Dynamic scheduling of multiclass queueing networks

This thesis presents several dispatch policies for multi-class queueing networks and studies their stability properties. The discrete proportional processor sharing (DPPS) dispatch policies are shown to be stable as long as each server's traffic intensity is less than one. Any policy can be embedded into a DPPS policy and the new policy is always stable. Largest weighted upstream unbalance (LWUU) policies and largest weighted total unbalance (LWTU) policies are two families of policies, which use downstream or upstream information. For more restricted network models, we show that both LWUU and LWTU policies are stable.; In a batch processing network, multiple jobs can be formed into a batch to be processed in a single service operation. A batch policy specifies which class of jobs is to be served next. Throughput of a batch processing network depends on the batch policy used. When the maximum batch sizes are equal to one, the corresponding network is called a standard processing network, and the corresponding service policy is called a dispatch policy. This thesis shows that any normal dispatch policy can be converted into a batch policy that preserves key stability properties. Examples of normal policies include static buffer priority (SBP), first-in–first-out (FIFO) and generalized round robin (GRR) policies.; A series of simulation studies is conducted in a three-product-five-station network. The simulation results show that DPPS dispatch policies and their induced batch policies have good performance.