Task

Objectives

The objective is to make the student familiar with the common tools and techniques for investigating the dependability and traffic handling properties of telematic systems. Emphasis is put on the carrying out of systematic and well documented investigations, as well as extraction and presentation of the findings in a style suited for publication.

Tasks

The lab consists of four tasks outlined below. The first task is somewhat larger than the other tasks which have an approximately equal workload.

All tasks are based on the simplified network server cluster sketched in the figure below. The students may choose to study either the performance or the dependability aspects of the system.

It is required that tasks 1, 2, and 3 are done satisfactorily in order to pass the subject.

The full task definition can be downloaded here. Below is an outline of the tasks.

  1. Simulator. Based on the description of the network server cluster, its environment and operation policies, a DEMOS based simulator shall be developed. With this simulator, a series of experiments shall be carried-out. The results will be used in the succeeding tasks.
    The students should decide what functionality they must implement to be able to produce the output required for Task 2 and 3 (i.e. decide what aspects are related to the performance or dependability part of the system).

  2. Data analysis. The objective of this task is to perform statistical analysis on a log of events which may stem from the operation of a system, a simulation, measurement equipment etc. Simulation traces are used.
    • If dependability is chosen, question 1 and 2 of Task 2 are mandatory.
    • If performance is chosedn, question 3 and 4 of Task 2 are mandatory.
  3. Evaluation. The objective of this task is to compare the results obtained from task2 to the results from some simplified theoretical models to find out the pros and cons of those models.
    • Traffic analysis. The traffic handling capabilities of the network server cluster are modelled and analysed, and results compared with simulations. This task is mandatory if performance is chosen.
    • Dependability analysis. The dependability of the network server cluster is modelled and analysed. Both block schemes and state diagram (Markov model) techniques shall be used, and the results are compared with simulations based on less rigid assumptions. This task is mandatory if dependability is chosen.