Sentinel based Architecture for Wireless Network
Management (Final Year Project)
In this project, we are trying to build an architecture where
both Clients of a Wireless Network and dedicated Stationary Sensors collaborate to monitor the Wireless Network. Our primary focus is to improve the
efficieny of 802.11 hand-offs. Using this architecture we also handle various other problems such as.
� Finding Hand-off Regions and Improving the 802.11 Hand-offs
� Location Determination and Mobility Prediction
� Fault Detection and Diagnosis
We are also using this architecture
for network fault diagnoses. The
HIPC paper
discusses on the Architecture and hand-offs decisions based on some basic
parameters like Neighbor graphs, nearest Hand-off points and AP Load.
Pidgin: Offline Mail Client (in C#.NET)
I implemented an offline mail client with an interactive User Interface. This mail client synchronizes the offline store with the mail server whenever there is a change. SMTP mail servers were used. The modules implemented were
� Message Store Provider and various search features.
� Support for Address Book
� Background Synchronizer which synchronizes the Personal store with the mail server
Image Analysis and Strategy making for Robotic Football (in
C#.NET)
This project analyses a robotic football field using the image taken from the top of the field and forms an efficient strategy for the robots motion using graph theory concepts. The basic modules were
� Image processing to approximate the robot and ball positions and find the situation of the game (attacking or defending).
� Forming various frames of complete graph with robot and goal and try to find the best frame using shortest path, cut-set and other heuristics.
� Graphical Interface which shows run-time analysis.
Moment based Tamil Character Recognition (in C#.NET)
This project implements a
moment based nearest
neighbour classification technique for Tamil Character Recognition. I used Hu's
moments and BW ratio as features. It works effectively with printed Tamil
characters. There is training phase to train any character set before we start
recognising that character set.The basic modules were
� Image Pre-processing which includes image thresholding, noise removal.
� Segmentation of line and characters using horizontal and vertical projectors.
� Character thinning and size normalisation.
� Feature Extraction using Hu�s invariant moments.
� Nearest neighbour classification.
Other Projects
-
Lexical Analyser, Parser and
Intermediate Code Generator for C like Language (in C)
-
Dynamic Brach Prediction using Tournament Predictors(in SS SMT simulator)
-
Assembler and Macro processor for SIC (XE) (in C)
-
Implementation and Analysis of various AI search techniques to solve the 8-puzzle game
-
Accounting and Registration software of the Computer Society of Anna University (in VB)
-
An Electronic timer (Electronics project)
