Designing ARINC653 Partition Constrained Scheduling for Secure Real Time Embedded Avionics

Being a high end embedded system, an avionic system calls for stringent real time constraints as well as secure guarantees. In terms of logical architecture, avionic systems have recently grown into the form of Integrated Modular Avionics (IMA) from the traditional federated avionics system whose redundancy level is overwhelming for modern large aircrafts. The key idea of IMA system lies in the rules of time and space partitioning, which guarantees system predictability and reliability. However, existing industrial practices of IMA partition and priority settings usually incur significant waste of resources, which would eventually lower the performance of IMA tasks in terms of latency or throughput. This issue was not properly addressed by previous researchers who assumed settings of priority variances and fixed partitions, which differ from practical applications. In this paper, a secure real time scheduling scheme with partition readjustment is proposed with inputs of features exhibited by tasks under partition. In our scheme, the resource costs are reduced by merging and restructuring partitions without compromising hard real time constraints. The simulation results of actual flight missions show that significant improvement by our method in terms of the average response time of tasks as well as number of partitions.