5 minutes of some stuff we do

1
5 minutes of some stuff we do

• R.D. McLeod, et alia
• Department of ECE, IIC
• University of Manitoba
• Manitoba, Canada

2

• Introduction Improving Performance
• Part 1 The Transport Layer Phatpackets
• Part 2 Wireless MAC
• Part 3 Summer Projects Security Emphasis
• Part 4 Distributed Intrusion Detection
• Part 5 Game Theory Novel Techniques
• Part 6 Satellite Access Remotely deployed
  networks

3

• Consider using larger packet sizes

4

• Our protocol is called Phatpackets
• Phatpackets high performance protocol
• Open source initiative
• Rate based, more adaptive and controllable
• Emphasis on developing prototype transfer program
  at the application layer
• Maximize throughput within the resource limits of
  the collaborating hosts
• End to end performance enhancement

5

6
(No Transcript)
7

• 1. MAC modification to prevent Head of the Line
  Blocking.
• All nodes are not created equal
• 2. Energy aware probabilistic routing
• Variation of an old theme
• Exploit Multipath
• Battery consumption as new cost metric

8

• Nathack Allowing access to applications through
  a firewall without having a 14 year old set up
  your firewall for you.
• mTCP masquerading UDP over TCP
• Multiple IP address download accelerator
• Distributed Intrusion Detection (GEMs)
• Remote OS maintenance (TRLabs)

9

• This project is an attempt to deploy a network of
  intrusion boxes.
• GEMS (Geographical Event Mapping System)GEMS is
  the first Canadian academic computer security
  research done on a large scale that proposes to
  eventually exam how intrusions move across the
  Internet and what measures can be developed to
  predict or model their behaviour as they
  propagate.

10

• Phase 1, more along the lines of a packet monitor
  attempting to detect intrusion attempts.
• Long term goals are to evolve towards application
  layer monitoring or at least along the lines of
  application layer filtering.
• The application layer intrusion detection
  differentiates this project with others as we
  will be attempting to detect application layer
  malicious behavior.
• The intrusion detection system should be more
  robust as it does not depend on user intervention
  or collaboration.
• The system should be more customizable as it is a
  dedicated attached network device with a single
  purpose.

11

• GEMS (Geographical Event Mapping System) Long
  Term

2005/06 100 NOEDS (volunteers)
2006/07 500 NOEDS
12

• Models being developed that attempt to detect
  lying in agent based systems.
• Long term applications would be within networks
  that share some information but not complete
  information.
• Eg. P2P, BGP

13

• Here we are just initiating a project to develop
  a system of collaborating communication robots.
• Their access technology will be satellite.
• Satellite performance can be greatly improved by
  better transport layer protocols and some measure
  of QoS.
• Would like to obtain a ground station connected
  to CAnet for real experimentation.

14
(No Transcript)