الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 79 |  |  |
| | | from | 79 | | |
Abstract
The complexity of embedded real-time systems has increased, and is expected to handle growing number of diverse applications. Most of these applications have large diversity in execution times of their tasks. Traditional scheduling techniques, such as Fixed-Priority Scheduling, Earliest Deadline First Scheduling (EDF), Rate Monotonic Scheduling (RMS), etc., do not satisfy the requirements of such applications. In most traditional scheduling techniques, one or group of tasks may dominate the CPU resources regardless of its criticality, which is called monopolism. In this context, the timing requirements of each application in the system should be isolated and guaranteed.
The Hierarchical Scheduling Framework (HSF) is an efficient solution for scheduling tasks of complex real-time systems. To avoid the interference between independent subsystems (applications) and guarantee a budget for each processor without preemption, HSF supports the concept of temporal isolation where each subsystem executes only in its server (virtual partitioning period).
This thesis proposed, designed and implemented an Adaptive Hierarchical Scheduling Framework based on EDF scheduler (AHSF-EDF), which creates and guarantees a virtual temporal isolation for each subsystem, thus resolving the problem of monopolization. AHSF-EDF implemented an adaptive budget controller that automatically and periodically adjusts the budget of each subsystem, to reserve resource wasting, based on Chebyshev’s estimator; a prediction algorithm for tasks execution times. Implemented into the kernel of TI-RTOS on a resource constrained platform, experiments show that the proposed scheme provides good performance for different applications with dynamic tasks under normal and overload conditions.
As an enhanced version of AHSF-EDF, we proposed an Adaptive Hierarchical Scheduling Framework based on EDF with Virtual Deadline (AHSF-VD) that dynamically adjusts the CPU budget of each server. Different relative deadlines are assigned to tasks depending on their criticality modes. Dual-criticality levels (low-criticality LO and high-criticality HI) are considered, where the virtual relative deadlines for high-criticality tasks are generated by greedy tuning algorithm. The proposed AHSF-VD framework implemented into the kernel of TI-RTOS, and tested in a real platform, is found to guarantee the minimum budgets for high-criticality servers during overload periods, and ensure that high-criticality tasks meet their deadlines with no miss ratios at the expense of low-criticality tasks.
List of Publications
• Hesham Hussien, Eman Shaaban and Said Ghoniemy. ”Adaptive Hierarchical Scheduling Framework for TiRTOS”. In The International Journal of Embedded and Real-Time Communication Systems (IJERTCS), Volume 10, Issue 1, Article 7, 2019. 121117-045223.
• Hussien, H., Shaaban, E., & Ghonaimy, S. (2018, December). Mixed-criticality Hierarchical Scheduling for TI-RTOS. In 2018 13th International Conference on Computer Engineering and Systems (ICCES) (pp. 279-283). IEEE.