Stability of MaxWeight-({\alpha},g)

TL;DR

This paper proves the throughput optimality of the generalized MaxWeight-({\alpha},g) scheduling policy in single-hop switched queueing networks, extending the classical MaxWeight policy to more flexible bandwidth sharing scenarios.

Contribution

It establishes throughput optimality for MaxWeight-({\alpha},g), a generalized, myopic scheduling policy, broadening the applicability of MaxWeight policies to bandwidth sharing networks.

Findings

MaxWeight-({\alpha},g) is throughput optimal.

The policy generalizes classical MaxWeight.

Applicability to bandwidth sharing networks.

Abstract

We consider a single-hop switched queueing network. Amongst a plethora of applications, these networks have been used to model wireless networks and input queued switches. The MaxWeight scheduling policies have proved popular, chiefly, because they are throughput optimal and do not require explicit estimation of arrival rates. In this article, we prove the same throughput optimality property for a generalization of the MaxWeight policy called the MaxWeight-({\alpha},g) policy. These throughput optimal, myopic scheduling policies allow for scheduling choices similar to those found in bandwidth sharing networks -- a further well studied model of Internet congestion.