Co-evolution of Popularity and Content Spread in Mobile Social Networks
Srinivasan Venkatramanan and Anurag Kumar
With proliferation of smart hand-held devices (smartphones and tablets) and an increasing demand for high-bandwidth media content, it is necessary to investigate peer-peer content delivery systems in greater detail. In this work, we consider a population of mobile users carrying their smart devices, thus forming a mobile social network, in which content can be transferred opportunistically at pairwise meetings. We also model the evolution of content's popularity i.e., the number of users who are currently interested in receiving a particular content. Inspired by models from epidemiology, we use fluid limits to characterize this co-evolution, and provide optimal mechanisms for content delivery.
Competition over Timeline in Social Networks
Srinivasan Venkatramanan and Anurag Kumar
Online social networks like Facebook (1billion+ users), Twitter (500million+ users) have amassed a significant fraction of the global Internet audience. This has triggered several brands and organizations to increase their investments in advertising through social networks, instead of other traditional forms of Internet advertising (banner ads, sponsored search, etc.) Most social networks employ a reverse chronological timeline structure to display the content to end users. Due to limited user attention, there is competition for this timeline space between various brands (content creators). In this work, we model this competition by a non-cooperative game among such content creators, and study the characteristics of the resulting Nash equilibrium.
Intelligent Systems for Road Traffic Control
Pramod MJ, Samrat Mukhopadhyay and Anurag Kumar
With the increasing complexity of road traffic across the world as well as in India, the need for Intelligent systems that can control traffic lanes can hardly be exaggerated. In this work, we model traffic lanes with large vehicles like cars and buses as special type of queues and find the delay performances of these queues.We use these models to formulate an optimization problem with objective to minimize the total delay experienced by arriving traffic at a signalized intersection. We characterize the optimal amount of green or "GO" times that will be necessary to provide to each lane. We also study the effect of burstiness of traffic for two-lane scenarios. We find a Stochastic Approximation Algorithm to find and tune these optimal green times online. We also study the effect of motorcycles and other small vehicles which can make a group in the lanes.
QoS aware Survivable Network Design
Abhijit Bhattacharya and Anurag Kumar
The objective of our work is to design QoS and energy aware, survivable multi-hop wireless sensor network by deploying a minimum number of additional relays and sinks (to minimize deployment cost) to communicate data generated at sensor nodes (placed in strategic locations, to monitor various parameters of interest; they can act as relays as well) to the infrastructure nodes/sinks. These relay placement problems are, in general, NP-Hard, and our focus is on designing efficient approximation algorithms that provide reasonably good solutions within a reasonable time.
Impromptu Deployment of Wireless Sensor Network
Arpan Chattopadhyay and Anurag Kumar
There are situations in which a wireless sensor network (WSN) needs to be deployed in an impromptu
or as- you-go fashion. One such situation is in emergencies, e.g, situational awareness networks deployed
by first-responders such as fire-fighters or anti-terrorist squads. As-you-go deploy- ment is also of interest
when deploying networks over large terrains, such as forest trails, particularly when the network is
temporary and needs to be quickly redeployed in a different part of the forest (e.g., to monitor a moving
phenomenon such as groups of wildlife), or when the deployment needs to be stealthy (e.g., to monitor
fugitives). Motivated by these more general problems, we consider the problem of as-you- go deployment
of relay nodes along a line, between a sink node and a source node, where the deployment agent starts
from the sink node, places relay nodes along the line, and places the source node where the line ends.
Our basic setting is as follows. A person walks along a line (e.g., a forest trail), making link quality
measurements with the previous relay from various locations, and deploys relays at some of these locations,
so as to connect a sensor placed on the line with a sink (base station) at the start of the line. While making
the placement decisions, the agent uses his prior knowledge of the statistical nature of the propagation
environment (e.g., path-loss, shadowing etc.). The objective is to minimize the expected cost of the
network, where the cost is a combination of the transmit power of the nodes, the link outage probabilities
and the number of relays used. Such cost function captures the trade-off among the transmit power,
network performance and the number of relays in the network.
We consider several variations of the problem, such as: (i) the distance to the source node from the
base station (sink node) is not known exactly, but its mean is known, versus the case where nothing is
known about the distance, (ii) one deployment agent versus multiple agents, (iii) 1-connected network
versus k-connected network, (iv) prior knowledge about propagation environment versus absolutely no
We mostly use tools from Markov Decision Process (MDP) theory to tackle these problems.