7th International Conference on Collaborative Computing: Networking, Applications and Worksharing

Research Article

Decentralized Work-In-Process Optimization in Cooperative Resource Allocation

Download590 downloads
  • @INPROCEEDINGS{10.4108/icst.collaboratecom.2011.247084,
        author={Doraid Dalalah},
        title={Decentralized Work-In-Process Optimization in Cooperative Resource Allocation},
        proceedings={7th International Conference on Collaborative Computing: Networking, Applications and Worksharing},
        publisher={IEEE},
        proceedings_a={COLLABORATECOM},
        year={2012},
        month={4},
        keywords={resource allocation parallel computing primal dual optimization cooperative control},
        doi={10.4108/icst.collaboratecom.2011.247084}
    }
    
  • Doraid Dalalah
    Year: 2012
    Decentralized Work-In-Process Optimization in Cooperative Resource Allocation
    COLLABORATECOM
    ICST
    DOI: 10.4108/icst.collaboratecom.2011.247084
Doraid Dalalah,*
    *Contact email: doraid@just.edu.jo

    Abstract

    Abstract—Resource allocation entails deciding how to split a resource of restricted availability among various demands in a way that optimizes current objectives. In this paper, we focus on one type of distributed resource allocation problems in which a distributed system comprising networked heterogeneous agents and processors/servers where the agents strive to boost their efficiencies by issuing more work transactions for higher throughput. Each agent can issue work transactions which comprise a set of tasks that have to be completed by a networked set of servers. The agents get more utilities as their transactions delivery rate increases, however, network administration entails a set of constraints on the tolerable delay on the transactions. An optimization model is constructed to characterize the system in which the available servers are allocated to the present agents. The optimization model is solved in a decentralized way so that the agents can work separately to maximize the global benefit measure of the network. The presented model numerical solution is put together using ARENA simulation package and experimented for different network topologies. The simulation results show fair allocation of the resources whereas the anonymous agents work in parallel to achieve optimality.