Work in Progress

From ResiliNetsWiki

Jump to: navigation, search

Introduction

This page discusses the various aspects of my research project including terminology, major components, research tracks, current simulations, expected results and my PhD proposal.

Terminology

This section clarifies on some of the network terminology that we will use throughout the project

Challenged Networks : It is a network that is limited or is at a disadvantage in one or more of the following network characteristics

Channel capacity
Connectivity
Error rate
Mobility
Other resources like energy, memory, computing power
Traffic/load
Faults/disruptions
Natural or man-made attacks

In other words, all the networks besides fully connected, wired, Internet like networks fall under the category of challenged networks. These networks are challenged in one aspect or the other that inhibits their performance. This umbrella of challenged networks covers commonly known categories such as delay tolerant networks, disruption tolerant networks, mobile ad-hoc networks, sensor networks, MANETS, etc.

Research Components

In this section we document the various directions of the project and the anticipated work in each direction.

This work involves three major components:

Network taxonomy
Resilience metrics
Performance evaluation

Network Taxonomy This is the first part of my research. It involves identifying all the parameters that define a challenged network. We believe that various networks such as sensor networks or MANETs or any other wireless networks are all the same on some fundamental level. They differ only in the quantative values of certain network parameters. The objective of this part of the project is to identify these parameters and develop metrics with which one can define (almost) any challenged network. For example by changing a candidate metric, say, connectivity one can transform a MANET into a partitioned episodic connected network (in other words, a delay tolerant network).

The crucial note here is that the metrics should be comprehensive enough to define most of the networks and at the same time should be relatively simple and easy to understand and use.

At the end of this work, we hope to be able to present a clear spectrum of the challeneged network space using fundamental network characteristics.

Resilience Metrics This is the second part of my research. In this effort, we will develop quantative metrics to determine the resiliency of a network. A combination of these metrics will be required to address various aspects of resiliency. Also, the resiliency of a network will be based on the normal network metrics ( from the above section ) and the desired performance of the network in adverse conditions. The adverse conditions will be specified using the network taxonomy developed in the first part of the research. For example, a candidate resiliency metric, say, percentage traffic delivered will define the percentage of the traffic that can still be successfully delivered to its destination when the node density (a network metric) reduces by x% due to power failures. An overall resiliency function based on the individual resiliency metrics may be developed.

At the conclusion of the work, we will gain better understanding of the resilience in challenged networks and represent it using multivariable functions.

Performance Evaluation Platform In order to prove the use of network taxonomy in general and the effectiveness of the network metrics specifically, we will conduct ns2 simulations to evaluate the performance of popular routing protocols with varying network metrics. This shall provide a standard method of presenting the performance of the routing protocols and also help compare different routing protocols on a common platform. For example, consider a ns2 simulation on the performace of say flooding protocol when the network density changes from very sparse network to highly dense network or when the channel bandwidth changes from few bps to several hundred Mbps or when the mobility model changes from random way point to scheduled mobility. We are interested in how the protocol behaves as the fundamental metrics of the network change. Similar simulations will be conducted with AODV, DSR protocols to study their performance over the entire challenged network spectrum.

This effort will help us better understand the perfomance of the existing routing protocols so that they could be re-used in evolving networks without going through a complete research cycle or worse yet, a trial-and-error procedure to select the best protocol.

While network metrics help us to clearly define a given nework, the performance evaluation platform provides a easy way of understanding the performance of various protocols in that network region.