Maximum-Throughput Access Control in Wireless LANs Through Max-Weight-Inspired Policies

Abstract

The design of efficient medium access control (MAC) protocols that are backward compatible to legacy IEEE 802.11-based protocols is a central issue in the literature. This paper draws inspiration from utility-driven max-weight policies that lead to the maximum throughput region in synchronized slotted systems. It introduces two important enhancements to the IEEE 802.11 MAC protocols, which are not considered in current IEEE 802.11 protocol versions and 802.11x enhanced versions: consideration of the transmitter-receiver link state and the transmitter backlog size. Our objective is to characterize the impact of these parameters on access control and ultimately rely on them to enhance the capacity region and, thus, the total throughput, compared with legacy 802.11x MAC protocols. We consider a scenario of multiple nodes attempting to access an access point (AP). Each node is characterized by a link quality to the AP, which is essentially mapped to the physical (PHY)-layer rate that can be supported and a queue length that evolves based on the packet-arrival process at the node and the transmissions. Motivated by the max-weight access control policy, we aim to increase the total uplink throughput while ensuring bounded queue lengths by making slight modifications to the legacy IEEE 802.11 protocol. Specifically, we make the contention window (CW) dependent on queue size and PHY-layer rate. The key idea is that a node with a large PHY transmission rate and large queue size should tend to use a smaller CW so that it gets higher chances of channel access. We take a first step toward this approach and suggest heuristic queue size and link-state-aware rules for defining the CW. Experimental performance evaluation shows significant enlargement of the capacity region and substantial performance improvement in the total throughput under our policy, compared with current protocols.