CptS/EE 562 Literature Survey Assignment

Spring, 2004

10% of final grade

Bottom Line

Assigned: Thursday February 26, 2004

Subject Chosen and approved by instructor: Thursday, March 11 (lecture after midterm, right before spring break).  I want to make sure we don’t have overlap.

Due: Tuesday, March 23, 2004 in class (firm deadline).  Hand in hardcopy to class, and also email the instructor a copy (.pdf or .doc or .dvi preferred) too. 

Note: Student presentations will start March 23, and I before spring break will ask one or two of you to present that week.

Overview

In this assignment you will apply what you have learned in class to analyze 5-7 research papers (“primary papers”) in a related area.  You will also skim 3-5 papers (“secondary papers”) cited in these primary papers to get a better feel for the topic, and summarize them in a paragraph or so each.  The basic idea is to analyze and summarize the state of the art in this field.

In all places in your report, you are expected to use technical terms carefully, as discussed in class and in Dependable Computing and Fault Tolerance: Concepts and Terminology.  You should also be careful to have your citations in standard citation formats, and include URLS when they are available.

You will also give a presentation later in class, and it will be best to stick with the topic from your literature survey.  We will discuss this later in class.

Topic

You will choose your own topic and find the appropriate papers.  If the topic is too broad to be adequately covered in the paper numbers discussed above, then focus in on one aspect of it.  If it is too narrow, generalize the topic.

Organization of Your Report

Your report should be organized as follows:

  • Abstract: a 150-200 word summary summarizing the technology/issues/problems (e.g., Byzantine quorum systems, high-performance multicast, etc) and summarizing the state of the art in that area.   The first words of your abstract should be: “This report summarizes the state of the art in XXX as represented by papers [X,Y,Z, …]”, where XXX is the technology/issue/problem and [X], [Y],  [Z], … are references below.
  • Overview of Technical Issues: a half or full page explaining the technology and the issues involved in the research.  A figure is appropriate but optional (probably half of the students’ reports can use one, half not).
  • Paper summaries: discuss the results, limitations, and strengths of each paper.  2 to 3 pages per paper.
  • Analysis: 2 to 4 pages comparing and contrasting the papers reviewed.  You will be expected to apply good judgement and your knowledge of fault tolerance, not just rehashing what the papers say.  For example, your judgement about if the papers are making good points or are arguing about a “distinction without a difference”.
  • Conclusions: 100-300 words summarizing the report.
  • References: all papers mentioned in standard citation form.  I prefer the format [JKZ93] for something with authors’ last names starting with ‘J’ and ‘K’ and ‘Z’ and year of publication 1993, not the [1] format (I have a horrible memory so I need the help remembering which paper you are citing when reading your reports).  If there are more than 3 authors, do something like [JKZ+93].  The citation of course has the full author list, but the “label” for it (e.g., “[JZK+93]”) only has the first 3 authors’ initials, so the References section looks aesthetically pleasing!

You can cite the textbook too if it provides useful background on a specific issue, but include the section number probably to be helpful.   This cannot be a primary source, and you should have at most one textbook among your secondary sources.

If you cite any web pages as the source, unless an author is listed, use the name of the organization (BBN Corp, UCSB for UC Sanata Barbara, etc), then the title of the web page, then it URL.  Assume the date of publication is 2004, unless it says last updated in 2003 or something like that.

Include in the reference the URL for the paper if it is not in one of the major conferences or journals.  Say “Online: URL.” where of course “URL” is the actual URL.

A sample format will be emailed to the class in .doc format at least 2 weeks before the due date.

Note that if you have the commercial version of Adobe Acrobat (not the free reader) then you can extract an image from a .pdf paper.  This can be very handy for adding a figure from one of your papers (with attribution, of course).  If you do not have access to Acrobat, I can arrange this in my research lab.

Grading

Grading criteria will be discussed later.  We are trying to survey the state of the art, and concentrating on papers in the last few years.  Therefore, if I find a key paper in any of the sources listed below that is in the last few years, and you did not find it, and it is more relevant than one of the papers you chose as primary, then your grade will suffer.

Paper Sources

The sources below have two kinds of links: one for the conference or workshop itself (which usually contains the program, including the paper titles), and also the source for the conference papers (ACM or IEEE).  Some conferences don’t have online sources (e.g., expensive publishers like Springer and Wiley), but you can almost always find the paper online with google or looking at the author’s web pages.

The WSU library has subscriptions for the online publications for the IEEE and ACM.  They will not likely include most conferences and workshops in the last 6-12 months, so you will have to search for the paper like a Wiley or Springer paper.

(In general, if one cannot find a paper in their library or online, it is standard practice to email the author, and they are obliged as a matter of professional courtesy to send you an electronic or paper copy.  Few authors mind this, and zero of the grad student authors!  If you cannot, then the WSU library’s inter-library loan system can probably get you a copy, but I do not know how long it will take.)

We are focusing on papers published in the last 2-3 years, but if there is a need to include in your paper set a paper from as early as 1995, that is OK.  Below are sources for the last 2-3 years only.  Most conferences and workshops are every year at the same time each year (within a week), but some are every 18 or 24 months.

Most conferences are IEEE, and their overall paper web page is here.  You can look at tables of contents, abstracts, etc but have to have the IEEE library configuration/password to access the full papers.  The WSU library has a subscription to this.  Also, note that the conferences often have associated one-day or half-day workshops or “Fast Abstracts” or “Works in Progress” sessions that have shorter papers online that may be of interest (but do not count as a full paper in terms of your review number of papers).

Before I go into the list, note that a HUGE source of finding related work is CiteSeer.  It probably won’t be needed for you on this project, but it can tell you who has cited a paper, which might give you an idea of related work.

Top-Tier Conference and Workshops Completely or Mostly on Fault Tolerance

The International Conference on Dependable Systems and Networks (DSN), IEEE/IFIP (Note: FTCS and DCCA merged to become DSN a few years back.)

Some Middle-Tier  or Other Conference and Workshops Completely or Mostly on Fault Tolerance

Symposium on Reliable Distributed Systems (SRDS), IEEE

Workshop on Object-Oriented Real-Time Dependable Systems (WORDS), IEEE

Top-Tier Conferences and Workshops Often with Some Fault Tolerance

International Conference on Distributed Computing Systems (ICDCS), IEEE

International Conference on Distributed Systems Platforms (Middleware), IFIP/ACM. Note this is not every year.

  • Middleware 2003 (conference web site),  papers not online but I might have the hardcopy if you cannot find on the author’s web page.
  • Middleware 2001 (conference web site), papers not online but I have the hardcopy if you cannot find on the author’s web page.  Hey, my doctoral advisor’s (Schlichting) paper got best paper award!)
  • Middleware 2000 (conference web site, papers not online but I have the hardcopy if you cannot find on the author’s web page)
  • Middleware 1998 (conference web site, papers not online but I think I have the hardcopy if you cannot find on the author’s web page)

Some Middle-Tier or Other Conference and Workshops with Some Fault Tolerance

International Symposium on Distributed Objects & Applications

Journals

The one (brand new) journal that is appropriate is

There are really no other journals dedicated to fault tolerance, other than the IEEE Transactions on Reliability which has almost nothing to do with fault tolerant distributed systems.  Further, unlike many other fields, experimental “systems software” researchers do not often bother with journal publications, because of their long lead times (1-3 years), so conferences are by far the preferred way to have “impact” with your research.  To say that for experimental systems programmers journals are “almost a joke” is a bit too strong, and would offend some from other fields (or those that run the journals), so I won’t!  They do have the one strong virtue of no (hard) page limitations.

Still, some top-tier journals that occasionally have fault tolerance papers in them are:

Some Example Paper Sets

I may provide a few example paper sets the Tuesday after the midterm exam, but I might not (I have not decided yet).  But part of this exercise is for you to learn how to find papers in recent research literature from a given topic in fault tolerance.  So more likely I will just provide some suggested subjects, likely that Tuesday.

The best way to do this is to read through the tables of contents in the top-tier conferences for the last 3 or so years.  Then take it from there…