E-mail: sandipc [AT] cse [DOT] iitkgp [DOT] ernet [DOT] in


Current Research Works


High Throughput Wireless Backbone Networks



Wireless data access networks gradually migrating towards high data rate communications and Gigabit wireless technologies. IEEE has standardized the high throughput wireless communication through amendments over the well-established and commercially successful IEEE 802.11 standard, like IEEE 802.11n, IEEE 802.11ac, IEEE 802.11ad etc. The standards for high throughput wireless communications present a number of new technologies at the PHY and MAC layer - MIMO, multi-user MIMO (MU-MIMO), Channel bonding, Frame aggregation etc, for improving physical data rates. However, all these technologies have their pros and cons and requires thorough investigation and analysis for their practical deployments. If commercially successful, the high throughput wireless data access technologies can be used to design wireless back-haul networks through mesh networking technologies, that can replace the traditional wireless distribution system (WDS) through an effective means.

In this research work, we are investigating following research aspects over high throughput wireless technologies -


Navigation System for Developing Countries


UrbanEye: A Public Bus Navigation System - http://www.cnergres.iitkgp.ac.in/urbaneye/
CrowdMap: Crowdsource based City Traffic Map Generation - http://www.cnergres.iitkgp.ac.in/crowdmap/

Public transport in suburban cities (covers 80% of the urban landscape) of developing regions suffer from lack of information in Google Transit, unpredictable travel times, chaotic schedules, absence of information board inside the vehicle. Consequently, passengers suffer from lack of information about the exact location where the bus is at present as well as the estimated time to be taken to reach the desired destination. We find that off-the-shelf deployment of exisitng (non-GPS) localization schemes exhibit high error due to sparsity of stable and structured outdoor landmarks (anchor points).

Through rigorous experiments conducted over a month however, we realize that there are a certain class of volatile landmarks which may be useful in developing efficient localization scheme. Consequently, in this paper, we design a novel generalized energy-efficient outdoor localization scheme which efficiently combines the volatile and non-volatile landmarks using a specialized data structure, the probabilistic timed automata. Our system uses speed-breakers, turns and stops as landmarks and estimate the travel time with a mean accuracy of 3 mins and produces a mean localization accuracy of 50 meters. Results from several runs taken in two cities reveal that the system consumes significantly less energy as compared to several other localization techniques, including GPS.

After providing a successful navigation system, we target to tackle the "gaming behaviour" of the bus drivers. As there lies a huge competition between different bus transport corporations, the drivers adapt different strategies, such as picking up more passengers by getting ahead of other buses, intentionally stopping on signals, and filling up the time quota, which leads to rash driving. This leads to a very uncomfortable journey and may lead to injuries specially to senior citizens.

We Provide an alert system, against the gaming behaviour of such buses. The application will alert the user about the speed breakers and turns along with their impact intensity (gentle, medium or high) and bus stops with the likeliness of bus to stop at that bus stop. Also, the application will alert about the driver's intention such as the driver is trying to pick up more passengers, or the driver is trying to finish his time quota etc.

UrbanEye Demo: Check here

We are also developing CrowdMap, an intelligent data logging module for smart-phones and a server side processing mechanism to extract road and bus route information, and to annotate them over the city map. CrowdMap seamlessly discovers the bus routes, and embeds the annotated route information on the city map.


Scalable Identification of Mobile Video Applications by Constructing App-specific Traffic Signatures



Mobile applications, especially video applications, contribute a significant proportion of Internet traffic. Video apps comprise of Over-the-Top video apps, like YouTube, NetFlix, HotStar, live streaming apps, like Times Now, Star Sports, and video messaging apps, like Skype, Google Hangouts. Ability to distinguish these video apps based on traffic generated from these apps finds many use cases, starting from enterprise policy enforcement in BYOD settings, to resource provisioning by network operators who only has visibility to traffic flows. Since app traffic appears as any other typical HTTP traffic, therefore, port-based identification, or even deep packet inspection techniques, are limited in accurately identifying the app source for the traffic. However, streaming and interactive video often have distinguishing traffic characteristics that may help in classifying the app source. In this work, we analyze the traffic patterns from traces collected from video apps, and utilize the insights to construct classifiers that can identify video app type in a mixed traffic flow. We demonstrate our solution in a realistic setting where we can classify video app sources with high accuracy.


Mobile Access Networks - Traffic Shaping and Optimization



The growth of mobile broadband network has witnessed a rapid pace in the recent years. According to the Cisco VNI report, mobile Internet traffic will increase 66% in between the years 2012-2017. At the same time, the traffic pattern for mobile Internet is going through a rapid change from data based communications to multimedia based communications. According to the Cisco report, mobile network speeds will increase by a order of seven in the year 2017, while two-thirds of the world's mobile traffic will be video. The application characteristics for video traffic significantly differs from the data traffic characteristics. Further the application traffic demand is gradually shifting from down-link centric to up-link centric, where new applications and utilities are coming up with high bandwidth requirements for up-link traffics. Therefore the next generation mobile broadband system designs should go through a hard multi-objective optimization criteria that has the ability to improve the user experience (QoS/QoE) while simultaneously redusing energy consumption and carbon footprint. Therefore we require traffic shaping and improved application development that can exploit the underlying network protocols to save energy consumption without affecting user experience.This project aims at designing effctive application traffic management and optimization mechanisms for next generation mobile access networks for smarter access and Qos/QoE supports.


Cognitive Netwoks and Green Community Wireless Backbone



As radio spectrum usage paradigm moving from the traditional command-and-control allocation scheme to the open spectrum allocation and access scheme, wireless networks meet new opportunities and challenges. Accordingly, the concept of cognitive wireless mesh networking (CogMesh) has been introduced in the research community, and researchers have started to address the unique problems in such a dynamically networking environment. Basically, CogMesh is a self-organized distributed network architecture combining cognitive wireless access technologies (e.g. Cognitive Radio) with the mesh (ad-hoc) structure in order to provide an integrated and converged service platform over a wide range of heterogeneous networks. CogMesh is based on opportunistic spectrum access (OSA) and featured by self-organization, self-configuration, self-protecting and self-healing. Within the CogMesh framework, in order to achieve the OSA goal, we would like to explore several new directions of future prototyping for CogMesh framework.

On the other hand, the recent developments in `Software Defined Radio' (SDR) have opened several new research directions in Green Computing and Green Networking. Reduction in transmission power may introduce higher channel error rate, but can also reduce interference in the network. Therefore, power can also be considered as a trade-off parameter during rate adaptation. Further, adaptive power control can also be used during the design of the scheduling and the forwarding protocols, as the (interface, next hop) pairs can be selected based on the minimum power consumption and maximum performance gains simultaneously. This way, the joint power control, scheduling, forwarding and rate adaptation mechanisms can be designed to improve further the performance of IEEE 802.11s protocols in a mesh backbone network, that we would like to explore.


DSDN - Distributed Software Defined Networking Platform



Software Defined Networking (SDN) has opened a new Pandora's Box in the computer network research communities. SDN uses a centralized view of the global network through a software managed controller, that uses a network operating system to create a virtual global view of the network. Controller in a SDN separates the control plane from the data plane, and significantly reduces network management complexity, while improves network scalability and performance.

However, SDN was traditionally designed for data center networks, which is inherently centralized. Making the concept of SDN operable in a distributed environment poses several research challenges. Nevertheless, the concept of SDN is equally important for dynamic and mobile wireless networks that require several management level reserach and design challenges - like auto-configurability, seamless handover, network convergence etc. SDN con solve many of these problems by separating the control operations (control plane) from data communication (data plane). In this work, we target to design a distributed network controller, and integrated several network management functionalities over it.