Dheeraj Golchha Jain: Master's Thesis, Jul 2007

Title: Wi-Fi NetMon: Performance Observation, Anomaly Detection and Diagnosis in Long Distance Wi-Fi Networks

Abstract:

The last few years have seen lot of work for providing network connectivity in remote rural areas. An alternative of wired network has been to use off the shelf 802.11 equipment with high gain antennas to provide network connectivity to rural areas. The attractiveness of this setting comes from cost effectiveness and low power consumption. Although there are many such network deployments, a very little has been done for monitoring and management of these networks. An initiative has been taken at IIT Kanpur for monitoring and management of long distance Wi-Fi networks called Wi-Fi Netmon.

In our work, we focus on performance observation, anomaly detection and diagnosis of long distance Wi-Fi networks from central location. We present a centralized, client - server based architecture for monitoring of long distance Wi-Fi networks. We provide a set of experiments such as packet error rate, UDP throughput, TCP throughput, get configuration and set configuration to observe performance and to retrieve or configure wireless parameter values of links on these networks from central location. We use these experiments to observe performance of links in long distance Wi-Fi networks and detect and diagnose problems in case of poor performance of links.

We focus on detecting and diagnosing problem known to be commonly occurring on these networks. The problems that we consider are power reset, link misalignment, insufficient transmit power, presence of hardware quirk, interference detection, packet dropping at receiver and MAC level ACK timeout. We design and implement a debugging algorithm for observing performance and detecting and diagnosing the problems stated here.

We also evaluate our work by creating a wireless link in the lab. We introduce the problems in the link and check whether we are able to detect and diagnose these problems through the debugging algorithm. We successfully detect and diagnose the problems power reset, link misalignment, insufficient transmit power and interference detection.


Bhaskaran Raman
Last modified: Wed Feb 13 17:50:19 IST 2008