Flow Control and Scheduling for Shared FIFO Queues over Wireless Networks

TL;DR

This paper analyzes FIFO queue stability in wireless networks, develops a convex inner-bound for the non-convex stability region, and proposes algorithms that significantly improve throughput.

Contribution

It introduces a convex inner-bound for FIFO queue stability regions and develops flow control and scheduling algorithms that optimize throughput.

Findings

The stability region is non-convex but can be approximated by a tight convex inner-bound.

The proposed algorithms outperform traditional queue-based control and max-weight scheduling.

Simulation results demonstrate significant throughput improvements in wireless FIFO queue networks.

Abstract

We investigate the performance of First-In, First-Out (FIFO) queues over wireless networks. We characterize the stability region of a general scenario where an arbitrary number of FIFO queues, which are served by a wireless medium, are shared by an arbitrary number of flows. In general, the stability region of this system is non-convex. Thus, we develop a convex inner-bound on the stability region, which is provably tight in certain cases. The convexity of the inner bound allows us to develop a resource allocation scheme; dFC. Based on the structure of dFC, we develop a stochastic flow control and scheduling algorithm; qFC. We show that qFC achieves optimal operating point in the convex inner bound. Simulation results show that our algorithms significantly improve the throughput of wireless networks with FIFO queues, as compared to the well-known queue-based flow control and max-weight scheduling.