Network Performance

This is a 30-lecture course taught to masters students and fourth-year undergraduates at UCL. I teach basic tools for modelling computer networks: looking at data, using appropriate simulations, and analysing mathematical models.

• Course details & arrangements
• Handouts & resources
• Problem sets & past papers
• Further reading

Course outline

• Section 0. Introduction

  [pdf]

• Section 1. Random numbers

  [pdf]
  §1.1 Introduction, §1.2 Describing random variables, §1.3 Common distributions, §1.4 Generating random variables, §1.5 Fitting distributions, §1.6 Working with distribution functions, §1.7 Independence, §1.8 Working with random variables, statistical multiplexing, §1.9 The normal distribution, §1.10 Abstract random variables.

• Section 2. Random processes

  [pdf]
  §2.1 Markov chain, §2.2 PageRank, §2.3 Other calculations with Markov chains, §2.4 Formal properties of Markov chains, §2.5 Markov process, §2.6 Stability of Markov processes, §2.7 Poisson process.

• Section 3. Markov job models

  [non-final pdf]
  §3.1 Processor sharing model of TCP, §3.2 Erlang link, §3.3 PASTA property, §3.4 FIFO queue, §3.5 Kendall notation, §3.6 Network examples, §3.7 Symmetric queues, §3.8 Quasi-reversible networks.

• Section 4. Tools for analysing random processes

  [non-final pdf]
  §4.1 Drift models, §4.2 Erlang fixed point, §4.3 Dynamic alternative routing, §4.4 Operational laws.

• Section 5. TCP

  [non-final pdf]
  §5.1 Windowed flow control, §5.2 Jacobson's algorithm, §5.3 Drift model, §5.4 Fixed point, §5.5 Teleology, §5.6 Teleology and routing.

• Section 6. Conclusion

  [non-final pdf]

Pre-requisites.