Skip to main content
Department of Information Technology

Projects

Integration of Mobile IP and Ad hoc Networks

Introduction

Mobile IP promises to enable global mobility in the Internet while maintaining uninterrupted connectivity. This is acheived by tunneling a mobile host's traffic from its current location in the Internet to the host's home network. This gives the illusion that a mobile host is always reachable "at home".

Mobile IP relies on the availability of Home and Foreign Agents. Foreign Agents in visited networks are usually discovered by regular agent advertisment messages or by a mobile host's agent solicitation message. This system is built around the fact that mobile hosts can reach the agents over one hop in the network layer, i.e., without intermediate routers forwarding the message. However, this is not the case in ad hoc networks where agents/gateways may be located multiple hops away from the mobile hosts. Furthermore, the availability of several agents may complicate the agent selection process for mobile hosts since hosts on the same multihop path may ultimately want to use different gateways.

Goal

The goal of this project is to improve the implementation and integration of Mobile IP and the AODV ad hoc routing protocol and to understand the interactions between the protocols. Experiments should be performed to measure the performance and hand-over time when switching between gateways.

Pre-requisites

The students should be familiar with working in a Linux environment and preferably have experience with Kernel hacking. Programming is done in the C programming language. The students will be given implementations of AODV and Mobile IP and code that is an initial attempt at integrating them. The hardware that will be used are LinkSys routers and Linux laptops.

Number of students

2-4

Contact person

Erik Nordström

Hardware Requirements

2 Linksys routers acting as Foreign agents, 1 laptop (Mobile Host), 1 Stationary machine (Home agent).

MLN/Knoppix

Introduction

In order for students to perform networking labs they often need to be in a physical lab during scheduled hours. The need for scheduled hours
is because you often want to block the lab environment from the surrounding network. If the labs could be run on one physical machine booted from a CD with the lab environment, this cd image could be distributed to students, which then could do the labs at home or in the lab outside scheduled hours.

Goal

The goal is to produce a disc image that can be either netbooted or
booted from CD http://www.knoppix.org/. This image should contain MLN
http://mln.sourceforge.net/ and preconfigured setups for labs.

Pre-requisites

Multiple subprojects can be designed, so not all knowledge is needed by all participants.

  • Linux knowledge
  • Programming experience

Number of students

2-6

Intended student group

N/A

Hardware requirements

1-2 PCs, 1 CD burner

Contact person

Olof Rensfelt

Modem Over Skype

Introduction

Skype is a VoIP system which has become very succesful recently. It offers encrypted voice communication between users and has a very good reputation for its sound quality. We would like to try to run data traffic over the encrypted sound channel by using a software modem. We expect to run into trouble since the codec used is heavily optimised for speech.

Goal

The goal is to transfer data between two machines running skype by hijacking the sound input/output from skype and redirect it to a software modem.

Pre-requisites

Linux Operating System knowledge.

Number of students

2-3

Hardware Requirements

2 PCs

Contact person

Olof Rensfelt

Performance evaluation of Voice over IP and IEEE 802.11

Introduction

The purpose of this project is to evaluate the quality of VOIP/WLAN on the newly installed centralised WLAN, both by subjective measurements as well as quantative measurements. Voice has real-time requirements and is vulnerable to delay and losses. The IEEE 802.11 has the capacity for carrying voice but does not have any means to make bandwidth reservations for real-time traffic. Instead, the network must be dimensioned so that it will not be overloaded. This dimension include how close the base stations are located. Voice over IP is also sensitive to the delay caused by hand-over between base stations and packet loss caused by weak signal strengths. Wireless voice clients, i.e. telephone, are equipped with different voice codecs that compensates for losses and delay jitter buffers that handles temporary delays.

In this project you will work with state-of-the-art wireless clients and infrastructure. They may include Skype, CISCO IP-phone and PocketPC based PDAs from the start-up company OptiMobile. The company provides VOIP over WLAN with seamless handover to GSM when there is poor or none WLAN coverage. The advisor of this project will come from the company.

The project consists of designing a set of performance measurements, some Linux or PocketPC programming and exercising the experiments. You will learn how IP telephony works and the characteristics of wireless hot spots.

Goal

Design and run test cases for the evaluation of VOIP over the existing WLAN. Standardized subjective schemes will be used as well as quantitative measurements of packet loss, delay, voice gaps and throughput. Stress testing under background load combined with and without the QoS-classes
that the WLAN-infrastructure offers.

Prerequisites

None. Experiences from PocketPC or Linux will be helpful but not necessary. Two LinkSys routers, one PDA with IEEE802..11 and GSM interface, one laptop with Linux driver.

Number of students

4

Intented student group

IT/DV/MN

Contact person

David Lundberg

Using sensor data to generate random bitstrings

Introduction

Secure communication requires a cryptographic key to encrypt and sign messages. Some devices do not have user interfaces that allow to enter or install keys. We are looking for alternative ways to generate cryptographic keys, for example through shaking a device and using the unique shaking pattern as a key.

Goal

In this project we want to generate (pseudo) random bitstrings that can be used as cryptographic key. Instead of applying mathematical algorithms, we investigate to use measurement data from sensors like movement sensors to create random bitstrings. We will test the resulting bitstrings with statistical randomness tests.

Pre-requisites

creativity, interest in experimenting, interest in mathematics.

Number of students

2

Intended student group

N/A

Hardware requirements

2 sensors, computer to work on the sensor data (off-line)

Contact person

Christian Rohner

Intermittent networking testbed

Introduction

Imagine you are hiking on Kungsleden in the mountains and want to send an SMS to your friends at home. Unfortunately there is no GSM/3G coverage. However, there are other people hiking, and some of them are going towards a village or a mountain peek where there is coverage for sending an SMS. Why not ask them for a favour?
This scenario is one example of intermittent networking where node encounters and mobility replace direct connectivity.

Goal

We want to build a testbed for intermittend networking: A first goal is to implement the Prophet routing protocol to distribute messages in a "smart" way. We will do that on laptops with WLAN, possibly also Bluetooth. The purpose of the testbed is to do controled experiments where we can learn how messages get distributed. Detailed logfiles and analysis tools are therefore also part of the testbed. Depending on the interest and the number of students we might extend the testbed to Bluetooth and Java capable mobile phones.

Pre-requisites

creativity, interest in programming and doing experiments

Number of students

2-4

Intended student group

N/A

Hardware Requirements

2-3 mobile phones, 1 Linksys with Bluetooth module

Contact person

Christian Rohner

Evaluation of the Internet Indirection Infrastructure (i3) on PlanetLab

The Internet Indirection Infrastructure (i3) is an overlay network used to allow a "rendezvous" communication style. This means that endsystems in the Internet do not communicate directly, but rather via an indirection point. The planetlab is a global, distributed testbed which is used by academics and researchers all over the world to test and deploy new research software. i3 is implemented for the planet lab and we are interested in investigating how well it behaves. Specifically, to attempt to duplicate the research findings detailed in the i3 papers.

Goal

The goal of this project is to get i3 running on the planetlab, on as many nodes as possible and to test how well it performs by using the metrics detailed in the i3 paper. Any additional information gathered about how well i3 performs is also extremely useful.

Prerequisites

Linux systems programming using C

Number of students

3-4

Intended student group

N/A

Hardware requirements

Planetlab account (not really hardware ;)

Contact Person

Richard Gold

Decentralized Bit Torrent client

Introduction

BitTorrent's approach to file download is based on the concept of a tracker that allows clients to locate members of the peer to peer graph interested in that file and holding copies of it. The idea is to investigate a distributed approach to tracker implementation building redundancy into the BitTorrent protocol and potentially reducing the bottleneck imposed by a single tracker.

Issues which will need to be investigated when implementing a decentralised tracker include
- consistency between tracker copies (is it necessary, if necessary how strictly consistent?)
- performance implications

Goal

To investigate the potential benefits of providing a decentralised tracker implementation, potentially based on a distributed peer network that maintains tracker information. The target platform for the implementation is PlanetLab.

Background Reading

http://www.bittorrent.com/bittorrentecon.pdf

Pre-requisites

Some exposure to P2P systems, interest in P2P implementation, socket programming.

Student Group

The project is suitable for a group of 4-5 students in either DV or IT programmes.

Contact Person

Arnold Pears

Updated  2005-04-06 12:48:25 by Richard Gold.