A learning automata based framework for task assignment in heterogeneous computing systems
Raju D. Venkataramana, Meghna Ranganathan · 1999
In this paper, a framework for task assignment in heterogeneous computing (HC) systems is presented.This framework is based on a learning automata model.The proposed model can be used for dynamic task assignment and scheduling and can adapt itself to changes in the hardware or network environment.An important feature of this framework is that it works for anv cost metric.This could either be a general metric like minimizing the total execution time, or an application specific metric.The HC system itself is modeled using a task flow graph (TFG) and a processor graph(PG).The TFG models the application while the PG models the network of processors.The learning automata model is constructed by associating every task in the TFG with a variable structure learning automaton [7].The actions of each of these automata correspond to the nodes in the PG.The objective is to optimize the cost criterion by guiding the learning of the automata.The paper presents different heuristic techniques to achieve this.The performance analysis of these techniques are discussed with reference to a specific cost metric.