CS-204: Advanced Topics in Networks

Michalis Faloutsos

Time and Place: Check with Terri and the grads archive list
and the csgrads mailing list, since it is BOUND TO CHANGE again.
For now (Jan 25): Olmstead Hall 1136
Time: Tu-Thur 9:40-11:00

NOTE: The page for last's years class contains a lot of information
                on conferences and journals, tools, and some scripts.

Follow the project link to see some samples of projects.

By Feb 8th, you should have a group and a preproposal (1 or 2 pages).
Take the preproposal seriously: you should identify an area, goals,
methodology/approach, expected results.
I am doing this to ensure that you are on track.

Tu, Feb 15th: we will most probably have the quiz/exam on the textbook.
Keep in touch and check this page again for possible changes.
Tips: a list of acronyms you should be aware of , thanks to Yong Cao's help.
The issue is not to memorize the origin but understand what they really mean and what is their importance.

TU 16th March: The projects are due.             <-------- EVEN NEWER ------
After popular demand, the deadline for the projects was moved.
Note that now this becomes a hard deadline.
Exceptional cases of emergency may be considered, but
poor planning or coursework are not valid excuses.
See also comment for literature surveys below.

Thu 15th March: we will have the exam on the papers.
NOTE: Read the papers we have covered.
My original goal was to have you read all asterisks papers but I DON'T.
This though increases my expectations from the project.

Friendly tips:
Start reading the textbook. You should be done with it by early Feb.
Start reading the papers. Don't get stuck in details, try to understand the innovation and the main concepts.
Start thinking of a project. The two previous activites will help you.


Paper  Discussion Schedule

  I intend to start after the quiz i.e. Thu 17 Feb. Things may change, check this page again.
 Tip: Read the papers with the (*) under the related heading.
  We will discuss papers in the order with which they appear.
 You should be able to discuss it and/or present the paper main concepts.

First week: multicast
Second week: Internet traffic
Third week: Network Topology - Wireless
 Fourth week: Architecture - Congestion Control
 (week = two lectures)



DESCRIPTION OF COURSE SCOPE

This is a course for people with a) serious interest in networks,  and
b) non-trivial background in  networks. The goals is to help
potentially interested students  see what networking research is all about, and
 to neworking students prepare for their research
(i.e. start their thesis work) and the networks depth exam.

Expected workload: I will do my best to make sure that an average student
(with the appropriate background) will be able to hanlde this class, plus
 a similar workload grad class, and a seminar.

Please read the warning.

 Note that the class is quite different and more demanding than that in  Winter 2000. 


OVERVIEW

The class will be divided in two parts: the lecture part and the
 research part.

Lecture Part.  (first half)
 Overview of computer networks with Internet as a case study.
We will examine fundamental principles and most recent developments.
 At the end students will have a pretty good idea how Internet works.

Research Part. (second half)
 We will study a fair amount of papers 15-25 from the latest conferences
 and journals. Students will be expected to read a number of papers (1-2)
 for each class and discuss it. Students are expected to show up in class for
 this part (see below the 10% participation).
Here is the initial list of papers. I will be adding things soon.

Literature Surveys: Literature surveys are acceptable but a pure literature survey
is not going to get maximum grades (sorry Chris).
A good literature survey:
 * identify the important papers: starting from the good conferences and journals
 is a good strating point (SIGCOMM, INFOCOM, ICNP, MOBICOM, MOBIHOC)
* synthesize the information: pure listing of papers is not a literature survey
    you have to highlight similarities, classify papers and approaches, compare them
* it has to be clear, well written, with one major topic in mind, develop
    top down (problem - main classes of approaches - subclassess of its class- etc)

A literature survey becomes someting more when it goes beyond surveying and it identifies
open problems,  proposes new methods or combination of existing methods etc.
A reasonable lit-survey typically has at least 20 papers a good one around 35
and an extensive one 50 or more. HOWEVER, these numbers are indicative:
First, it is the quality of papers that you use that matters.
Second it depends on the topic: the goal is to cover a particular subject.
 Tip: select an appropriate topic so that the amount of papers that you read are
  adequate.
Surveys just like other projects will be judged on a) topic interest, b) completeness,
c) innovation (new perspectives and insights), d) quality of writing.

Prerequisites: Students are expected to have taken an undergraduate class in
networks or equivalent (related working experience, self study, incredible
brain power, or determination).
I may exercise my right to not accept a student  if I can sense they won't make it.
Typically, I let students assume the responsibility of their actions, because I believe
they know better their limitations.

Student Evaluation

60% Non-trivial research project
30% One exam  or two 15% exams (in-class with 4-page personal notes).
10% Class Participation

WARNING: The project will be a non-trivial original piece of work. The end result should
be at the level of a decent workshop paper.  Literature surveys are acceptable. However,
they have to be thorough, synthesize the read material, and offer an interesting perspective
on the state of the art.
I will offer several ideas for projects but identifying a topic is mainly the student's duty.
Note: In previous courses, I have been very flexible on this issue. This time I will be strict.
I will be very happy to help with the report, or explain what are the expectations
but the expectations have to be met at the end. "I didn't know that I had to go that far"
is not an excuse.
Here are some suggestions and guidelines for the projects.

Here is my some guidelines of how a well written paper should look like: The structure of a succesful paper
Every case is different, by when you diverge too much from the given structure
you should think twice and probably have a very good reason for it.


In more detail....

Lecture Part

Text-book (strongly suggested but optional):
   Computer Networking:
   A Top-Down Approach Featuring the Internet
   by James F. Kurose and Keith W. Ross
   (reference texbook)
   It could be found on-line last year here:
   http://gaia.cs.umass.edu/kurose/Contents.htm

We will cover the following chapters from the text-book:

Ch. 1 Introduction (All)
Ch. 2 Application Layer (All. Except 2.6)
Ch. 3 Transportation Layer (All. In brief: 3.4)
Ch. 4 Network Layer and Routing (All)
Ch. 5 Link Layer and Local Area Networks (ONLY: 5.1, 5.3, 5.5.1, 5.5.2, 5.7, 5.8)
Ch. 6 Multimedia Networking (All.)

Research Part

    For an idea of the type of papers that you will have to read
    see the 204 site for Winter 2000.
    See also the requirement of the Networks depth exam to get an
    idea of the type of papers that we will cover.  Note that the course
    material is only s subset of the Networks depth exam.