The design and implementation of an advanced knowledge-based process scheduler

Conventional process schedulers in operating systems divide a machine's CPU resource among processes using a fixed scheduling policy, in which the utilization of a computer system (e.g., a real-time or a timesharing system) is a major concern rather than content or behavior of a process. As a result, the CPU resource is likely to be used in an inefficient manner, or the processing time of a process might be extended unnecessarily. In this paper, we therefore propose a process' behavior-based scheduler that delays process switching in order to allow the object process to continue its execution even though its timeslice has already expired, when it is predicted from an advanced knowledge called PFS (Program Flow Sequence) that the object process needs a little bit more CPU time before it voluntarily relinquishes the CPU. This allows the processing time or the process switching cost of the object process to be reduced.