SchedCust: Design & Development of scheduling policy customization framework for ARM based System on Chip

Jasleen Kaur, SRN Reddy
2020 Procedia Computer Science  
Selecting a right scheduling strategy for the specific embedded application has a significant impact on system's performance. To adapt to the requirements of a particular embedded application, the embedded operating system can be customized manually or automatically. In this paper, the standard process for customizing the CPU scheduling policy in an open source Linux operating system (OS) has been discussed. The paper also focuses on the design & development of the CPU scheduling customization
more » ... ramework for ARM based Raspberry Pi 3, tagged as SchedCust, that allows the CPU scheduler to be configured to execute a particular scheduling policy. The SchedCust framework helps the users to automatically develop the scheduling policy specific custom operating system (OS) for the selected processor. The experimental results have been computed for FIFO and RR scheduling policy, with respect to four distinct parameters. The data collected from user testing showed that the manual process of developing a custom OS takes more time and efforts when compared to the automated process. Abstract Selecting a right scheduling strategy for the specific embedded application has a significant impact on system's performance. To adapt to the requirements of a particular embedded application, the embedded operating system can be customized manually or automatically. In this paper, the standard process for customizing the CPU scheduling policy in an open source Linux operating system (OS) has been discussed. The paper also focuses on the design & development of the CPU scheduling customization framework for ARM based Raspberry Pi 3, tagged as SchedCust, that allows the CPU scheduler to be configured to execute a particular scheduling policy. The SchedCust framework helps the users to automatically develop the scheduling policy specific custom operating system (OS) for the selected processor. The experimental results have been computed for FIFO and RR scheduling policy, with respect to four distinct parameters. The data collected from user testing showed that the manual process of developing a custom OS takes more time and efforts when compared to the automated process.
doi:10.1016/j.procs.2020.04.068 fatcat:usrozn4lxvdkpnqkrlifl6yqly