2004


TCENet Meeting, Anbarasan(2KC05).R, Kalicharan.S(2KC38), Lalitha Devi.k(2KC48)

TCENet Meeting is intranet communication software, which includes three main utilities. Thay are File Transfer: This mainly concerns with transferring of files between two registered users. Instant Message Transfer: This facilitates the users to log in, locate their friends in the friends list, invoke a message box when they are online and finally send messages using hat message box. Peer-to-Peer Video Transfer: This requires one of the parties to have a video capturing equipment like a web camera or a digital video camera. Here video is transferred as stream. But the pre-requisite for this transfer is the higher band-width.

Design of a Micro-kernel suitable to drive a Distributed Operating System Amalan Joe Steeve I (03CS01)

"A Microkernal based Operating System Design" has always been an object of interest to Operating System Designers. Though the concept of Microkernals has been around for more than three decades, only lately actual implementations are being realized. This thesis brings out a Microkernal design which must be simple to implement and robust in operation. Though the design conditions have been biased for implementing a Distributed Operating system, the desing is flexible enough to implement a Simple stand alone OS too. The thesis also brings out the caveats in other microkernals and analyses how the propoased system should perform better. The porposed system is titled ProX kernal.

Tuxedo Chess Engine Ananth Shrinivas .S (2KC04)

Game playing is one of the classic problems of Artificial Intelligence(AI). Chess in particular is the popular domain for practicing AI techniques. The large branching factir of chess makes an exhaustive game-tree search based on the minmax strategy infeasible. Hence an approximate, limited depth search with many enhancements to the minmax algorithm is commonly used. With a limited depth, the quality of game playing engine increases with teh search depth. Recently parallel search of game-trees has been widely researched to fine ways of increasing search depth. A number of algorithms for parallel search of a game tree have been developed an analysed. This research has resulted in Deepblue - A VLSI hardware assisted, Parallel Chess Engine beating the Human World Chess Champion Garry Kasparov. Meanwhile, the field of super computing has taken a new dimension in the form of clusters. A cluster is a network of workstaions able to deliver performance comparable to super computers, espicially for MIMD based parallel processing applications. Clusters are becoming increasingly popular for high performance scientific and engineering computing. Beowulf Linux Clusters are expicially preferred for thier wase of installation and maintenence coupled with high performance at a low cost. Tuxedo is a Paralle Chess Engine that is designed for a Beowulf Cluster. It uses AI techinique while searching the game tree in parallel. It performs significantly better than it's non parallel implementataion, in terms of search time and depth. Consequently game play quality of the parallel engine is better as proved by the ELO estimates. Thr parallel algorithms designed and used for Tuxedo is based in exixting sequential and parallel game-tree search algorithms and by exhibits good scalability and robustness. Tuxedo is implemented in C and C++ and is portable to POSIX compatible platforms. It is distributed under the terms if GNU GPL