Efficient Parallel Iterated Greedy Algorithm for Solving Task Assignment Problem in Heterogeneous Systems

Efficient allocation or assignment of tasks has been a constant problem for research in the domain of Combinatorial Optimization. With Task Assignment being an NP-Hard Problem for more than 3 processors, considerable effort has gone into developing heuristic algorithms that provide an efficient and approximate solutions that match the expected values. Currently trending algorithms utilize Graph Matching and Graph Partitioning techniques to achieve the assignment of Tasks. We have presented an algorithm that is different from the Graph Based methodologies and instead uses Greedy Algorithm to achieve the effective task assignment. This is a Parallel Iterated Greedy Algorithm which minimizes the cost of computation and at the same time provides faster convergence to a value that is closer to the theoretically computed value. The algorithm was implemented using Message Passing Interface (MPI)