WiFi networks have become ubiquitous. However, due to the nature of the
radio-wave medium, the performance of 802.11 is unpredictable and highly
dependent on the environment. This problem is fundamental to 802.11's
decentralized, signal-based airspace arbitration mechanism. When devices have
incomplete and inconsistent channel conditions for an overlapping
interference domain, their signals alone cannot ensure a fair competition for
airspace. As a result, competing flows may suffer from unfair bandwidth
distribution if the shared airspace is congested.
A useful tool to visualize and diagnose problematic wireless networks is the
set of devices interfering with each other at a given time. We say two
devices a and b interfere when one of two possible situations occur. First,
a is able to sense b's radio signals, though not necessarily decode them,
resulting in a unable to send data. Second, a and b aren't in radio range,
but their destination devices are, resulting in packet collisions. We call
such a set of mutually interfering devices the interference neighbourhood.
We present wypy, an online system which merges trace-files and produces a map
of interfering devices contained within the trace. wypy is able to identify
pairs of devices exhibiting either hidden or exposed terminal interference
using a pipeline that consists of trace merging and reconstruction, filtering
of simultaneously sending devices, throughput and delay signal calculations,
and a test for interference correlation. We evaluate wypy using an in-lab
testbed set up in known interference scenarios.
| Identifer | oai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:BVAU./515 |
| Date | 05 1900 |
| Creators | Lotun, Reza M. E. |
| Publisher | University of British Columbia |
| Source Sets | Library and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada |
| Language | English |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Format | 413872 bytes, application/pdf |
Page generated in 0.0029 seconds