[Paper Review] Fairness Provision in the IEEE 802.11e Infrastructure Basic Service Set

Feyza Keceli, Inanc Inan+1 more|ArXiv.org|Apr 13, 2007

Wireless Networks and Protocols26 references3 citations

TL;DR

This paper proposes a model-assisted, measurement-based dynamic EDCA parameter adaptation algorithm for IEEE 802.11e infrastructure BSS that achieves weighted fair resource allocation between uplink and downlink flows of the same Access Category while preserving inter-AC prioritization. The algorithm dynamically adjusts Contention Window (CW) and TXOP limits at the Access Point based on network conditions and traffic type (UDP vs. TCP), ensuring fairness independent of round-trip delay and maintaining QoS for real-time flows.

ABSTRACT

Most of the deployed IEEE 802.11e Wireless Local Area Networks (WLANs) use infrastructure Basic Service Set (BSS) in which an Access Point (AP) serves as a gateway between wired and wireless domains. We present the unfairness problem between the uplink and the downlink flows of any Access Category (AC) in the 802.11e Enhanced Distributed Channel Access (EDCA) when the default settings of the EDCA parameters are used. We propose a simple analytical model to calculate the EDCA parameter settings that achieve weighted fair resource allocation for all uplink and downlink flows. We also propose a simple model-assisted measurement-based dynamic EDCA parameter adaptation algorithm. Moreover, our dynamic solution addresses the differences in the transport layer and the Medium Access Control (MAC) layer interactions of User Datagram Protocol (UDP) and Transmission Control Protocol (TCP). We show that proposed Contention Window (CW) and Transmit Opportunity (TXOP) limit adaptation at the AP provides fair UDP and TCP access between uplink and downlink flows of the same AC while preserving prioritization among ACs.

Motivation & Objective

• To address the inherent uplink/downlink unfairness in IEEE 802.11e infrastructure BSS where downlink flows receive significantly less bandwidth than uplink flows under high load.
• To develop an analytical model that calculates EDCA parameter settings (CW, TXOP) to achieve a desired utilization ratio between uplink and downlink flows for a given AC.
• To design a dynamic, measurement-based adaptation algorithm that adjusts EDCA parameters in real time based on simple network measurements, improving fairness in real-world scenarios.
• To differentiate adaptation strategies between UDP and TCP flows due to their distinct transport-layer behaviors, particularly in ACK-dependent feedback mechanisms.
• To ensure the solution remains fully compliant with the IEEE 802.11e standard while allowing non-power-of-two CW values at the AP for fine-grained fairness control.

Proposed method

• Develops a simple analytical model to compute EDCA parameter settings (CW and TXOP) that achieve a target utilization ratio between uplink and downlink flows for a given Access Category.
• Proposes a model-assisted dynamic adaptation algorithm that uses real-time measurements of transmission durations and flow behavior to adjust CW and TXOP values at the AP.
• Differentiates adaptation logic between UDP and TCP flows: UDP flows are treated with fixed CW and TXOP settings, while TCP flows use CW and TXOP values derived from the analytical model to balance fairness and throughput.
• Enables the AP to use any CW value (not restricted to powers of two), allowing fine-tuned control over fairness ratios, unlike stations which must use powers of two due to beacon field constraints.
• Employs a feedback mechanism that monitors flow completion times and delays to validate fairness and QoS maintenance under dynamic load conditions.
• Validates the approach through extensive simulations across various traffic mixes, including VoIP, video, and data flows, under varying network delays and load levels.

Experimental results

Research questions

• RQ1 How does the default IEEE 802.11e EDCA configuration lead to unfair bandwidth sharing between uplink and downlink flows in an infrastructure BSS?
• RQ2 Can an analytical model accurately predict EDCA parameter settings (CW and TXOP) that achieve a desired utilization ratio between uplink and downlink flows?
• RQ3 How can dynamic EDCA parameter adaptation be designed to maintain fairness in real-world WLANs where ideal assumptions break down?
• RQ4 What are the distinct impacts of UDP and TCP transport behavior on fairness in uplink/downlink flow allocation, and how should the adaptation algorithm account for these differences?
• RQ5 To what extent does the proposed dynamic algorithm maintain QoS for real-time flows (e.g., VoIP, video) while ensuring fair data flow access?

Key findings

• The default EDCA configuration results in uplink/downlink unfairness, with each downlink flow receiving only 1/(n+1) of the total channel access, where n is the number of stations, leading to significantly lower downlink throughput and higher delay.
• Sticking to fixed EDCA parameters, even with optimized settings, fails to achieve fairness regardless of the number of active flows or network delay variations.
• The proposed model-assisted dynamic algorithm achieves fair resource allocation between uplink and downlink flows across a wide range of scenarios, including high loads and varying wired link delays.
• Short TCP flows experience significantly reduced completion times and no connection timeouts under the proposed algorithm, demonstrating effective short-term fairness.
• The algorithm maintains QoS for real-time flows (e.g., VoIP, video) with minimal delay increase—within acceptable limits—while ensuring fair data flow access.
• The solution is fully compliant with IEEE 802.11e and allows the AP to use non-power-of-two CW values, enabling fine-grained fairness control not possible with standard implementations.