Cross-Layer Techniques for Efficient Medium Access in Wi-Fi Networks - PhDData

Access database of worldwide thesis




Cross-Layer Techniques for Efficient Medium Access in Wi-Fi Networks

The thesis was published by Zhushi, Astrit, in November 2022, UCL (University College London).

Abstract:

IEEE 802.11 (Wi-Fi) wireless networks share the wireless medium using a
Carrier Sense Multiple Access (CSMA) Medium Access Control (MAC) protocol.
The MAC protocol is a central determiner of Wi-Fi networks’ efficiency–the
fraction of the capacity available in the physical layer that Wi-Fi-equipped
hosts can use in practice. The MAC protocol’s design is intended to allow
senders to share the wireless medium fairly while still allowing high utilisation.
This thesis develops techniques that allow Wi-Fi senders to send more data
using fewer medium acquisitions, reducing the overhead of idle periods, and
thus improving end-to-end goodput. Our techniques address the problems we
identify with Wi-Fi’s status quo. Today’s commodity Linux Wi-Fi/IP software
stack and Wi-Fi cards waste medium acquisitions as they fail to queue enough
packets that would allow for effective sending of multiple frames per wireless
medium acquisition. In addition, for bi-directional protocols such as TCP,
TCP data and TCP ACKs contend for the wireless channel, wasting medium
acquisitions (and thus capacity). Finally, the probing mechanism used for
bit-rate adaptation in Wi-Fi networks increases channel acquisition overhead.
We describe the design and implementation of Aggregate Aware Queueing
(AAQ), a fair queueing discipline, that coordinates scheduling of frame transmission
with the aggregation layer in the Wi-Fi stack, allowing more frames per
channel acquisition. Furthermore, we describe Hierarchical Acknowledgments
(HACK) and Transmission Control Protocol Acknowledgment Optimisation
(TAO), techniques that reduce channel acquisitions for TCP flows, further
improving goodput. Finally, we design and implement Aggregate Aware Rate Control (AARC), a bit-rate adaptation algorithm that reduces channel acquisition
overheads incurred by the probing mechanism common in today’s
commodity Wi-Fi systems. We implement our techniques on real Wi-Fi hardware
to demonstrate their practicality, and measure their performance on real
testbeds, using off-the-shelf commodity Wi-Fi hardware where possible, and
software-defined radio hardware for those techniques that require modification
of the Wi-Fi implementation unachievable on commodity hardware. The techniques
described in this thesis offer up to 2x aggregate goodput improvement
compared to the stock Linux Wi-Fi stack.



Read the last PhD tips