Current Projects

  1. TVWS trials -- jointly with IIT Hyderabad (funded by DeitY)
  2. Structured sharing of resources in a Device-to-Device (D2D) Network (funded by Intel USA)

Completed Projects

  1. SPARC: Spectrum-Aware rural connectivity - Use of sub-GHz bands for rural communication in India using Cognitive Radio (funded by DeitY, GoI)
  2. 4G-LTE pico cell performance evaluation (funded by Cisco Pvt. Ltd.)
  3. Measuring performance of 2G/3G networks (funded by Ford Foundation and DST, GoI)
  4. Next Generation WiFi and Internet of Things (IoT) (funded by Cambridge Silicon Radio India Pvt. Ltd.)
  5. Middleware for Disaster Management (Indo-UK project funded by DST)
  6. WiMAX Mesh
  7. Personal Navigation System
  8. Robust data collection in sensor networks
  9. OLPC WiFi mesh networks
  10. Scalable Ad Hoc Networks
  11. Network Probing
  12. Traffic Modeling and Analysis
  13. Multiscale Sampling Theory
  14. Multiscale Queuing Theory
  15. Long-Term Document Security


Design and Development of a Rapidly Deployable WiMAX Mesh Network (funded by Ministry of Communications and IT, GoI)

The objective of this project is to design, build, and test a wireless mesh network that provides high performance in terms of throughput and latency to enable triple-play applications. The proposed network shoould be rapidly deployable, provides increased coverage and performance, allows nomadic mobility of mesh nodes, has lower infrastructure and operational costs, and is robust to node failure. It is our belief that such a network is best deployed as a time-division multiple access (TDMA) wireless mesh based on WiMAX offering the required range and traffic prioritization. Such a network would find application in military, public safety, and in emergency situations requiring fast network deployment.
Paper: WISARD, 2010


Indoor Global Positioning System (funded by Marvell India Pvt. Ltd.)

The goal of this project is to build a self-contained indoor navigation system that can accurately locate the position of a user within buildings. Applications for tracking vehicular and individual movement are typically based on the Global Positioning System (GPS). GPS makes use of satellite signals for positioning and does not perform well indoors because these signals get significantly attenuated. Hence the need for a novel solution.Our navigation system will consist of several sensors that measure various physical properties such as acceleration, orientation, height etc., which together provide an estimate of the change in position of the device in three dimensions over time. This change in position when combined with the last known accurate GPS-based position of the object then gives an estimate of current position.


Scalable Mobile Ad Hoc Networks

A Mobile Ad Hoc Network is a self-configuring network of mobile nodes connected by wireless links. Ad Hoc networks require little infrastructure which makes them the choice solution for communication during disaster relief efforts, in underdeveloped parts of the world with little infrastructure, and on battlefields. Several of the proposed routing protocols for ad hoc networks perform well only with small networks of a few hundred nodes. The SAFARI project at Rice University aims to design a suite of ad hoc networking protocols that enable large networks that consist of tens of thousands of nodes.
Paper: WiOpt, 2006


Optimal Sampling Strategies for Multiscale Processes

We design strategies to optimally sample multiscale stochastic processes in order to estimate their global average optimally. This has implications for Internet measurement, sensor network design, environmental monitoring, etc. A multiscale process consists of a set of univariate random variables that are organized like the nodes of a tree. Nodes at the bottom are called leaves and the topmost node is the root. We associate each node with the average of a physical process over some region with nodes at higher scales corresponding to larger regions. The root thus represents the global average of the process and the leaves represent local samples. The question we address is: Among all possible sets of leaves of size n, which set gives the best linear estimate of the root in terms of minimum mean squared error?
Paper: Institute of Mathematical Statistics Lecture Notes - Monograph Series, 2006


pathChirp: Efficient Available Bandwidth Estimation for Network Paths

Knowledge of the available or unused bandwidth on a network path can be crucial for optimizing network performance. pathChirp is a new active probing tool for estimating the available bandwidth on a communication network path. Based on the concept of self-induced congestion, pathChirp features an exponential flight pattern of probes we call a chirp. Packet chirps offer several significant advantages over current probing schemes based on packet pairs or packet trains. By rapidly increasing the probing rate within each chirp, pathChirp obtains a rich set of information from which to dynamically estimate the available bandwidth.
Paper: Passive and Active Measurement Workshop, 2003


STAB: Locating available bandwidth bottlenecks
The Spatio-Temporal Available Bandwidth estimator (STAB) is a new probing tool that locates thin links - links with less available bandwidth than those preceding them on a path. By localizing thin links, STAB facilitates network operations and troubleshooting, provides insight into what causes network congestion, and aids network-aware applications such as grid computing. The tool combines the concept of "self-induced congestion", the probing technique of "packet tailgating", and special probing trains called "chirps" to efficiently locate the thin links.
Paper: IEEE Internet Computing, 2004

mwm tree  

A Multifractal Wavelet Model with Application to TCP Network Traffic

Traffic models that capture important features of observed Internet traffic provide synthetic data for simulation purposes as well as give key insights into the causes of the dynamics in traffic. We develop a new multiscale modeling framework for characterizing positive-valued data with long-range-dependent correlations (1/f noise) such as network traffic. Using the Haar wavelet transform and a special multiplicative structure on the wavelet and scaling coefficients to ensure positive results, the model provides a rapid O(N) cascade algorithm for synthesizing N-point data sets. We derive a scheme for matching the model to real data observations and, to demonstrate its effectiveness, apply the model to network traffic synthesis.
Paper: IEEE Transactions on Information Theory, 1999

multiscale analysis  

Small-Time Scaling Behavior of Internet Backbone Traffic

The intriguing discovery of "self-similarity" in network traffic by Leland et al sparked considerable research to determine the causes of this behavior. In this work we concentrate on the analysis of backbone traffic at sub-second time scales. We observe that for a majority of the traffic traces, the (second-order) scaling exponents at small time scales (1ms - 100ms) are fairly close to 0.5, indicating that traffic fluctuations at these time scales are (nearly) uncorrelated. Some traces, however, do exhibit moderately large scaling exponents (approximately 0.7) at small time scales. To identify the network causes of the observed scaling behavior, we analyze the flow composition of the traffic along two dimensions -- flow size and flow density. Our study points to the dense flows (i.e., flows with bursts of densely clustered packets) as the correlation-causing factor in small time scales.
Papers: Computer Networks'05, MASCOTS'09, IFIP Networking'10

queue only 1  

Multiscale Queuing Analysis of Long-Range-Dependent Network Traffic

Many studies have indicated the importance of capturing scaling properties when modeling traffic loads; however, the influence of long-range dependence (LRD) and marginal statistics still remains on unsure footing. We study these two issues using a novel multiscale approach to queuing analysis. The queuing analysis provides a simple closed form approximation to the tail queue probability, valid for any traffic model and any given buffer size. Our results clearly indicate that the marginal distribution of traffic at different time-resolutions affects queuing and that a Gaussian assumption can lead to over-optimistic predictions of tail queue probability even when taking LRD into account.
Paper: IEEE/ACM Transactions on Networking, 2006.


Long-term document security (funded by Ministry of Communication and IT)

Documents such as birth certificates and professional degrees must last a lifetime. We would like be these to be stored and made available electronically. Clearly, a one time encryption with digital keys (with a short lifetime of few years) is not going to keep the documents safe. In this project we designed renewable encryption schemes to keep the documents secure. In addition, we developed a prototype professional degree server which can make such certificates available to employers.