ROUTING AND BUFFER ALLOCATION MODELS FOR A TELECOMMUNICATION SYSTEM WITH HETEROGENEOUS DEVICES

A branching system which consists of several parallel nonidentical communication devices with finite input buffers is considered in this paper. The system is designed and operated under a stochastic congestion control policy. Several job routing and buffer allocation decision models are analyzed in order to maximize the aggregate system throughput, or to minimize the storage space costs. The polynomial and pseudo-polynomial optimization algorithms which are presented here take advantage of the structural properties of this system. The results obtained are also of particular relevance in the design of load balancing policies for distributed computer systems.