Fair scheduling in wireless packet data networks
Handbook of wireless networks and mobile computing
| Hi-index | 0.00 |
The increasing number of wireless data users and the deployment ofbroadband wireless networks have brought the issue of providingfair channel access to the forefront. Wireless fair schedulingalgorithms must deal with unique domain-specific issues such aslocation-dependent and bursty channel error. In the past, a numberof scheduling disciplines have been proposed for approximatingfluid fair queueing in wireless channels, but using only theARQ-based link-layer retransmit model (Bhagwat et al., in: IEEEINFOCOM, 1996; Lu et al., in: ACM SIGCOMM, 1997; Ng et al., in:IEEE INFOCOM, 1998; Lu et al., WINET 6, 2000, pp. 323-343;Nandagopal et al., in: ACM MOBICOM, 1999).In this work, we generalize past work in three key ways: (a) weuse the notion of utility-based fair scheduling, wherein the goalof the scheduler is to serve flows in a way that maximizes theaggregate utility of the system, (b) we combine the techniques ofFEC and ARQ in a generalized framework in which differentscheduling disciplines can be realized by mixing and matchingcomponent algorithms within the framework, and (c) we support thenotion of application level fairness when flows see differentchannel error probabilities.The increasing number of wireless datausers and the deployment of broadband wireless networks havebrought the issue of providing fair channel access to theforefront. Wireless fair scheduling algorithms must deal withunique domain-specific issues such as location-dependent and burstychannel error. In the past, a number of scheduling disciplines havebeen proposed for approximating fluid fair queueing in wirelesschannels, but using only the ARQ-based link-layer retransmit model(Bhagwat et al., in: IEEE INFOCOM, 1996; Lu et al., in: ACMSIGCOMM, 1997; Ng et al., in: IEEE INFOCOM, 1998; Lu et al., WINET6, 2000, pp. 323-343; Nandagopal et al., in: ACM MOBICOM, 1999). Inthis work, we generalize past work in three key ways: (a) we usethe notion of utility-based fair scheduling, wherein the goal ofthe scheduler is to serve flows in a way that maximizes theaggregate utility of the system, (b) we combine the techniques ofFEC and ARQ in a generalized framework in which differentscheduling disciplines can be realized by mixing and matchingcomponent algorithms within the framework, and (c) we support thenotion of application level fairness when flows see differentchannel error probabilities. Specifically, we explore the combination of adaptive FEC andARQ-based schemes within the utility-based wireless fair schedulingparadigm. We show through simulation and analysis that thiscombination enables the scheduling disciplines to optimize theaggregate system utility and achieve application level fairness,while handling varying channel errors effectively. Specifically, weexplore the combination of adaptive FEC and ARQ-based schemeswithin the utility-based wireless fair scheduling paradigm. We showthrough simulation and analysis that this combination enables thescheduling disciplines to optimize the aggregate system utility andachieve application level fairness, while handling varying channelerrors effectively.