indul- gence and may result in the loss of life. Therefore, paying attention to the dependability of embedded systems is an essential phase in developing embedded systems. Software dependability reflects the degree of the user’s trust in the system. Fault Injection (FI) is a superior mean of evaluating the system’s dependability. An artificial fault is inserted into different locations in the system, monitoring its propagation and observing the sys- tem’s behavior in the presence of the fault. FI techniques can be mainly categorized into four different techniques: Hardware implementation-based Fault Injection (HFI), Soft- Ware Implementation-based Fault Injection (SWIFI), hardware Emulation-based Fault Injection (EFI), and Simulation-based Fault Injection (SFI); each has merits and de- merits.
The SWIFI approach is investigated in our thesis using virtual platforms such as Quick EMUlator (QEMU) software. QEMU emulates multiple architectures, including I/O peripherals, in order to execute a complete operating system or boot firmware. In ad- dition, throughout this thesis, we make it a point to ensure that the fault simulation is carried out in a manner that is compliant with the ISO 26262 functional safety of road vehicle standard. The standard describes the principles and tests that must be conducted in order to certify software failure-proof. QEMU-based FI permits injecting fault in the early steps of design and implementation with no need for the availability of a physical prototype of the target system, providing sufficient controllability and ob- servability in addition to accurate results. Here, we propose a fault injection framework based on virtual platforms to evaluate the dependability of embedded software for ex- ternal peripheral devices. The work proposed in this thesis concentrates on emulating the ARM environment and a number of external peripheral devices. The framework is used to inject permanent stuck-at and intermittent fault models into external peripheral devices.