In this project, I am interested in designing TCAM conscious algorithms for answering frequent element and top-k queries over data streams. TCAM or Ternary Content Addressable Memory is characterized by constant time look-up operations in hardware. The current research is focussed on designing an efficient algorithm that can efficiently leverage this look-up property of the TCAMs.

Hide Abstract

The increasing popularity of social networks has initiated a fertile research area in information extraction and data mining. Although such analysis can facilitate better understanding of sociological, behavioral, and other interesting phenomena, there is growing concern about personal privacy being breached, thereby requiring effective anonymization techniques. If we consider the social graph to be a weighted graph, then the problem of anonymization can be of various types: node identity anonymization, structural anonymization, or edge weight anonymization. In this paper, we consider edge weight anonymization. Our approach builds a linear programming (LP) model which preserves properties of the graph that are expressible as linear functions of the edge weights. Such properties form the foundations of many important graph-theoretic algorithms such as single source shortest paths tree, all-pairs shortest paths, k-nearest neighbors, minimum cost spanning tree, etc. Off-the-shelf LP solvers can then be used to find solutions to the resulting model where the computed solution forms the weights of the anonymized graph. As a proof of concept, we choose the shortest paths problem and its extensions, prove the correctness of the constructed models, analyze their complexity, and experimentally evaluate the proposed techniques using real social network data sets. Our experiments demonstrate that not only does the proposed technique anonymize the weights, but it also improves the k-anonymity of the graphs while scrambling the relative ordering of the edge-weights, thereby providing robust and effective anonymization of the sensitive edge-weights.

Hide Abstract

This project involves designing a tool that would help monitor a wireless network in real time. This is an important problem as with the increasing popularity of the wireless networks and their large scale deployment, managing such networks has become a challenging task. To be particular, if there are problems in operation of these networks, then diagnosing them and troubleshooting them manually has become a big problem. Most of the existing methods involve off-line analysis of the network performance, but that is of little help when it comes to dealing with the problems in real-time. Real-time analysis is required to ensure reliable operation of the network and guarantee user satisfaction.
We have designed a tool, that would help the users/ network adminitrators to analyse the network in real time. This tool consists of a sniffer, that passively monitors the network, and a GUI that displays the various metrics of the network as observed by the sniffer. Details of the project can be found in the technical report.

Hide Abstract

Wireless Mesh Networks are an emerging field of research in the recent years. But very little has been done for providing QoS enabled routing in Wireless Mesh Networks. This work consists of designing a routing protocol for Wireless Mesh Networks that provides "strong" QoS guarantees in terms of Bandwidth and Delay. This protocol determines the routes on demand and selects a route that can strictly comply with the bandwidth and delay requirements. The protocol takes the overall robustness of the routes when selecting a route amongst multiple candidate routes.

Hide Abstract

Various Ad-hoc routing protocol implementations suffer form certain implementation problems like Communication Gray Zone, Fluctuating Neighbors etc. We hope to provide a pragmatic approach towards improving the performance of the routing protocols in this respect. A new module has been implemented that keeps track of the stability of the neighbors to make route decisions and is also capable of detecting the presence of new nodes in its neighborhood, even if they are not participating in any data transfer, and explore possible routes through it. We are also exploring the possible applications of this new module in Multipath routing. In addition to obtaining the stability information, it gathers other neighbor statistics to help the routing entity choose the best route from a set of available routes.

Hide Abstract

In this project we are working on improving the throughput / goodput of the TCP sources. We are exploring a change in the congestion avoidance algorithm of the traditional TCP-Reno flavor by using a constant congestion window to compensate for the packets lost in wireless losses, as well as congestion losses. The entire experimental work is mainly simulation based; all simulations are performed using the network simulator ns-2 (VINT Project of UCB, LBL & Xerox PARC). We are also exploring the impact of this modified congestion control algorithm on other aspects of TCP performance such as Fairness.

Hide Abstract

In this project, as a group of 2, we had set up a Multi- Hop Ad-Hoc network. As of now the network connects the two sites of the Centre for Mobile Communication & Computing (CMCC) Lab within the campus, thereby creating a Wireless corridor. Gradually, it would be expanded outside the campus, with many added functionalities. We had also developed a GUI based Client-Server application that would help monitor the status and performance of the other nodes currently present in the network.

Hide Abstract

In this project, as a group of 2, we deployed a Wireless Distribution System (WDS), in which an isolated Access Point (AP) "not connected to the infrastructural network" connects to another AP which is connected to the Wired Network to provide services to its users. Hence a Distribution System over wireless is set up and the remotely placed AP can deliver services, even without any link to the wired network.In IEEE 802.11 terminology a "Distribution System" is system that interconnects so-called Basic Service Sets (BSS). A BSS is best compared to a "cell", driven by a single Access Point (one of those circles in the diagram below). So a "Distribution System" connects cells in order to build a premise wide network which allows users of mobile equipment to roam and stay connected to the available network resources. A distribution system can Wired (typically Ethernet), or Wireless (using the radio device inside the Access Point). One important aspect of WDS (this in contrast to other existing wireless AP to AP connection schemes used in for instance outdoor installations) is the fact that a single PC card in the Access Point can assume multiple roles at the same time. It can "drive" a cell (as in wired connected APs), and as such connects wireless clients to the infrastructure, and it can maintain up to six different wireless connections to other Access Points. For that to be possible the operational (frequency) channel will need to be the same for the cell that is controlled by the AP and for the wireless links to the other APs.

Hide Abstract

This work was carried out during the summer training at IGSI, Kolkata in which I had performed some research & developmental work towards integrating SIP phones (based on Session Initiation Protocol) with SAP enterprise portal. Our goal was to integrate a SIP phone with the other collaboration features of SAP Enterprise portals. This would equip the SAP portals to make or receive phone calls. The main objective of the project was to design a portal component that can be integrated with the SAP Collaboration Features in order to make a call from a computer to a mobile or basic landline phone without using PSTN. The goal is to use the Session Initiation Protocol (SIP) for this purpose. I had tried to incorporate a SIP phone as one of the SAP Collaboration Room features. SAP is one of the most popular ERPs - short for enterprise resource planning, a business management system that integrates all facets of the business.

Hide Abstract