The first three exercise sheets have been handed in for our ananual undergrad course on Internet networking basics (see the NPA course page for further details). Here are some observations:

• The problem of plagiarism arised again, although we stated very clearly that this has never been allowed and actions will be taken.
• It seems like students have realised that we find the original sources in the Internet from where a text has been copied easily. I see more and more solutions containing citations. Unfortunately, the amount of text to be copied and cited seems to be unclear. Supporting ones argumentation by citing relevant and high quality sources is a clear benefit for a solution and very highly regarded. However, I also found cases where an entire solution has been copied from the web and completely cited. However, a solution should have a own contribution which is significant. Only a single citation cannot be graded.
• We use a self-implemented plagiarism detector to check if students copy solutions from each other. We highly encourage group work, but we require each individual student to be able to explain the solution in his own words. This is self-check to verify if things have been completely understood and thus supposed to be helpful for the students. The results are extremely positive so far; we did not found any copied solution so far.
• Foreign students in the German exercise groups tend to produce more correct German than some of the native speakers.
• We migrated the course to the Moodle e-learning platform hosted campus wide by TU Berlin and used by most of the courses. It has a lot of promising features like anonymous polls, user friendly discussion boards, online grading and grade notification etc. However, the discussion board could be used a lot more frequently. Questions tend to be raised when it is already to late to answer them (e.g. problems with the assignment).
• We offer the course in two languages simultanously, which means we hold the lecture twice and proivde an English tutorial session. However, as in the last year we started a poll asking for feasible time slots to place the tutorials and prefered languages. It was suprising to see that English was much less required than in the last year.

One of my student reported a funny situation when he was using whois to solve the exercises in my group on Friday:

$ whois google.com

(…)

Server Name: GOOGLE.COM.SUCKS.FIND.CRACKZ.WITH.SEARCH.GULLI.COM
IP Address: 80.190.192.24
Registrar: EPAG DOMAINSERVICES GMBH
Whois Server: whois.enterprice.net
Referral URL: http://www.enterprice.net

The course registration for the Internet Routing Seminar is over and a good amount of students have registered for the course. The next step consists of assigning papers to students according to their choices. However, based on their choices I have created a small, non-representative paper popularity.

Paper Popularity

For the course registration, students have to pick five out of 23 papers and assign priorities to them (priority 1, …, 5). So every student has five distinct choices for possible topics. Based on these choices, I created a paper popularity rating by simply weighting the choices by their priority (amount of choices for a particular paper in priority 1 is weighted with a higher score than the same paper appearing in priority 5). I used the following weights: priority w_1=5, w_2=4, w_3=3, w_4=2, w_5=1. The weight is then computed by s = \sum_i=1^5 {a_i * w_i}, where a_i is the amount of votes per priority for a paper and w_i is the weight (see the last sentence). Following this procedure, I end up with the following paper priorities, where s indicates the score:

1. s=23 BGP routing policies in ISP networks
2. s=17 Building an AS-topology model that captures route diversity
3. s=15 A Light-Weight Distributed Scheme for Detecting IP Prefix Hijacks in Real-Time
4. s=14 An Analysis of the Skype Peer-to-Peer Internet Telephony Protocol
5. s=13 BubbleStorm: Resilient, Probabilistic, and Exhaustive Peer-to-Peer Search
6. s=12 OpenFlow: Enabling Innovation in Campus Networks
7. s=12 APT: A Practical Tunneling Architecture for Routing Scalability
8. s=10 Ant colonies for Adaptive Routing in Packet-switched Communications Networks
9. s=10 Incentive-Compatible Opportunistic Routing for Wireless Networks
10. s=9 Avoiding transient Loops during IGP Convergence in IP Networks
11. s=9 Ariadne: A Secure On-Demand Routing Protocol for Ad Hoc Networks
12. s=8 In VINI Veritas: Realistic and Controlled Network Experimentation
13. s=7 Interdomain traffic engineering with BGP
14. s=4 A Measurement-based Study of the Skype Peer-to-Peer VoIP Performance
15. s=3 Designing Extensible IP Router Software
16. s=3 HLP: A Next-generation Interdomain Routing Protocol
17. s=2 On Count-to-Infinity Induced Forwarding Loops in Ethernet Networks
18. s=1 Unified Energy-Efficient Routing for Multi-Hop Wireless Networks
19. s=1 Power Awareness in Network Design and Routing

19 papers have been selected by students, 4 have not been considered. Thus, 82 % of the pre-selected topics have been taken into account.

The new term starts at TU Berlin and our group offers an advanced seminar dealing with Internet routing (see details here). Today was the preparatory meeting where I introduced the basic concepts of this seminar and its topics. Slides covering all the details can be found here.

In contrast to previous seminars, I agreed to introduce a new feature; students can suggest papers as we do not want to stop them from working on interesting topics, if they can’t agree on the papers collected and suggested by us. I wonder how many students will take advantage of this, or it might turns out to be too much unessessary overhead (although I prepared a list of relevant conferences and linked their programs, students will have to investigate some amount of time in looking through them, which will of course not result in any grading benefit).

There is a tendency of more and more people from academia getting involved in blogging and thus provide valuable and interesting insights into their daily life as researchers. Christian Spannagel, assistant professor at PH Ludwigsburg, is an researcher who can be considered as an open scientist. Open scientists intend to make processes more transparent to the public, e.g. by blogging about interesting things. Similar to the idea of open scientists, there is also a tendency of having “open teaching“. Christian Spannagel is an ardent advocate of this method and opens up his teaching philosophy to the public where he wants to reflect his teaching methods, discussions and feedback of talks and classes.

I’m happy to see that others follow a similar path and provide more insight into their courses. Course CS263: Wireless Sensor Networks, which is a graduate seminar held in the spring term 2009 at Harvard University by Matt Welch, is covered by an official blog to post notes and musings of papers discussed in the class. The blog is highly interesting as it makes discussions within a seminar public. Speaking from my own experience, students come up with lots of interesting musings during discussions on a particular paper which are normally “lost” but are archived and published in this way.

It would be great if these examples would catch on, e.g. by having more reports on scientific conferences. As one example I would like to mention the NSDI’07 blog coverage.

Reference:

• Harvard CS263 Blog
• Blogging as a research project (about the course blog)

German blog posts on the concept of an open scientist:

• Der öffentliche Wissenschaftler
• Öffentliche Wissenschaft: Eine Bastelanleitung
• Der öffentliche Wissenschaftler: Diskussion zu Vorträgen
• Feedback zum öffentlichen Wissenschaftler
• Feedback zu meiner Lehre
• Ermunterung zum bloggenden Wissenschaftler
• Öffentliche Wissenschaft im Web 2.0

Today we had the last day of our blockseminar on Internet Measurement (see my post on day 1) and heard the following talks:

• nCap: Wire-speed Packet Capture and Transmission Improving Passive Packet Capture: Beyond Device Polling [Slides] [Student Paper] [Original Paper 1 Paper 2]
• DisCarte: A Disjunctive Internet Cartographer [Slides] [Student Paper] [My Blog Post] [Original Paper]

After completing the talks, we had a quite fruitful discussion on presentation style of every talk that will be hopefully helpful for the attending students. In general, I was really impressed by the quality of the talks. Some were really brilliant but all of them were unexpectedly good in general.

I really enjoyed this seminar and hope our advanced seminar on Internet Routing offered next term will be as interesting as this one.

Today was the first day of our two days blockseminar on Internet Measurement, in which I supervised two students. During the seminar, we addressed the following topics (papers) by talks held by students attending the seminar along with a discussion on the topic afterwards:

• Characterizing Files in the Modern Gnutella Network: A Measurement Study [Slides] [Student Paper] [Original Paper]
  Which files are shared on Gnutella and what are their characteristics? Besides studies that derived traces by hosting peers dedicated to provide measurement data, this paper describes data derived from crawls of the Gnutella network.
• Rarest First and Choke Algorithms Are Enough [Slides] [Student Paper] [Original Paper]
  This paper discusses why BitTorrent performs well and states that the Rarest First Algorithm and the Choke algorithm are enough to provide reasonable fairness, diversity of the content pieces and performance. Roughly speaking, Those are the key features that differentiate BitTorrent from other peer-to-peer file sharing protocols.
• Leveraging BitTorrent for End Host Measurements [Slides] [Student Paper] [Original Paper]
  How optimistic unchokes—provided by BitTorrent and essential for its functionality—can be exploited to perform end host measurements; a dedicated and modified BitTorrent client called BitProbes downloads two megabytes of data from peers—by acting as a freerider and not uploading downloaded data—and uses this communication for conducting host measurements.
  Some points that have been discussed: (1) the authors claim that downloading but not storing the data is enough to avoid legal issues. Is that really true? (2) During a sample 7 days crawl, the authors covered about 20% of the available autonomos systems (AS) in the Internet. What does this number mean? Is it a high coverage, or a low one? For the answe, one has to keep in mind that not all AS are likely to host BitTorrent clients (like enterprise networks).

• Unconstrained Endpoint Profiling (Googling the Internet) [Slides] [Student Paper] [Original Paper]
  How documents indexed by Google can be used to label IP addresses with applications run by a particular host
  The discussion mainly focused on the question whether the proposed method is really unconstrained as the title of the paper claims. Some key points: (1) The propsed method relys on Google, but the Google index varies (regional filtering etc.). (2) Existance of the deep web: not every available document is indexed by a particular search engine. (3) How dynamic are IP addresses? What if we want to label IPs of access providers which usually map to a set of users that used it in the past? (4) Can we trust data provided by the third parties (e.g. faked access log files etc.)?
  We agreed that this methodology seems good to discover trends but details have to be taken with a pinch of salt.
• I Tube, You Tube, Everybody Tubes: Analyzing the World’s Largest User Generated Content Video System [Slides] [Student Paper] [Original Paper1 Paper 2]
  What kind of videos are shared on YouTube and what is their access characteristics. See my blog post on this from October 2007.
• The Flattening Internet Topology: Natural Evolution, Unsightly Barnacles or Contrived Collapse? [Slides] [Student Paper] [Original Paper]
  This paper analyses a trend of big content provider building up WANs and tend to bypass Tier 1 providers to save transit costs and increase performance which flattens the Internet topology

For references to the original papers, the student papers (mostly in German) and slides, see the seminar webpage. The talks had a very high quality and the discussions were pretty interesting. So I’m really looking forward to day II.

We’re done with correcting and grading the exams; white smoke is escaping from our university building The results are quite ok with a failure rate of 20% (students which did not pass).

There is a positive correlation between the points earned in the exercises and the exam points (correlation coefficient 0.62); students which made an effort to solve the exercises carefully on their own (!) typically scored high in the exam.
The fact that the correlation coefficient is not close to 1 may be interpreted in two ways; (1) students are typically stressed under examination conditions and some may cope with those conditions better than others, and, (2) some students may not solved the exercises on their own and may have handed in solutions made by fellow students (we actually caught some students cheating).
General advice: do the exercises and do them carefully on your own! Think about the problems and ask questions in case you did not fully understand a particular concept. This is actually a truism every student may be aware of, but it turns out to be true once more.

In retrospect I have to say that it has been a nice course with good students. I enjoyed the time with them in the exercise sessions. Some handed in brilliant solutions that amazed me. Some amazed me by the progress they were making. Some stimulate one by the questions they ask. Few frustrated one when caught them cheating.
Big universities are often criticised for their anonymity; a student is just a number, a face in the crowd. As you can see, this is not entirely true.

However, when thinking back, here’s a list of some common misunderstandings I faced in the curse of the semester (maybe this helps when preparing for exams):

• Although ARP is used to establish a mapping of layer 3 (IP) to layer 2 (MAC) addresses, it is not used to lookup entries in a forwarding table (to which port do I have to forward packets for X and so on)
• half-open in the sense of TCP does not mean that a connection is open half of the time
• In contrast to MAC addresses, IP addresses are typically not rewritten when traversing a router (excluding the case of NAT)
• MAC addresses can be used only within a particular LAN segment. One cannot address hosts located in a different LAN.
• A common place for misunderstandings is the protocol stack: which protocol relys on which (e.g. ARP does not rely on UDP) and where is it located within the stack (e.g. DNS is no transport layer protocol)

There are probablly much more, but I don’t remember them right now…

The exam is over now and I have a good feeling by the feedback we got from students. It seems like that the problems were appropriate but the time was a little bit to short in order to solve everything. Moreover, no one returned the exam in the first n minutes because he was scarified by its problems, which is a good sign Only one student returned his exam before the time was over.

However, although we did extensive testing, we didn’t found all of the bugs that were in this exam. There were some very minor things that could confuse students at first place (e.g. there was a misleading host symbol, but reading the problem description should make things clear).

The only strange thing I noticed so far: one student did not put his identity (name etc.) on the exam, so he returned it anonymously. However, as he was the only one and we have a attendance list as well as sequentially assigned exam ID’s that point to the left and right neighbor along with seating arrangements, we could guess who the anonymous student is likely to be.

However, we realised that the proposed scheme of introducing exam IDs in order to parallelise the correction process breaks if we allow students to put solutions for multiple tasks on one single sheet of additional paper.

We now have to form teams doing the correction of a particular problem…

One semester draws to a close. We will have the Network Protocols and Architectures exam tomorrow, and, after one week of work on the exam, most of the things are prepared now. It is always amazing to see the significant amount of time that has to be put into designing and testing (!) a small set of reasonable questions. What are reasonable questions? Are they trivial or to hard to solve, given the knowledge obtained in the course? Are they solvable in the given time? Do they cover the main aspects of the lecture sufficiently? What is the usability of the exam; does one has to turn pages unnecessarily often in order to solve a problem, are important things sufficiently highlighted, is the text easy to read/understand, …? Is every unwanted ambiguity removed that would make the correction unnecessarily complicated by allowing a too wide variety of possible answers? Does the English translation of the problem description map 1:1 to the German original? There are a lot of things to consider when designing exams.

In order to guarantee the quality of the exam, we had a iterative approach; two people were mainly responsible for designing and typesetting the exam, two senior people were reviewing the problems, others were solving it while taking the time. After having some iterations of this procedure, I think that we have a pretty stable exam now.

Moreover, I introduced the concept of exam IDs; each exam will have a unique id that is printed on all sheets that will be part of the grading. There will be a cover sheet that contains the ID and a form where the student can provide his personal data. Besides the cover sheet, there won’t be any personal date (like a name) on any other sheet. This will allow us to do two things: 1) exams can be corrected in an anonymised way; the corrector does not know about the identity of the student and thus cannot be biased. 2) the correction process can be easily parallelized by forming groups that correct only one specific problem of the exam without having the student to write his name on every sheet that we provide. The ID can be easily printed in advance, but it’s much harder with names as handing out the exams before the exam starts would take much longer then.
Drawbacks: printing takes much longer (I needed to write shell scripts that replace predefined exam ID tags in the postscript code with the actual ID and sent them to the printer, etc.). Moreover, the exam parts, corrected by different groups, have to be combined at the end. But this will scale with the amount of people correcting it.

After having spend like one entire week on preparing the exam, I’m now really excited about its result. Will the students have problems with the task? Is it really not too hard to solve? Will there be any complications? Lets see. I think I’m as exited as the students