Elsevier

Computer Communications

Volume 75, 1 February 2016, Pages 39-49
Computer Communications

A control theoretic approach to achieve proportional fairness in 802.11e EDCA WLANs

https://doi.org/10.1016/j.comcom.2015.11.002Get rights and content

Abstract

This paper considers proportional fairness amongst ACs in an EDCA WLAN for provision of distinct QoS requirements and priority parameters. A detailed theoretical analysis is provided to derive the optimal station attempt probability which leads to a proportional fair allocation of station throughputs. The desirable fairness can be achieved using a centralised adaptive control approach. This approach is based on multivariable state-space control theory and uses the Linear Quadratic Integral (LQI) controller to periodically update CWmin till the optimal fair point of operation. Performance evaluation demonstrates that the control approach has high accuracy performance and fast convergence speed for general network scenarios. To our knowledge this might be the first time that a closed-loop control system is designed for EDCA WLANs to achieve proportional fairness.

Introduction

Enhanced Distributed Channel Access (EDCA) was proposed in IEEE 802.11e-2005 standard to support QoS enhancement and service differentiation for WLAN applications [1]. It extends the basic Distributed Coordination Function (DCF) by classifying traffic flows into four different Access Categories (ACs), namely voice, video, best-effort and background. Traffic with higher QoS requirements, e.g. shorter delay deadline, is assigned a higher priority, and hence, on average, waits for less time before being sent to the channel. This mechanism is beneficial for high-priority traffic. Compared to the DCF, EDCA sacrifices the performance of low-priority traffic to some extent to provide QoS support for high-priority traffic. When the network is saturated with a large proportion of high-priority flows, an extremely unfair scenario will appear, in which the channel will be almost completely occupied by high-priority flows, e.g. VoIP or video streaming flows, however low-priority traffic, such as email or web browsing data, will suffer severe starvation.

Resource allocation in EDCA WLANs has therefore been the subject of considerable interest. The objective is to seek for a fair allocation of network resources (e.g. throughput, airtime and etc.) amongst different traffic types, and meanwhile, guarantee the specific QoS requirements and service differentiation. This paper considers proportional fair allocation of station throughputs amongst ACs for provision of distinct average delay deadlines and priority parameters. The 802.11e EDCA standard specifies four contention parameters to distinguish priority levels, which are minimum Contention Window (CWmin), maximum Contention Window (CWmax), Arbitration Inter Frame Space (AIFS) and maximum Transmission Opportunity (TXOP). A set of default values for the four parameters are recommended in the standard for each physical (PHY) layer supported by 802.11e. As the default values do not take into account the varying WLAN conditions, and thus lead to suboptimal performance and no fairness guarantees, in this paper we find the optimal CWmin value that leads to proportional fair allocation of station throughputs while assuming AIFS and TXOP taking the recommended values and CWmax=CWmin. The optimal CWmin value corresponds to an optimal station attempt probability which is derived from the proportional fairness analysis.

In order to implement the derived proportional fair allocation in practice, a centralised adaptive approach which uses multivariable state-space control theory is then proposed. The WLAN is represented as a discrete multi-input multi-output (MIMO) linear time-invariant (LTI) state-space model. A state feedback control method, the Linear Quadratic Integral (LQI) control, is used to tune the CWmin value to drive the station attempt probability to the optimum so as to maintain a fair throughput allocation. We have demonstrated in simulations that the proposed control approach is adaptive to general network scenarios with high accuracy and fast convergence speed. To our knowledge this might be the first time that a closed-loop control system is designed for EDCA WLANs to achieve proportional fairness amongst ACs.

The remainder of this paper is organised as follows. Section 2 gives a comprehensive review on the state-of-the-art research on fairness and control theory approaches to solve network problems. Section 3 presents the theoretical analysis to derive the optimal station attempt probability which leads to proportional fair allocation of station throughputs given the constraints on the average delay deadlines for different ACs. Section 4 describes a centralised adaptive control approach which can realise the proportional fair allocation derived from Section 3 in real networks. Section 5 evaluates the performances of the fairness algorithm and the proposed centralised control approach, and Section 6 concludes the paper.

Section snippets

Fairness

Fairness has been the subject of a considerable body of literature on 802.11 WLANs [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19]. The unfairness behaviours may be caused by a number of factors, e.g. hidden terminals, exposed terminals, capture, uplink/downlink unfairness, asymmetric radio conditions and multiple data rates and etc., which have been investigated in [2], [4], [5], [6], [7], [8], [11], [13]. There also exist distinct fairness

Network model

We consider a single-hop 802.11e EDCA WLAN with one AP and n client stations, as depicted in Fig. 1. The channel is assumed to be error-free for all supported PHY rates. Traffic flows are classified into N different ACs. We assume that each client carries flows of a single AC, so there are no virtual collisions in our setup. The number of stations in the ith AC is ni. The total number of stations is thus n=i=0N1ni. The analysis can be readily generalised to encompass situations where client

Centralised closed-loop control approach

In this section, we design a centralised adaptive control approach to implement the desirable proportional fairness in real networks. Based upon the analysis in Section 3, the proportional fairness is achieved when the station attempt probability parameter α reaches its optimum value α*. The variable α is only determined by the minimum contention window Wi with AIFS and TXOP taking the recommended values and CWmax=CWmin. Our approach uses a multivariable closed-loop control system to tune W to

Throughput and delay performance

The main objective of this work is to achieve proportional fair allocation of station throughputs while satisfying specific delay constraints of different ACs. To verify if the proposed fairness algorithm meets this objective, we first evaluate the throughput allocation and delay performance. The results are obtained using Matlab based on the throughput and delay analysis in Sections 3.2, 3.3 and the proportional fairness algorithm described in Section 3.4. As the throughput and delay analysis

Conclusions

This paper considers using a closed-loop control approach to achieve proportional fair allocation of station throughputs in a multi-priority EDCA WLAN. The optimal station attempt probability that leads to proportional fairness is derived given the average delay deadline constraints of different ACs present in an WLAN. To achieve the desirable proportional fairness, a centralised adaptive control approach is proposed. The WLAN is represented as a discrete MIMO LTI state-space model. The LQI

Acknowledgements

This work was conducted as part of the project ISS-EWATUS (issewatus.eu) and has been financially funded by European Commission through the FP7 program (grant agreement no: 619228).

References (38)

  • C. Wang

    Achieving per-flow and weighted fairness for uplink and downlink in IEEE 802.11 WLANs

    EURASIP J Wirel. Commun. Netw.

    (2012)
  • W. Li et al.

    AP association for proportional fairness in multirate WLANs

    IEEE/ACM Trans. Netw.

    (2013)
  • F. Kelly

    Charging and rate control for elastic traffic

    Eur. Trans. Telecommun.

    (1997)
  • A. Banchs et al.

    Proportional fair throughput allocation in multirate IEEE 802.11e wireless LANs

    Wirel. Netw.

    (2007)
  • J.F. Lee et al.

    Proportional fairness for QoS enhancement in IEEE 802.11e WLANs

    Proceedings of the IEEE Conference on Local Computer Networks

    (2005)
  • V.A. Siris et al.

    Optimal CWmin selection for achieving proportional fairness in multi-rate 802.11e WLANs: test-bed implementation and evaluation

    Proceedings of the First International Workshop on Wireless Network Testbeds, Experimental Evaluation and Characterization

    (2006)
  • P. Serrano et al.

    Optimal configuration of 802.11e EDCA for real-time and data traffic

    IEEE Trans. Veh. Technol.

    (2010)
  • B.A.H.S. Abeysekera et al.

    Dynamic contention window control scheme in IEEE 802.11e wireless LANs

    Proceedings of the IEEE VTC

    (2009)
  • R. Cheng et al.

    A new scheme to achieve weighted fairness for WLAN supporting multimedia services

    IEEE Trans. Wirel. Commun.

    (2006)
  • Cited by (6)

    View full text