Rate Adaptation and Resource Allocation for Wireless Networks

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering


PhD Thesis Defence

Title: "Rate Adaptation and Resource Allocation for Wireless Networks"

By

Miss Qiuyan Xia


Abstract

IEEE 802.11-based wireless networks have become increasingly popular due
to the recent availability of affordable devices providing multi-rate
capabilities. In this thesis, we study rate adaptation and resource
allocation methodologies for 802.11-based wireless networks. Rate
adaptation is a key functionality of 802.11 devices allowing them to cope
with time-varying channel conditions and to maximize the throughput of
point-to-point wireless links. We propose a novel rate adaptation
algorithm, "Smart Sender" for 802.11 Wireless Local Area Networks (WLANs),
which utilizes both statistics and signal strength measurements to select
the transmission rate that maximizes the link throughput. We also
implement and evaluate the algorithm in commercial WLAN devices, which
proves that Smart Sender responsive, has low overheads, and is robust to
collision errors in various wireless environments.

We further study rate adaptation for Vehicular Ad hoc Networks (VANETs)
with an 802.11 physical layer. Different from traditional methods
employing periodic channel quality estimations, we propose to use a
learning algorithm MTRA (Model Tree based Rate Adaptation) which trains a
packet error rate model for different data rates. In real time, MTRA takes
multiple inputs from the environment and outputs a rate decision
satisfying specific packet error rate requirements. This approach can
effectively utilize the available environmental information (distance,
SNR, speed, and so on) for timely rate adaptation, hence avoiding the
estimation delay experienced by the legacy methods. By locating the
optimal rate for current channel conditions as soon as possible, MTRA
offers superior throughput performance compared to other popular methods.

While rate adaptation exploits the multi-rate capability of an individual
user, it also results in the well-known phenomenon of "performance
anomaly" among multiple rate-different users sharing the WLAN. Therefore,
efficient resource allocation among multiple users is critical to the
system performance. For downlink resource allocation, we develop a
weighted fair scheduling based on adaptive rate control (WFS-ARC)
framework for throughput optimization while satisfying temporal fairness.
For resource allocation with TCP flows in a WLAN, we design a dual queue
management (DQM) scheme for TCP congestion avoidance and uplink/downlink
fairness provisioning. DQM also exploits opportunistic scheduling for high
link efficiency.

For future work, we will study resource allocation methodologies for
wireless networks such as IEEE 802.11n WLANs and IEEE 802.16 mesh
networks. We believe the approaches discussed in this thesis could
contribute to future research on these next generation wireless networks,
by providing a versatile and flexible structure for various systems.


Date:			Thursday, 3 July 2008

Time:			10:00a.m.-12:00noon

Venue:			Room 3501
			Lifts 25-26

Chairman:		Prof. Philip Mok (ECE)

Committee Members:	Prof. Mounir Hamdi (Supervisor)
			Prof. Brahim Bensaou
			Prof. Qian Zhang
			Prof. Ross Murch (ECE)
			Prof. Hussein Mouftah (Inf. Tech. & Engg.,
						Univ. of Ottawa)



**** ALL are Welcome ****