Ph.D. students
Information for Ph.D. students
General Information
Ph.D. students follow the slightly extended version of the course under course codes FEP3210 and FEP3211.
FEP3210 gives 9ECTS credits and required the following:
- minimum of grade B
- one presentation during the course on one of the course topics.
FEP3211 gives additional 3 credits. For this you have to prepare a piece of software that can be used as demonstrator during the course.
Ph.d. student presentations
The presentations are scheduled for the second lecture of a given topic, that is, lectures 4, 6, 8, 10 and 12. Select a topic and contact me as soon as possible, the latest one week before your planned presentation. Once you have selected a topic, your name will appear on this side. You can not select a topic that is already assigned.
The length of the presentation should be ca. 20 minutes. For the presentation you can use my laptop, you can forward me the slides before the lectures if you prefer. Prepare a presentation that the others are able to follow, that is, with many illustrations. If you need help with the material you are presenting, please contact me.
Some suggestions.
When you present a paper, start with giving a general picture of the paper, discussing:
- What is the problem the papers deals with, how is it related to or build on the lecture material.
- Why is this problem interesting,
- How does the paper deal with this problem.
Show some interesting results. Explain them in detail.
If you present theorems with complicated proofs: give the idea of the proof without the details.
Topics and papers:
Lecture 4: Multiple access protocols, CSMA and CSMA/CD
R. Rom and M. Sidi, "Multiple access protocols", Springer-Verlag
Explain what CSMA is, how it could improve the performance of Aloha, explain nonpersistence and 1-persistence. Take the nonpersistent slotted case and present the throughput analysis from the book.
Explain what CSMA/CD is, how it can improve the performance of CSMA, skip the throughput derivation but show some results.
Lecture 6: Congestion and error control, TCP
J. Padhye, F. Firoiu, D. Towsley, J. Kurose, "Modeling TCP throughput: a simple model and its empirical validation," Sigcomm, 1998. Read sections 1, 2-2.1, 3, 4, 5.
Introduce the problem the paper addresses. Explain "what", "why, "how".
Explain briefly how TCP Reno works (including slow start, fast retransmit, etc.)
List and explain the assumptions used in model 2.1 – why are these assumptions necessary and how are they valid? How does the model differ from the one presented on the lecture?
Explain the analysis in 2.1 and related performance figures from section 4
Lecture 8: Scheduling
Presentation 1
A. Parekh, R. Gallager, “A Generalized Processor Sharing Approach to Flow Control - The Single Node Case,” IEEE Transaction on Networking, 1993, Vol.1, No.3.
Read I-III-before part A, check V,VI,VIII
Summarize GPS, PGPS, Discuss Figs. 1,2 and Table 1, prove the main results in III (before part A). Make sure that the audience understands the examples you show. Less may be more.
Presentation 2
H. Zhang, “Service Disciplines for Guaranteed Performance Service in Packet-Switching Networks,” Proceedings of the IEEE, Oct, 1995, pp. 1374-1385 (I, II, III A,B,F,G)
Present results of III.A,B,G (parts related to A,B), IV.A,B,D,G (parts related to A,B)
Present the proposed work conserving and work non-conserving solutions. Explain the related figures with the examples. Discuss the solutions, for example, what information is needed for the scheduling and how complex the decision is. Discuss the delay, delay jitter bounds and buffer requirements. How do they depend on the length of the transmission path, in terms of switches crossed? Are they correlated with the rate of the flow? Are the results intuitive?
Lecture 10: Fairness
L. Massoulie, J. Roberts, "Bandwidth sharing: objectives and algorithms,“ IEEE Infocom 2000 (Sections I, II, III)
I suggest a possible division of the presentations, but this is only a suggestion. You may want to prepare together.
Presentation 1
Introduce briefly the problematic of fairness and distributed fairness control.
Explain the different fairness definitions in Section II and show with a numerical example how they work on the parking lot scenario. Compare their performance in terms of throughput on long and short routes and overall throughput.
Presentation 2
Consider end-to-end window based control in Section III C. Explain, how, together with some scheduling solution at the nodes it achieves given fairness objective. Present one-two objectives that are relatively easy to understand.
Lecture 12: Multimedia
[1] N. Laoutaris, I. Stavrakakis, “Adaptive playout strategies for packet video receivers with finite buffer capacity,” IEEE ICC 2001.
[2] M. Yuang, Y.G. Chen, C.L. Shen, “Dynamic playout smoothing method for multimedia applications,” IEEE ICC, 1996. (main reference in [1])
Present [1] I-IV. Make sure you explain the Markov-chain describing the system clearly. Give the state transition probability matrix and diagram for the case of Poisson arrivals (detailed in [2]).