Dynamic scheduling for a flexible processing network

We study flexible processing systems where there are multiple paths for a certain type of customer to follow through the system. This paper analyzes an open processing network model with discretionary routing. The underlying methodology is the heavy traffic approximation, which has been well publicized by many researchers. Allowing priority sequencing in addition to dynamic routing as our control policy, we show that the balanced workload routing rule does not necessarily give us an optimal solution. In general unbalanced workload routing with priority sequencing is shown to give us better performance than a balanced one. Through some simulation results we compare our recommendation with other scheduling policies for a flexible processing network.