Sergio Servetto

Grading | Projects | Reading Materials | Syllabus | Back to Teaching

ECE 446: Digital Communications over Packet-Switched Networks

Announcements:

5/22, 2:45am: The course is over. Please find below the grades in your final exam. The letter grades you will find in "Just the Facts" in a couple of days. Thank you very much for having been such a great class! It was a pleasure working with you during this spring.

5/22, 2:45am: Letter grades have been assigned already. These are the considerations I observed in assigning these grades:

We computed numerical grades as follows:
- We computed one grade for the homeworks: we normalized all homeworks to 100 points, dropped the worst two, and averaged the remaining six.
- We computed one grade for the midterm: we took the maximum between the original grade and the average of the two grades (original and redo).
- We used the final grade listed below.
In the grading policy for the class stated here, midterm/homeworks/final each carried a weight of 20%. So we averaged these three grades, to obtain a tentative ranking in the class.
In this tentative ranking grades above 80 were assigned letters in the A range, grades above 70 were assigned letters in the B range, and grades below that were assigned letters in the C range. (For this ranking, the mean was 71 and the standard deviation was 11.) Note however that this does not incorporate projects/presentations yet.
The final letter grade was bumped up/down or remained unchanged based on three factors: my evaluation of your project report, my evaluation of your presentation, and my evaluation of your slides. I found it very hard to assign a numeric value to the project in the same way that I would assign a numeric value to your final. Instead, I graded projects as:
- Good. This means you had: a well organized report, with clearly identifiable points to make on the topics of our class, well delivered and concise presentation not going significantly overtime, clear slides. Overall a well defined and well rounded project.
- Ok. This means I thought you did a good job, but one or two aspects of your project could have been improved.
- Bad. This means I thought your project was disorganized, or it was not clear what you were trying to convey, or your presentation was confusing, or there was some other substantial deficiency.
With my evaluation of the project, depending on how good/bad I thought your project work was, I bumped up/down your tentative letter grade. This is the final grade that you will be able to see in "Just the Facts".
Distribution of letter grades: A: 2; A-: 5; B+: 3; B: 3; B-: 3; C+: 3; C: 5; C-: 2.

5/22, 2:45am: Grades for the final exam (FakeID, grade):

14: 32
10: 40
05: 46
09: 47
24: 50
27: 56
23: 56
22: 57
13: 57
29: 58
30: 61
12: 61
17: 64
20: 66
08: 66
21: 67
15: 68
25: 68
19: 69
28: 73
18: 73
16: 76
06: 78
26: 84
07: 85
11: 90

Mean: 63. Standard deviation: 13.

Older announcements:

5/16, 4:40pm: More on the final. It appears I made a mistake. The exam will last for 2.5 hours (12-2:30pm), not 3 as I had said originally. I apologize for the mistake. Also, like the midterm, the final is open everything-but-network-connections.
5/15, 11:45pm: On practice problems for the final. Guys, I have received a number of inquiries over the last few days about practice problems for the final. There are some fairly detailed guidelines below on how to prepare for the final, and you can expect a final that closely resembles those guidelines. There will be no extra problems for practice. Just look at those guidelines, and make up your own practice problems. That should be plenty of preparation for this exam. Good luck!
5/10, 1:50pm: Topics to focus on when studying for the final exam:
- Basic probability: binomial/geometric/Poisson distributions, how to use them.
- Layered Network Architecture.
- Layer 2 in LANs, Layers 3/4 in TCP/IP Networks.
- The Aloha protocol, TCP's algorithm for window flow control.
- Dijkstra, Bellman-Ford, and computation of routing tables.
- The operation of a switch.
Guidelines for how to study:
- For the descriptive parts: be certain to get your descriptions right! (e.g., don't confuse a layer 2 bridge with a layer 3 router, don't mix functions implemented at different layers, etc).
- For the algorithms parts: be ready to simulate by hand the execution of any of these algorithms. Make up your own examples of a graph with a root and link delays, and compute routing tables by hand. Or of a multiaccess network with three nodes, and run Aloha by hand. Or ... Just know how these algorithms work very well -- there is a reason why I made you write simulations, and this is it.
- For the performance analysis questions: anything that I will ask you will involve setting up either a geometric or a binomial or a Poisson variable, and computing a mean or a variance. Make sure you understand in what cases each one of these applies.
Coordinates of the final exam: Tue May 17, 12-3pm, 403 Phillips.
5/09, 1pm: All missing slides for lectures available.
4/16, 1am: Notes on TCP sockets: [Download]; notes on UDP sockets: [Download].
4/16, 1am: Midterm solutions: [Download].
4/12, 2:25pm: Van Jacobson's article referenced in today's lecture: [Download]
4/05, 11:10am: Information on projects available (see Projects).
4/05, 10:20am: Syllabus updated.
3/08, 5:15pm: Time for the recitations: Wednesdays at 6pm.
3/03, 11:30am: In today's lecture, we will have Mr. Paul Schuh, one of our gurus who run the ECE network, come and talk to us about the design and configuration of this network.
3/03, 11:30am: As announced by Yiorgos last Tuesday, since yesterday night (Wed 3/02), we have a recitation section in which the goal is to go over the solutions to the homeworks after you turn them in. We will discuss today in class if there are better times to do this, but for now, this recitation will take place on Wednesdays at 7pm.
2/16, 7:30pm: Two people who developed some of the very first protocols for the Internet (that we will be studying in this class), are going to receive the ACM Turing Award (like the Nobel Prize in computer science). See article in the Technology section of the New York Times.
2/15, 7am: Some interesting articles from the Technology section of the New York Times:
On policy issues related to the Telecom industry.
On video signals delivered to cell phones.
2/11, 6:30am: My apologies for the cancellation of class yesterday, will discuss on Tuesday. For now:
Read the lecture notes posted yesterday (lecture #5).
Turn in your graded hw2 on Tue 2/15, together with hw3.
2/09, 6:35pm: I have changed the mechanism to restric access to the solutions to one that requires you to type in your netID and password. Let me know if you still have trouble getting access to these solutions.
2/08, 2pm: A seminar that looks interesting:
Group Dynamics in Silico: Complex Networks and the Human Flock,
by Prof. Macy of Cornell's Department of Sociology, this Thursday after class. Go if you can.
2/08, 8:15pm: Solutions for homework #2 posted.
2/08, 4:15am: Homework #3 posted, Yiorgos' lecture (#2) scanned and posted.
2/06, 11:45pm: An interesting article from the Technology Section of the New York Times, relevant to our class.
2/03, 8pm: Lecture #3, solutions and guidelines for hw#1 posted.
1/26, 1:45pm: Lecture #1 posted.
1/25, 1am: Grading policy and syllabus updated, homeworks #1 and #2 posted.
1/25: First lecture.

Lectures:

Lecture #1 (Course presentation).
Lecture #2 (Network Mechanisms -- TA's lecture).
Lecture #3 (Network Mechanisms / Layered Architecture).
Lecture #4 (The OSI Reference Model).
Lecture #5 (Ethernet / Token Ring -- cancelled).
Lecture #6 (Ethernet and other technologies).
Lectures #7 and #8 (Internet protocols).
Lecture #9 (SONET/WDM).
Lecture #10 (DSL/CATV -- TA's lecture).
Lecture #11 (Paul Schuh's presentation on the ECE Network).
Lecture #12 (Basics of cryptography and cryptographic protocols).
Lecture #13 (Firewalls, Attacks and Countermeasures).
Lecture #14 (Recap of Part I).
Lectures #16 and #17 (Multiaccess).
Lectures #18 and #19 (Flow control).
Lecture #20 (Routing).
Lectures #21 and #22, Supplement (Graph algorithms -- TA's lectures).
Lecture #23 (Switching -- the problem).
Lecture #24 (Switching -- achieving 100% throughput).
Lectures #25 and #26 (Project presentations).

About the course:

Instructor:

Sergio Servetto.

TAs:

Ron Dabora.

An-swol Hu.

Georgios N. Lilis.

Mingbo Zhao.

Meeting times and place:

TuTh 2:55-4:10pm, 219 PH.

Office hours:

Instructor: Th, 4:10-5:30pm, 219 PH.

TAs: MWF, 1:30-2:30pm, 338 RH.

Prerequisites: ECE 301 or 320, and 310 (required), ECE 220 (recommended), and some knowledge of computer programming. Some useful related courses, recommended for either concurrent registration, or for the previous/following semester: ECE 425, ECE 426, ECE 467, ECE 468, CS 414, CS 519.