After successfully completing this course, the student can:
- assess quantitative aspects of dependable networked systems in practice, using one or more of the following methodologies: measurements, simulation, and model-based analytical methods for performance and dependability evaluation;
- explain the differences between the different methodologies, and reason about which approach(es) are useful in a given scenario
- write a research report.
The goal of this course is to assess quantitative aspects of dependable networked systems, such as wired and wireless networks, cloud applications and cyber-physical systems, using state-of-the-art quantitative system evaluation methods. Examples of such quantitative issues are related to performance (throughput, delay, losses), dependability (reliability, availability), and energy usage.|
The course is organized in the form of individual projects; for each project one staff member (of DACS) will take the role as supervisor. In general, the following four phases can be identified for each project.
First, the student studies general literature and chooses a specific application. Students may propose their own system/scenario, or come up with a nice system/scenario to be studied in cooperation with the DACS staffmember.
Secondly, one or more of the following three research approaches will be followed:
1. For a measurement-based approach the student will collect traces or use existing traces, e.g., TCP/IP (header) traces or Netflow traces.
2. Also a simulation-based approach can be chosen, using tools (such as NS2, OMNET++, etc.); this allows for studying systems that are not yet built, hence, that cannot be measured upon.
3. Analytical methods for performance and dependability evaluation may also be considered, especially if very new design alternatives or feasibility studies are considered.
Whatever choice of approach/es, further literature should be studied regarding the methods to be employed for the analysis, and tools might be studied (installation, practical experience gathering, etc.).
Third, a practical performance/dependability analysis should be performed, for the chosen system, and the chosen approach/es. Results to be obtained will be used to analyze quantitative system properties, such as throughput,delay, loss probabilities, reliability, availability and energy usage. Note that also the comparison of results obtained with different approaches does yield valuable insights. The purpose of the performed analysis is to understand, for example, dependability dependence between network components, occurrence of rare events, quality of service-provisioning of access links, admission control for fixed and mobile networks, TCP/IP or HTTP or WLAN usage in practice, etc.
The fourth part of the course is writing a research report. We aim for a style of reporting that allows submission of a research paper to a suitable workshop or conference.
The grade will be based on:
• the content (problem statement, research questions, literature, discussion, results, conclusions) and the structure of the paper (organization, readability, language use, references);
• the scientific quality (relevance, originality, complexity, contribution, validation);
• the process (level of independence, creativity, problem solving, planning, scientific reflection).
Assumed previous knowledge |Required materials-Recommended materials|
|ArticlesSelected papers, depending on topic (provided at the start of each project)|