Performance Evaluation of Infocommunication Systems

Introduction to Internet Traffic Theory

Week 1: Review of probability theory and stochastic processes

Fundamental theorems of probability theory which are extremely important in traffic modeling and performance evaluation of communication systems. The definition and interpretation of stochastic processes. Applications of stochastic processes in modeling.

Week 2: Traffic characterization by point processes

The definition of point processes. The application of point processes for internet traffic characterization. The stationarity of point processes. The description of point processes. The process of intervals between events. The counting process. The selection of origin in traffic modeling by point processes.

Week 3: Burst and correlation measures

Interpretation of burst and correlation and their impacts on Internet traffic. First and second order characterization of burst and correlation measures. Squared coefficient of variation. Peak to mean ratio. Probability distribution function. Moments. Correlation of interarrival times. Correlation of counts. Index of dispersion for intervals (IDI). Index of dispersion for counts (IDC). Variance time plot. Relationships between measures.

Week 4: Case study for traffic analysis: from measurements to modeling

Case study for selected traffic sources: renewal processes, correlated processes, aggregated traffic, voice traffic, video traffic. Burst and correlation analysis demonstrated for the case study traffic sources with lots of figures and interpretation of results. Study guides for traffic analysis by interpreting results of actual measured Internet traffic.

Week 5: Traffic models based on Poisson processes

Derivation, definitions and basic properties of Poisson processes. Traffic models based on Poisson processes: inhomogeneous Poisson process, batch Poisson process, Markov Modulated Poisson Process (MMPP) and their use for traffic modeling.

Week 6: Traffic characterization by renewal processes

Definition and basic properties of renewal processes. Characterization of traffic aggregation and traffic splitting in Internet routers. Traffic models based on renewal processes. Application guide for renewal-based traffic models.

Week 7: Advanced traffic models (voice, video, web, etc.)

How to choose traffic models for measured traffic? The goal and use of advanced traffic models. Web traffic models. P2P traffic models, gaming traffic models, VoIP traffic models, advanced video traffic models. How to set the parameters of traffic models.

Week 8: Introduction to fractal traffic theory

Fractal properties of Internet traffic. Self-similarity and long-range dependence. Heavy-tailed distributions in the Internet. Fractal traffic analysis. Fractal traffic models. Examples of using fractal traffic models.

Internet Traffic Management

Week 9: Overprovisioning and managed bandwidth

Traffic management philosophies. The overprovisioning approach. The managed bandwidth method. Advantages and disadvantages of traffic management approaches. Traffic characteristics of stream traffic and elastic traffic. Which approach to choose?

Week 10: Traffic control, connection admission control and traffic dimensioning

Traffic categories vs. traffic control approaches. Open loop control. Peak rate allocation. Rate envelope multiplexing. Rate sharing. Statistical multiplexing. Packet scale and burst scale congestion. Designing principles of traffic control methods. Case study: Gaussian approximation.

Week 11: Future internet design principles and case studies

Trends in future internet design and dimensioning. Lessons from the past for the future. Case studies chosen from current research papers.

Performance evaluation of TCP

Week 12: Evolution of TCP:  loss- and delay-based TCP and high-speed TCP versions

Principle of TCP (Transmission Control Protocol) congestion control. Loss-based TCP variants (Reno, BIC, CUBIC, etc.). Delay-based TCP variants (Vegas, FAST, etc.). Hybrid TCP versions (Compound TCP, Westwood , etc.), TCP versions for high-speed communications (H-TCP, etc.).

Week 13: Performance models of TCP

Models for TCP. A simple TCP model and its detailed analysis. Throughput calculation based on TCP models. Advanced TCP models. Application of TCP models in network design and dimensioning.

Week 14: Case study of TCP performance evaluation

Case study: networking examples for performance evaluations of TCP-based communication. QoS characteristics calculations. Throughput and latency calculations. Numerical examples of real internet communications solved by TCP models.