Optimization problems on the performance of a nonreliable terminal system

The aim of this paper is to investigate the effect of the different service disciplines, such as first in first out, processor sharing, Priority Processor Sharing, Polling, on the main performance measures, such as utilizations, response times, throughput, mean queue length. It has been shown by numerical examples that even in the case of homogeneous sources and homogeneous failure and repair times, the CPU utilization depends on the scheduling discipline contrary to the case of reliable terminal systems. All random variables involved in the model construction are supposed to be exponentially distributed and independent of each other.