FINAL PROJECTS:
IP Switching
IP Switching combines the speed and capacity of ATM switch with the
robust routing of IP. It implements IP directly over ATM switch
hardware. IP Switch works in connection-less mode, and it is connected by
shared pre-established PVC (permanent virtual channel) as default
channel with its neighbors. It contains a standard IP routing software,
which can forward packets hop-by-hop in connection-less mode. In
addition, it has a classifier software which can detect flow, a sequence
IP packets with same source and destination address, by examining traffic
characteristics in IP/TCP/UDP headers. Once a flow is detected, a SVC
(switched virtual channel) is dynamically allocated to the flow. The
flow state is maintained locally and periodically.
Two adjacent SVC belonging to the same flow are switched directly by
hardware to bypass the routing software. The major performance metrics
of IP switch are percentage of packets being switched and how many VC
space is needed to do so.
I plan to first build a IP switch by OPNET, then run real Internet trace to evaluate its performance. Particularly, I plan to explore the relation between the percentage of packets switched and needed VC space, and how different parameters affect its performance.
Simulation of video on demand using caching multi-cast protocol
I plan to design a system for streaming video on demand application.
I am going to design a media server to provide video on demand service.
The server communicates with hundreds of clients and send movies to them.
Basically, I need to find the network suitable for this kind of
applications.
In order to realize real-time video playback, the network should be able to
provide some QoS service, such as CBR transmission. IP network is not
suitable
for this application since it can not guarantee the data transmission speed.
For
video playback, speed is very important and late data will slow down the
playback.
I think ATM network is one possible choice and some other networks, such as
Cable
network, ADSL, HFC, SONET are also possible. The major task of this project
is
to design the structure of media server network and how to deliver the video
data
to clients. A network with ability to provide QoS has to be chosen for this
application. I am going to design the whole system structure and use some
network
simulation tools, such as OPNET or NS software to simulate the network
transmission.
I hope that by choosing a suitable network and designing some good rate
control
methods, the media server can dynamically adjust the system behavior to
accommodate unexpected network traffic to maintain guaranteed quality of
service.
Design of a network implementing static routing of IPv6 datagrams
I plan to simulate a system for static route lookups for IPV6 packets.
In the process I will be
assuming the size of the network. I would be dealing with only
IP datagram routing and the issue of framing the datagram in to
Ethernet/FDDI packets is not being done.
I am using OPNET for my simulations.
Performance with respect to the 'End to End delay', queuing
delay and link utilization would be done for Poisson traffic
and the network analyzed.
Apart from the simulations, I would also be presenting a general
overview of the IPV6 protocol in my project report.
Study of network topology and developing a topology analyzing tool
I am proposing to build
a tools to help me get one step closer to my original goal.
Network delays greatly decreases QoS and it is important to be
able to identify where the problem lies before we can model
the problematic section of the network in more detail.
I want to build a tool that work in a real network environment that can be run from the end user's computer. The tools should describe, visually, the path taken by packets from source to the destination. It also need to generate output suitable for the NS tool.
I intend to use C program, NS, NAM and script to build the tool. The completed tool will generate an animation of the packet flowing through the network by giving it just an IP address.
Transporting Internet protocol over ATM
ATM is considered as the technology that will make BISDN a reality in
the near future. But the most pressing need at present is facilitating
the possibility of making TCP/IP available over ATM. Due to the
popularity and the extensive use of TCP/IP in inter-networking
(especially in the Internet), the success of ATM as a networking
technology lies mainly in its support of IP on top of it.
I propose to study the currently suggested approaches to this issue as
my project. Since the SFU networking system operates on top of a ATM
back bone my study can center around how this issue is practically
handled hear at SFU. Since the study is going to be based on a
practical scenario, I think the knowledge that would be gained by the
study will be more realistic and practical. At the end of the study I
propose to do a simulation on how IP traffic is send over ATM.
Simulation Fiber Distributed Data Interface (FDDI) using Opnet
Since you've mentioned it several times, I've checked what is FDDI in an
Encyclopedia of Computer Networks. I didn't know much about it, but now it seems
very interesting. Also, since Opnet has FDDI model in its model library (under
LANs) it will be of great help.
I would like to create an FDDI network model with several nodes in a ring topology (maybe I will later include also a hub topology), and to evaluate the performance of FDDI protocol (throughput, end-end delay and link utilization) by changing various parameters (number of nodes, type of traffic, load, propagation delay etc.).
Also I will give an overview of FDDI protocol (stack), physical layer characteristics, frame structure, coding techniques, different topologies and areas of application.
IP-over-ATM vs. IP-over-SONET
IP has become the network layer technology of choice for building packet
networks since its simplicity and its growth together with the Internet,
and on the other hand, SONET is the first choice by carriers and likely to
be the physical infrastructure for interconnecting routers over the WAN.
However, there are still a couple ways to deliver IP package over SONET
backbone. The most two popular methods are IP over ATM and IP over SONET.
My project will do some research on the comparison between them, and use
Opnet to simulate an Provincial wide ISP network, where it is not easy to
judge which technology would be the best, and collect the simulation
results for analysis.
WWW site:
Project (MS Word file).
Frame Relay PVC's congestion control illustration
The Frame Relay is one of WAN technologies network and now its market
share is surging up. We all have learned from the lecture about
this part but not deep into it. My project will involve in
speculating this technology.
I will study the Frame Relay from its origination, its position in OSI model, its working mechanism and its application. I will give a general description and with key parts in detail.
One of the key parts I will focus on my project is Frame Relay's Congestion Control. Frame Relay has several congestion control mechanisms, one of them has been mentioned in our textbook but without further discussion, is Leaky Bucket algorithm.
This algorithm I will simulate by Opnet since Opnet has a full set of tools and parameters to build the Frame Relay network.
My task in simulation is to illustrate how this algorithm works to control the traffic on PVCs in the Frame Relay network. I will watch the parameters (CIR, Bc, Be) and have different volume-load data flow to show that.
Network security: sniffer detection:
With the rapid development of Internet, more and more important businesses
will be performed on the Internet, and Internet is becoming one indispensable
part of people's life. As a result, security is gaining more and more serious
concerns from the Internet users. Among many possible threats to the Internet
security, Internet sniffer is a very important one. It's actually a computer
program, which resides in a computer connected to the Internet. The special
thing of it is that it can copy packets from networks furtively even if those
packets are not addressed to it. Because of sniffer's passive behavior, it's
very hard to detect it.
In this project, I would like to examine possible ways to detect sniffers, look into their implementations, experiment them on the network in our department (if allowed), and improve them accordingly.
A distributed network simulation architecture
Exploring queue-based solution to providing quality of service guarantees
For the course project, I plan to build/program a simulation model on a
TCP/IP router with the capability of guaranteeing a maximum bandwidth
(Quality of Service - QOS) to paid premium customers.
- there will be 2 basic types of customers to the router (normal and
premium customers)
- normal customers do not have a guaranteed maximum bandwidth and are
subject to traffic congestion
- premium customers are guaranteed a maximum reachable bandwidth, even if
the traffic builds up
For the project:
- a basic TCP/IP packet router will be modeled
- define and model the QOS (traffic & congestion control) algorithm
- model loads (activities) to create the traffic for the simulation
Meta-stability in circuit switched networks
To provide better QoS, the network resources must be properly
managed or controlled. In this project, I plan to study a very
simple and fundamental phenomenon called "meta-stability" which
occurs in a circuit-switched network using dynamic alternate
routing. Meta-stability is the property of a set of states of
a dynamic system that can persist for a long time, i.e., in
network communication, the state of the network can remain for
a long time in a set of undesirable states where many circuits
are busy. (The another reason I am interested in this meta-stability
is because I've seen it in many other fields of applied science).
I will analyze the cause of this phenomenon and study the solutions
to this problem (e.g., truck reservation). In addition, I plan to
use NS simulator to simulate this phenomenon.
WWW site:
Java aplet.
The UNREAL simulation
For our project we would like to simulate the network traffic generated
by the game play of Unreal, Quake or some other network game. Usually
there is a game server and several clients. We would start by simulating
a simple network and continue to increase the complexity (number of
computers on the network and/or network traffic) until we start
detecting
problems such as network congestion, dropped
packets and high packet latency. Once problems reach the level that
prevents normal game play (e.g. intolerable latency)
we would look into different ways to solve the problems, such
as different routing algorithms or different buffer sizes. In addition,
it would be interesting to study how the networked game affects normal
operation of a business. For instance, at which point the traffic
generated by
games affects the other employees who use the network for work (it
would be great if for this we could find actual network topology of a
company and gather sample network traffic during business hours).
If you think that some other area of computing would be more suitable
then computer games we would be interested to do similar project dealing
with video broadcasting/conferencing or some other application that you
would suggest.
For the project we would like to use ns simulator.
Presentation: Power-point file.
Mapping the Internet
Because of the on-going rapid growth of the Internet, the need exists to
effectively map the Internet and other TCP/IP networks. Maps, which can
represent the infrastructure, geography or conceptual domains of the
network, can be used to isolate network faults, understand the topology of
the network, to track growth patterns of the network, and provide a metaphor
for organizing meta-information.
At their basis, all algorithms for mapping large, distributed TCP/IP networks such as the Internet rely on ICMP (Internet Control Message Protocol) messages for mapping purposes. Different approaches have been used to perform mapping with ICMP including mapping from a single node, using a distributed client-server architecture where client nodes map their immediate vicinity and report their findings to a central server which guides the clients in their mapping, and groups of distributed peers who map all they can and then combine the results from all peers after mapping has occurred.
The exact goal of my project has changed slightly since its original conception. Instead of looking at the ability of these algorithms to handle network outages, I am investigating the efficiency of the algorithms. I will implement the algorithms in Pearl or other suitable scripting languages and use them to attempt to map the SFU network. With each algorithm I will acquire a database of data which I will use to generate infra-structural maps of the network. By comparing the amount of the network successfully mapped, the time taken to perform the mapping, and the amount of redundant discoveries (where nodes are discovered more than once), I will be able to compare the efficiency of the approaches in mapping TCP/IP networks.
Implementation and performance measurement of leaky bucket and
dual leaky bucket for an ATM LAN
I have decided to model an ATM LAN in Opnet in order to be able to run my leaky
bucket processor on that. Currently the network model, which I'm examining leaky
bucket on is a simple one node network. I can combine both of the projects into
one and use the real traces for running the simulation.
A survey on the time delays in the Internet
Working on a telerobotics project which involves controlling a robot 'through the Internet, I need to deal with the problem of variable time-delays.
These delays become specially import
ant when force-feedback through the Internet is necessary.
Although I have just started gaining experience
with Opnet,my current intention is to use that tool
to examine the effects of variable time
delays over a complicated network, i.e. the Internet.
Distributed Denial of Service Attacks
For the ENSC 894 project, I will be studying the Distributed Denial of
Service Attacks which are plaguing the Internet currently. My project will
have two parts, the first part is to learn more about Distributed Denial of
Service Attacks and analyze any possible defenses devised for these attacks.
In the second part of my project, I will simulate a Distributed Denial of
Service Attack
on a router using Network Simulator (ns). This would be a similar attack
to the one placed on Yahoo! in early February, 2000. In my simulation, I
will study whether different routing
algorithms effect how legitimate users obtain service while a service is
under a DDoS attack.
Project: WWW site.
Simulation of an ATM local area network with Opnet
For my ENSC 894 project, I will try to learn the simulation tool "Opnet" and
will create a LAN (most probably an Ethernet) and simulate it with different
characteristics. The model for an Ethernet exists in Opnet and I can create my
own network with its own configuration, load, number of nodes, server
characteristics, traffic, etc and simulate it.
Quality of service in voice over IP applications
Internet-based telephony service will be another alternative telephone
service than the Plain Old Telephone Service (POTS) in the coming years. It
requires a range of protocols for transporting the real-time voice signal
across the Internet, for routing, for network management and billing, in
order to maintain a good quality of Service (QoS). It also requires a
signaling protocol to establish call connections in unicast and multicast
conferences, as well as other services such as call forwarding and transfer,
placing calls on hold, or voice mailbox services. Internet-based telephony
system is an interesting topic for our projects. There are different kind
data link layer technology can be studied for the voice application. Voice
Over IP is one of the hot topics.
In the Opnet simulation programs, there is a library used for study the voice application. The library has lots of features including:
For my interests, I would like to do some works on analyzing i) the Voice Over IP model for Internet-based telephone system with different configurable parameters such as call generation rates, encoder setting and unicasting and multicasting applications. ii) The quality of Service (QoS) under different configurable parameters and traffic.
Congestion control in TCP
TCP is usually used together with IP. IP defines the unreliable,
connection-less, best effort message packet delivery. The reliable
delivery mechanism is provided by the TCP which makes use a very
fundamental techniques known as positive acknowledgment with
retransmission. This is to require the recipient to communicate
with the source, sending back an acknowledgment (ACK) as it
receives data. The sender will starts a timer when it sends
a packet and retransmits the packet if the timer expires before
an ACK message arrives. This error control mechanism, however,
wastes a substantial amount of network bandwidth because it must
delay sending a new packet until it receives an acknowledgment
for the previously transmitted packet. TCP resolves this problem
by making use of sliding window. Sliding window is also used for
flow control in TCP.
In my project, I am going to use NS simulator to explore the underlying TCP congestion and error control algorithms such as slow start, congestion avoidance, fast retransmit and fast recovery. Different TCP implementations: Tahoe, Reno, New Reno, SACK TCP, and TCP Vegas (if time permitted) will be simulated. The performance in term of throughput will be presented.
Bandwidth economy of VBR-CELP over ATM
Networks look like a kaleidoscope into which the more you research, the more excited you become.
I really bewilder
which topic i will select from so many interesting ones such as video voice tone, voice over IPv6, etc.
Finally I pick up the most interesting and practical one--voice compression in ATM.
As we know, a remarkable fraction of our conversation via telephone is occupied by "silence", at least half I suppose. Meanwhile, we pay a lot of bytes in the packet for this inevitable and undesirable silence. For sure this type of silence is not sheer nothing which we can not kick them out totally. Therefore, along with the soaring speed of DSP and improvement of compression algorithm, voice compression come onto stage as a good solution.
Data compression is widely used in the field of image processing and transmission 'cause we only have finite resource like bandwidth and transmission rate for so great amount of image information. However, voice compression is underestimated by most of professionals. Under the circumstance that we have to transmit more and more information on the relatively poorer and poorer transmission resource, voice compression will become quite urgent sooner or later.
However, solution always comes with problem. If we adopt this idea, how do ATM, which pinpoint the 53 bytes cell, transmit variant data to which constant duration of voice compressed? How long is the compression delay? How many buffers does it need? ...
After all the problem above solved, I will simulate transmission of compressed voice on ATM to check out the QoS, efficiency. At last, I will some concern on economics of scaling.
Comparison of TCP protocols: an analysis of TCP congestion control algorithms
TCP dialects, i.e. Tahoe, Reno, new Reno, and Vegas,
have different congestion control intrinsically. By
comparison of the algorithm strategies and ns
simulation results, we will find out the cause of the
different performance of TCP congestion controls. We
plan to evaluate following algorithms: Slow-start,
additive increase/multiplicative decrease, Fast
Retransmit, Fast Recovery, Rate-Halving etc. Further,
we may improve some algorithm by adding some
modifications and build our own C modules. The results
of ns simulation will be presented by nam-1, Xgraph
tools.
OPNET based Simulation of a Cellular Telephone Systems
Project Objective:
Background:
TDMA/FDMA based Cellular Telephone Networks employ Spatial Frequency Reuse
and Spectral Efficiency. A mobile uses a
channel from the base station while within that cell and is handed off
to another cell when it moves out. Cell Shape & Grouping provides
trade-off between Cell Capacity & Interference. The mobility and call
generation parameters affect the blocking and dropping probabilities.
The main objective of this project is to model a cellular
telephone network, and subsequently measure the impact of mobility
and call generation parameters on the call level QoS parameters.
Project Plan:
Using OPNET's FDMA based AMPS (Advanced Mobile Phone System):
Simulation of TCP over wireless links using ns
All versions of TCP apply a common window adaptation procedure for congest
control. If a packet is lost, however, in TCP-Old-Tahoe, retransmission will
start only upon timeout, and newer versions of TCP - Tahoe, Reno, NewReno,
and Vegas, implement different fast-retransmit and fast-recover schemes.
I would like to use NS to simulate the performance impact of such schemes
over mobile radio link. Also I am interested in the effect of varying
wireless link conditions. I noted that the new NS 2.1b6 version had included
CMU wireless extension. It will be of great help to do the simulations.
Comparison of ABR, CBR, UBR over ATM networks
I am going to use Opnet to model video traffic over an ATM network.
The ATM Forum has specified some QOS requirements for this type of traffic. And those QOS requirements will have to be considered when designing networks or developing specific control access mechanisms for video traffic services.
For this reason, it is important to develop sophisticated traffic models to carry out accurate design, and to perform evaluation studies of networks supporting video services. I think that modeling compressed video traffic is an essential tool to analyze the performance of ATM networks.
I am going to use 100 to 500 digitized video frames from a movie, and choose
a video traffic model to evaluate the performance of an ATM switch with
multiple compressed video channels.
Presentation (ppt file):
Using Opnet to Model Video Traffic over ATM.ppt.