Next: Denotational Semantics
Up: Michaelmas Term 2002: Part
Previous: Artificial Intelligence II
  Contents
Computer Systems Modelling
Lecturer: Dr R.J. Gibbens
(rg31@cl.cam.ac.uk)
No. of lectures: 12
Prerequisite course: Probability
Aims
The aims of this course are to introduce the concepts
and principles of analytic modelling, operational
analysis and simulation, with particular emphasis on
understanding the behaviour of computer and
communications systems.
Lectures
- Introduction to modelling.
Overview of analytic techniques, operational analysis
and simulation. Little's law.
- Introduction to discrete event simulation.
Applicability to computer system modelling and other
problems. Advantages and limitations of simulation
approaches.
- Random number generation methods and simulation techniques.
Review of statistical distributions.
Statistical measures for simulations, confidence intervals and
stopping criteria. Variance reduction techniques.
- Operational analysis.
Simple operational quantities. Bottleneck
analysis. Applicability and assumptions.
- Simple queueing theory.
Stochastic processes, definition and examples.
The Poisson process. Markov chains. Advantages and limitations
of analytic approaches.
- Birth-death processes, general flow balance
equations.
Relation to queueing systems. The M/M/1 queue in
detail: solution for state occupancy, average queue
length, average residence time. General observations.
- Queue classifications, variants on the
M/M/1 queue and applications.
Extensions to birth-death models. Queueing networks.
- The M/G/1 queue and its application.
The Pollaczek-Khintchine formula; performance
measures.
Objectives
At the end of the course students should
- be able to build simple Markov models and
understand the critical modelling assumptions
- be able to solve simple birth-death
processes
- understand that in general as the utilization
of a system increases towards unity then the response
time will tend to increase -- often dramatically so
- understand the tradeoffs between different types of
modelling techniques
- be able to perform a bottleneck analysis of a system
- be aware of the issues in building a simulation of a computer
system and analysing the results obtained
Reference books
Jain, A.R. (1991). The Art of Computer Systems
Performance Analysis. Wiley.
Kleinrock, L. (1975). Queueing Systems, vol. 1. Theory. Wiley.
Leung, C. (1988). Quantitative Analysis of
Computer Systems. Wiley.
Ross, S.M. (2002). Probability Models for
Computer Science. Academic Press.
Next: Denotational Semantics
Up: Michaelmas Term 2002: Part
Previous: Artificial Intelligence II
  Contents
Christine Northeast
Wed Sep 4 14:43:05 BST 2002