Chapter 1.
Introduction
Wireless networks such as IEEE 802.11 WLANs [1] have recently been deployed widely with rapidly increasing users all over the world. As real-time applications such as VoIP and streaming video are getting more common in daily life, quality of service (QoS) guarantee over wireless networks is becoming an important issue. Generally speaking, QoS includes guarantee of maximum packet delay, delay jitter, and packet loss probability. To cope with this problem, a new enhancement of WLANs, called IEEE 802.11e [2], is introduced to support the QoS requirements of real-time traffic.
Fig.1 shows the example of IEEE 802.11e MAC architecture. The MAC protocol proposes a QoS-aware coordination function which is called Hybrid Coordination Function (HCF). This function consists of two channel access mechanisms. One is contention-based Enhanced Distributed Channel Access (EDCA) for prioritized QoS and the other is contention-free HCF Controlled Channel Access (HCCA) for
Chapter 1. Introduction
parameterized QoS. Because of the contention-free nature, HCCA can provide much better QoS guarantee than EDCA [3].
Beacon Interval
Service Interval Service Interval Service Interval CP
CFP CFP CP CFP CP
CFP: Contention Free Period (HCCA) CP: Contention Period (EDCA)
Fig.1. Example of 802.11e MAC architecture
HCCA requires a centralized QoS-aware coordinator, called Hybrid Coordinator (HC), which is commonly located in Access Point (AP). An AP with the HC function is called a QoS-aware AP (QAP). QAP has a higher priority than normal QoS-aware stations (QSTAs) in gaining channel control. QAP can gain control of the channel after sensing the medium idle for a PCF interframe space (PIFS) that is shorter than DCF interframe space (DIFS) adopted by QSTAs. After gaining channel control, QAP polls QSTAs according to its polling list. In order to be included in QAP’s polling list, a QSTA needs to make resource reservation for each traffic stream (TS) attached to it that requires QoS guarantee.
Resource reservation is accomplished by sending the Add Traffic Stream
Chapter 1. Introduction
(ADDTS) frame to QAP. In this frame, QSTA can give traffic characteristics a detailed description in the Traffic Specification (TSPEC) field. Based on the traffic characteristics specified in TSPEC and the QoS requirements, QAP calculates the scheduled service interval (SI) and transmission opportunity (TXOP) duration for each admitted TS.
Upon receiving a poll, the polled QSTA either responds with QoS-Data if it has packets to send or a QoS-Null frame otherwise. When the TXOP duration of some QSTA ends, QAP gains control of channel again and either sends a QoS-Poll to the next station on its polling list or releases the medium if there is no more QSTA to be polled.
The TXOP calculation of the sample scheduler provided in IEEE 802.11e standard document is based on mean data rate and nominal MSDU size. It performs well for constant bit rate (CBR) traffic. For variable bit rate (VBR) traffic, packet delay and loss may vary significantly for different TSs. Several schemes have been proposed recently to improve QoS guarantee while maintaining high bandwidth utilization [4]-[13]. As an example, the equal-spacing-based design, a variation of the famous rate monotonic scheduler, was proposed in [12]. In this design, there is no
Chapter 1. Introduction
need to have a common SI. Assume that there are n TSs and TS i is to be served periodically with period
T
i. It was shown that all TSs can be served with equal-spacing if and only if 1)T
i+1= k T
i i wherek
i is some integer larger than or equal to one and 2)1
equal-spacing-based design is a generalization of the sample scheduler and is only suitable for CBR traffic. A TXOP allocation scheme was proposed in [9] to handle VBR traffic with different delay bound requirements. An equivalent flow with delay bound of one SI is defined for a flow with delay bound of more than one SI to achieve inter-flow multiplexing gain. To reduce computational complexity, authors of [9] assumed that the arrival process of each real-time VBR traffic flow is Gaussian. This assumption may not be valid for real applications. Another design, called prediction and optimization-based HCCA (PRO-HCCA), which can handle VBR traffic was presented in [10], [11]. It takes delay bounds of different TSs into consideration in TXOP allocation. However, the PRO-HCCA scheduler has high implementation complexity because QAP has to maintain a partition list for each TS. Besides, the fact that every TS is polled individually in all service intervals implies considerable overhead for TSs with large delay bounds.
Chapter 1. Introduction
The purpose of this paper is to present a scalable HCCA scheme with per-QSTA granularity. In the proposed scheme, QAP maintains only one partition list for each QSTA even if it is attached with multiple TSs. The proposed scheme is then modified to reduce polling overhead for QSTAs that are attached with TSs having large delay bounds. For the modified scheme, different QSTAs are allowed to have different polling periods.
Numerical results obtained from computer simulations show that the proposed HCCA scheme and the modified one perform better than the PRO-HCCA scheduler. Since our designs are related to PRO-HCCA, we shall briefly review the scheduler in Chapter 2.
The rest of this thesis is organized as follows. Chapter 2 describes system model. We also review the sample scheduler and the PRO-HCCA scheduler. Chapter 3 contains our proposed scheduler and we modify our design to allow different polling periods for different QSTAs so that polling overhead can be further reduced. Numerical results are presented and discussed in Chapter 4. Finally, we draw conclusion in Chapter 5.
.
Chapter 2. Related Work
Chapter 2.
Related Work
2.1. System model
In the investigated system, transmission over the wireless medium is assumed to be divided into SIs and the duration of each SI, denoted by SI, is a sub-multiple of the length of a beacon interval
T
b. Moreover, a SI is further divided into a contention period and a contention-free period. We consider only uplink traffic because downlink traffic is completely known to QAP and, therefore, can be easily scheduled.We assume in this paper that the QoS requirement is specified with delay bound, which can be carried in the Delay Bound field of the TSPEC information. A packet is dropped if it violates the delay bound. There are N QSTAs, called QSTA1, QSTA2, …, and