4th International ICST Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks

Research Article

Flow-Level Stability of Channel-Aware Scheduling Algorithms

  • @INPROCEEDINGS{10.1109/WIOPT.2006.1666471,
        author={Sem  Borst and Matthieu  Jonckheere},
        title={Flow-Level Stability of Channel-Aware Scheduling Algorithms},
        proceedings={4th International ICST Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks},
        publisher={IEEE},
        proceedings_a={WIOPT},
        year={2006},
        month={8},
        keywords={},
        doi={10.1109/WIOPT.2006.1666471}
    }
    
  • Sem Borst
    Matthieu Jonckheere
    Year: 2006
    Flow-Level Stability of Channel-Aware Scheduling Algorithms
    WIOPT
    IEEE
    DOI: 10.1109/WIOPT.2006.1666471
Sem Borst1,2, Matthieu Jonckheere3,4
  • 1: Bell Laboratories, Lucent Technologies, P.O. Box 636, Murray Hill, NJ 07974-0636, USA
  • 2: CWI, P.O. Box 94079, 1090 GB Amsterdam, The Netherlands
  • 3: Department of Mathematics & Computer Science, Eindhoven University of Technology
  • 4: P.O. Box 513, 5600 MB Eindhoven, The Netherlands

Abstract

Channel-aware scheduling strategies provide an effective mechanism for improving the throughput performance in wireless data networks by exploiting channel fluctuations. The performance of channel-aware scheduling algorithms has mainly been examined at the packet level for a static user population, often assuming infinite backlogs. Recently, some studies have also explored the flow-level performance in a scenario with user dynamics governed by the arrival and completion of random service demands over time. Although in certain cases the performance may be evaluated by means of a Processor-Sharing model, in general the flow-level behavior has remained largely intractable, even basic stability properties. In the present paper we derive simple necessary stability conditions, and show that these are also sufficient for a wide class of utility-based scheduling policies. This contrasts with the fact that the latter class of strategies generally fail to provide maximum-throughput guarantees at the packet level.