Throughput analysis of IEEE 802.11e EDCA under heterogeneous traffic

Abstract

The amended version of IEEE 802.11e defines the enhanced distributed channel access (EDCA) for quality of service (QoS) connections. EDCA provides a priority scheme to classify different traffic categories by the arbitration inter-frame space (AIFS), and the initial and maximum contention window sizes. Most previous Markov chain analyses of EDCA have concentrated on the effect of contention window size, they neglected the effect of AIFS. In this paper, we consider of AIFS events among each back-off procedure and evaluate the saturation throughput of the IEEE 802.11e EDCA under heterogeneous traffic scenarios. The analytical model is based on the differentiated AIFS and uses the discrete time slot to analyses the external collision time. We take advantage of extensive simulation studies to validate the accuracy of the proposed model. Our models accurately estimate the effects of varied back-off window size and AIFS under heterogeneous traffic.

Introduction

Driven by the rapid growth of WLAN traffic, the IEEE 802.11 wireless LAN (WLAN) will play an important role in the future fourth-generation wireless communication. At the same time, the number of multimedia applications has increased tremendously, and these multimedia applications require some quality of service (QoS) support such as guaranteed bandwidth, bounded delay and jitter. Since the MAC (Media Access Control) layer is essential for QoS support, the IEEE 802.11e has defined the QoS-enhance IEEE standard. The 802.11e addresses the weakness of 802.11 [1] and proposes a new MAC layer coordination function called HCF (Hybrid Coordination Function). The HCF consists of parameterized QoS HCCA (HCF Control Channel Access) and prioritized QoS EDCA (Enhance Distributed Channel Access) [2]. The parameterized QoS HCCA mechanism is the centralized MAC protocol that each mobile station is polled by an access point (AP) to get transmission opportunities. However, HCCA is more complex and inefficient for normal data transmission. In contrast to the parameterized QoS HCCA mechanism, the priority QoS EDCA mechanism is a distributed MAC protocol in which each mobile station gets transmission opportunities according to a random back-off time. Though EDCA is easy to implement, it has difficulty achieving strict QoS for real-time applications, especially in the saturated case. To overcome these drawbacks, considerable research has been devoted to theoretical analysis of the performance of the 802.11e EDCF in order to develop an efficient means for QoS support in WLANs for a wide variety of applications.

To analyze the performance of the 802.11e, most studies have used theoretical analysis of the 802.11 DCF as the foundation of the performance analysis of the 802.11e EDCF. In the saturation case, Bianchi [3], [4] analyzed the performance of the 802.11 DCF and proposed a Markov chain as an approximately model of the binary exponential back-off procedure. Beyond the saturated throughput derived in Bianchi’s model, Foh and Zuckerman [5] took advantage of a Markov chain state to analyze the throughput and mean packet delay for the dependent single server queue. Calì [6], [7] used a p-persistent variant of DCF to analyze back-off strategies and approximated the capacity similarly as the analysis of Markov chain. In order to calculate the packet-collision probability, Tay and Chua [8] took advantage of an average value mathematical model to solve the maximum throughput in terms of collision probability. A variation of Bianchi’s model was proposed by Wu et al. [9] for the further consideration of retry limits. Clearly, all of previous DCF analyses were devoted to the back-off stage and the transmission retry limit. However, the main characteristics of the 802.11e EDCA are its adjustable back-off contention window (CW) and differential arbitration inter-frame space (AIFS) for multi-class access categories (ACs). To extend DCF analyses, the analytical studies of EDCF can be divided into p-persistent EDCF [12] and multidimensional Markov chains [10], [11], [13]. In the p-persistent analysis, Huei and Devtsilitios [12] used the results of bi-dimensions Markov chain and applied p-persistent CSMA/CD (Carrier Sense Multiple Access with Collision Detection) model to differentiate AIFS for ACs. Based on a multidimensional Markov chain, Xiao’s model [10] used a three dimension of Markov chain to estimate the capacity of 802.11e under the condition that channel busy probabilities of ACs were independent of the number of timeslots. In [11], Kong’s model considered the difference of AIFS in back-off procedures, but neglected AIFS while the back-off procedure reminded one timeslot. To accurately analyze the 802.11e model, Tao and Panwar [13] enlarged the 802.11e Markov chain to four dimensions by ACs. The four dimension of Markov chain can analyze effects of AIFS and the impact of internal collision but the calculation complexity is increased. Tao and Panwar [13] considered that the external collision time is approximate to the transmission time and neglected the impact of the external collision time. However, the external collision time in fact depends on the length of CTS (Clear To Send) frame in the optional RTS/CTS (Request To Send/Clear To Send) and the transmission time.

Based on this previous work, our proposal summarizes Xiao’s model [10] and Kong’s model [11] to build a Markov chain model for the EDCA analysis. The proposed developed model reflects the back-off and access procedures accurately by taking account of the back-off timer freeze that includes the external collision time, the transmission time, AIFS and CW parameters. The remainder of this paper is organized as follows. First, we briefly review DCF and EDCA mechanisms in Section 2. Second, the proposed analytical system model is presented in Section 3. Section 4 provides throughput analysis. Numerical and simulation results are given and discussed in Section 5, and Section 6 concludes the paper.