Distributed Scheduling Strategies for Processing Multiple Divisible Loads with Unknown Network Resources

The problem of scheduling multiple divisible loads on bus networks with unknown parameters is addressed in this paper. Most research in optimal partitioning of massive loads to be distributed among processors in a distributed system is under the assumptions of known and deterministic network resources. Lately, researchers have started addressing issues that are application oriented and go beyond those simple assumptions of load distribution. In this paper, a heterogeneous network with unknown network resources is considered, and several strategies with increasing refinements that minimize the total processing time of multiple divisible loads are designed. These strategies perform close to the real distributed systems more and more. Application comparison of these proposed strategies is carried out.