COM4515 Network Performance Analysis
|| This module considers the performance of computer
networks from a statistical aspect, using queuing theory. It
is shown that the performance of a computer network depends
heavily on the traffic flow in the network, and different
models of traffic and queues are used. These include
single-server queues, multiple server queues, and the
concept of blocking is discussed. Although the analysis is
entirely statistical, all the relevant background is
provided in the lectures, such that the course is entirely
self-contained. Problem sheets are provided in order to
assist the students with the course material. Students should be aware that there are limited places available on this course.
||Formal examination (2 hours)
||Dr Joab Winkler
- To introduce a statistical analysis of computer
- To show how queuing theory can be used to analyse
- To show the applications of queuing theory to other
areas of science and engineering
|| By the end of the unit the candidate should be able to:
- Analyse queuing networks and their application to
- Show how to design a computer network so that the
queues are within specified bounds
- Appreciate the importance of queuing theory to other
areas of engineering
- Review material:
- Networking Introduction.
- Random processes and probability theory.
- The Poisson distribution
- A simple network queue:
- M/M/1 queue
- Birth death processes
- Little's formula
- Richer queueing models:
- Queues in which the arrival and service rates are functions of the state system
- Queues with blocking
- Erlang delay.
- Erlang loss system.
- Erlang B and C curves.
- Overview of Simulation:
- Sampling theory.
- Obtaining samples from Markov chains.
||The course requires basic probability theory and knowledge of arithmetic and geometric progressions. This knowledge is basic and it is covered in all undergraduate engineering and physics degrees.
|| The teaching method is by lectures with numerous example
and problem sheets so that the students have a good
theoretical and practical understanding of the subject.
- 30 Hours of formal lectures, which includes problem classes.
||Problem sheets are are set and discussed in class and this
provides an opportunity for feedback.
- Data Networks (Second Edition) by Dimitri Bertsekas
and Robert Gallager
- Computer Networks by Andrew S. Tanenbaum