Identifying Wi-Fi Interference by End-Users Richard Meng, Xiang

Identifying Wi-Fi Interference by End-Users
Richard Meng, Xiang Ying Qian, Kyung-Hwa Kim, Henning Schulzrinne
Dept. of CS and EE, Columbia University in the City of New York

Motivation & Background

Goal

Identifying causes of WLAN performance degradation is
nontrivial
Most access points (IEEE 802.11b / IEEE 802.11g) are
deployed in the 2.4GHz wireless band, which causes
network interference
Most significant interference sources:

Difficult for end-users to identify the devices that cause
Wi-Fi interference
Existing solution:
Spectrum Analysis
Additional hardware is required (e.g., Wi-Spy)

Channel
Contention

Neighboring
Channels

Architecture

Monitors RF activity within a given frequency range
(Wi-Spy by Metageek, $84, metageek.net)

Our goal is to identify the source of Wi-Fi interference
without any hardware support.
Monitor and analyze the patterns of various
parameters related to Wi-Fi performance.
Obtain data from other collaborative nodes to see
whether others also observe the same interference
Train the pattern analyzer with the dataset obtained
from different environments and nodes
Provide users the best matched devices that are
supposed to cause the interference

Monitoring
802.11 packet
Wireless
Performance

Is there a microwave
oven nearby your
laptop?

Packet Statistics
Information
Sharing

Analyzer

Is there a Bluetooth
device nearby your
wireless router?

Information
Sharing

Analyzer

Monitoring
802.11 packet
Wireless
Performance
Packet Statistics

Non-WiFi interference

Preliminary Measurement

Implementation & Experiment

In our experiment, Bluetooth devices and microwave
ovens showed different patterns (e.g., the magnitude of
standard deviation of retry count, percentage of ACK
failures, i.e. microwave oven tends to cause more
unstable network condition)

Bluetooth Interference (HTC One X)
2.5

Count per frame

2

1.5

1

0.5

0

1

3

5

7

9 1 1 13 15 1 7 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 7 1 73 75 7 7 79 81 8 3 85 87 8 9

Time (seconds)
Failed

Retry

MultipleRetry

RTSSuccess

RTSFailure

ACKFailure

Microwave oven Interference
2.5

This result enables to identify some interference
sources.
However, the patterns are difficult to be resolved by
human.
A Machine learning method is needed to achieve more
accurate identification of interference sources

2

count per frame

Interference measurement
Existing waves makes measuring interference difficult
We measured network throughput, SNR, 802.11 retry
counter, and other variables to infer a characteristic of
current interference.
Implementation
On Linux: Analyze the information of Radiotap 802.11
header in the data link frames captured by integrating
Wireshark, Jpcap, and Alpacka library.
On Windows: Analyze built-in parameters in network
systems collected by Windows Native Wi-Fi API.
Experiment setup
802.11g Cisco Linksys AP (Channel 1,6,7)
Measurement on laptops
Experiment with and without microwave ovens /
Bluetooth devices
Sharing information using DYSWIS[1] framework

Results & Future work

1.5

1

0.5

0

References
1

3

5

7

9 11 13 15 17 19 21 23 25 27 29 31 33 35 3 7 39 41 43 45 47 49 51 53 55 57 59 61 6 3 65 67 69 71 73 75 77 79 81 83 85 87 89

TIME (SECONDS)
Failed

Retry