Large Deviations and Queues


16-lecture course for Part III [Part III schedules]


Lent term 2005, Mondays and Wednesdays at 11am [Reporter lecture list]




Lectured by D.J.Wischik.


Large deviations is a branch of probability concerned with rare events: their probabilities, and how they happen. The study of queues is concerned with estimating the distribution of queue length under random traffic, with characterizing traffic statistics, and with analyzing the performance of networks of queues and of scheduling systems like priority queues.

In the past 15 years, large deviations has been applied with great success to a variety of problems in queueing theory. The estimates it yields are simple, so that many of the details which make it hard to obtain exact answers disappear, yet they retain sufficient structure to make the answers relevant.

This course will give an introduction to abstract large deviations theory; but the focus will be on applying the theory to problems in queueing networks, particularly communications networks such as the Internet.

Pre-requisite mathematics.

Knowledge of probability, optimization, and basic topology, at about second-year level, will be assumed.

Suggested reading

Teaching materials

Supplemental notes

Problem sets

Further topics