ABSTRACT

This work attempts to mathematically formulize the computation of waiting time of any process in a static n-process, CPU-bound round robin scheme. That in effect, can calculate other performance measures also. An improvement in the existing round robin algorithm has also been worked out that provides priority to processes nearing completion. The suggested approach uses two ready queues, wherein a process is returned to the second ready queue after the completion of its penultimate round. This policy reduces the average waiting time and increases the throughput, in comparison to the conventional round robin scheme, while maintaining the same level of CPU utilization and no substantial increase in the overheads. The mathematical formulation of this policy is transparent enough; it provides, for each process, the actual order of shifting to the second queue and also the order of termination.