Dependency and utilization aware Task Allocation for Multi-core Embedded Processors

Multi core technology in processor improves computational capabilities of the Embedded Systems. As the number of cores increases, dependency, power management and memory or cache misses become major issues. In this paper, the aperiodic and periodic task allocations to multiple cores are determined based on utilization factor and dependencies among them. A dependency and utilization-aware task allocation algorithm is proposed to improve processor utilization and decreases memory access time. The periodic and aperiodic tasks and their priorities based on the safety criticality are considered for task allocation. They are allocated to likely cores based on the CPU utilization bound of the dependent tasks thereby improves the response time of the multi-core processor. The performance metrics in terms of utilization, response time and power consumption are observed. In this paper, for a given task set, the optimal number of cores required to schedule periodic and aperiodic tasks are estimated and the allocation table is obtained.