The other day I happened to take a look at the electronic version on the IEEEXplore of the first issue of the IEEE Transactions on Information Theory. There, I noticed something that I think has been lost in current journal publications and that I have been planning for this blog. What I found, was constructive research discussion and argumentation. What I mean is the following: take for example one of the papers in the issue, written by no other than C. Shannon. In the same issue, along with the paper, you can find a review by 6 readers and the ensuing answers from Shannon. This is not exclusive to Shannon’s paper; in fact, every single paper in the issue is treated in the same way. The end result is an issue extremely instructive, dynamic and entertaining.

Today’s reality is that published research basically goes through a peer review process where it seems to me that most of the constructive and interactive discussion is tightly constrained within the authors-reviewers-AE interaction. In this process, some of the constructive comments eventually are incorporated but the later process of argumentation and intellectual exchange between the authors and the interested readers are all but missing. Why has this happened? Well, I don’t think I have the answer but I do think that the volume of submissions compounded with an obvious limitation on how much can fit in a particular journal issue have translated in long publication cycles and space being only available for new contributions. I believe that this process has removed perhaps one of the most important steps in creative research.

Some may say that conferences are the right venue for this kind of interaction but many of you who have been at conferences would agree that the audience of such interaction is limited to the relative (to the eventual number of readers of a paper) few attendees to a particular session and, even more, in general the discussions are limited to a tight timing which frequently forces the session chair to “move the conversation offline”, where it really becomes private.

In reality, I think that the answer is in the technology we have helped developed. We can first settle with the reality that printed journals have some inherent characteristics that make intellectual debate on any paper very difficult. At the same time, we can embrace the fact that some recent changes are certainly helpful. By this I mean the electronic management of the submission and revision process, as well as the electronic publication. With this, there is only one element missing, and this is a dynamic, flexible and public medium that facilitates and hosts the debate.

I think the answer is right here, in a blog. A blog can catalyze discussion by initially posting a report, a comment or a review of a paper. Then, the blog is indented for open discussion where everybody can post as comments their own contribution to the debate. In fact, even anonymity of the person posting a comment could be maintained since anybody can use an alias. Those hiding behind an alias to post erroneous or ill intentioned post will be sorted out as a natural process because the posts are public.

Whit this said, I will try to make the debate of DSP ideas an important component of this blog. I plan to start doing so by selecting published papers, commenting on them and opening the debate to anybody interested. As always, I welcome any suggestion with regards to what to cover or how to approach it. I’m aware that for this idea to work I need a critical mass of readers and comment posters and that achieving this is a long journey, but any long journey starts with a single step. Besides, I truly believe that this could turn into a valuable medium to advance DSP technology that is worth trying.

Posted by mobile phone:
Since wireless technology owes so much to DSP, I thought that it was just fiting to create a mobile version of this blog. So, from now on, and thanks to the WordPress Mobile Plugin it will be possible to follow the Digital Signal Processing Blog from any web capable mobile phone. Enjoy it!

For some reason, the other day I remembered a presentation by Rajiv Laroia about 2 years ago. Rajiv is know to be very successful in applying DSP technology to solve tough communications problems. The presentation was about one of his inventions: Fhash-OFDM, one of the preeminent OFDMA technologies. The presentation covered the many advantages of this technology when applied to broadband wireless communication. Towards the end, I was convinced of the advantages of OFDM but I was wondering why an apparently simple technology hadn’t made it to the market earlier. The answer was that although the theory behind the system was known for a while, the technology to implement it hasn’t be ready until recently.

This got me thinking on how DSP has historically evolved over time and, specifically, the history of the FFT, the technology at the core of OFDM. It turns out that a little search on the web resulted in some interesting sites, such as the expected Wikipedia entry, another Wikipedia entry for the Cooley-Tukey FFT algorithm, one in the CS department at Dartmouth College, and one in the department of mechanical engineering at Oregon State University. All of them resulted in some interesting reading that helps to realize how a mathematical and engineering problem evolve over time into becoming an important tool of countless products. Also, for my personal train of thought, I was surprised to learn that the FFT can be traced back to Carl Friedrich Gauss some 180 years before the first DSPs appeared on the market. Nevertheless, Gauss’ contribution went almost unnoticed (not surprisingly , since it was published posthumously and in neo-Latin…) and was rediscovered in 1965 by Cooley and Tukey, about 15 years before the first DSPs appeared on the market. So, the timeline for the OFDM goes by the development of the FFT in 1865, following by the first DSPs in the late 70’s and early 80’s. After this, it took around 20 more years for the DSPs to be powerful enough and for other techniques (such as frequency synchronization ) to develop enough so OFDM becomes a reality. But this is a topic for another posting…

One of the reasons why I started this blog is that there are not many comprehensive web sites exclusively dedicated to DSP. One exception to this is the DSPrelated web site (http://www.dsprelated.com/). This site offers many useful resources such as the discussion groups, a web interface to the comp.dsp usenet news group, a sizeble list of DSP related links, books reviews, etc. Certainly, it is a site worth visiting regularly.

I have recently come across a number of exciting news and links that provide more sources to learn and do DSP-related research.

First I learned that many journals from EURASIP (The European Association for Signal and Image Processing) have started to offer their material under the “open access” model. One of these is the EURASIP Journal on Applied Signal Processing, which publishes papers aimed at both practicing engineers and academic researchers in topics such as communications, networking, sensors, radar and sonar, medical imaging, biomedical applications, astronomy, etc. These are great news because now everybody could access the papers here, which many are of really good quality (and I’m not saying this simply because I’ve been a reviewer and author for this publication, but because I do believe this is the case).

Since I’m talking about open access publications, one site that is worth checking from time to time is the Text Revolution web site, which provides links to textbooks that can be read online for free. DSP-related books can be found under the category of engineering or math.

Finally, Springer has recently announced a new journal due to come out in March 2007 with 4 issues per year. The journal title is “Signal, Image and Video Processing” and the Editor-in-Chief is Prof. Murat Kunt from EPFL / STI / ITS in Lausanne, Switzerland. Topics that will be cover include adaptive processing, biomedical signal processing, multimedia signal processing, communication signal processing, non-linear signal processing, array processing, statistical signal processing, modeling, filtering, multi-resolution, segmentation, coding, restoration, enhancement, storage and retrieval, colour and multi-spectral processing, scanning, displaying, motion detection and estimation, stereoscopic processing, etc. Papers may be published with open access at the choice of the authors.

It has been a very long time since my last post, let’s just say that the time demands of modern life have not been very kind to me lately… In any case, good news is that the Holidays are giving me that extra time to catch up with this blog.

Some weeks ago I intended to write about an interesting technology to store information that is heavily based on DSP technologies. The idea behind the technology is quite interesting and a nice example of outside-the-box thinking. In a nutshell, the idea is to store information on a piece of paper by using color coding. More interestingly, since I read the first article about this, there have been more discussions about the validity and practicality of this idea.

From my point of view, the discussion is itself interesting because it brings up a lot of issues related to several DSP technologies. From the creative standpoint, I believe that although there are many challenges with this technology in its current form, who knows where it can end and where the related DSP problems may take us to? I guess, it’s about the journey, not the destination.

<p class=”MsoNormal”>The deadline (September 29, 2006) is rapidly approaching to submit papers to the most important conference in signal processing: the <a href=”http://www.icassp2007.org/”>International Conference on Acoustics, Speech, and Signal Processing (ICASSP)</a>. Next year the conference will be held between April 15 and 20 at the <a href=”http://www.hawaiiconvention.com/”>Hawai’i Convention Center</a> in beautiful <a href=”http://www.co.honolulu.hi.us/”>Honolulu</a>. Based on the many exiting new developments going on in DSP, and the location, next year’s ICASSP already promises to be a success in number and quality of submitted papers, tutorials and special sessions. For more and up-to-date information check the official <a href=”http://www.icassp2007.org/CallForPapers.asp”>Call-for-Papers</a> website.</p>
<p><a href=”http://www.feedbase.net/Results.php” /></p>

The IEEE Signal Processing Society has recently announced the creation of a new research journal named “IEEE Journal of Selected Topics in Signal Processing“. By its name, I can imagine that this journal has been and will follow roughly the same path as the IEEE Journal of Selected Areas in Communications (JSAC) but specialized to signal processing topics. I guess that there may be some overlap between the two journals in topics related to applications of signal processing to communications, but this should work out just fine since there is always a lot of research and lots of new developments going on in this area. The Editor-in-Chief for the new journal is Prof. A. Lee Swindlehurst, who will be complemented by a great Editorial Board. The goal for the Editorial Board is to publish four issues in 2007, following by six per year after that, with an ultimate goal of monthly publication. There are already two active Call-for-Papers on the website.

I have recently learned about “Connexions”, which is essentially an online repository of notes and learning material of different subjects, but with a lot of content focused on DSP. What makes this site so valuable is that it follows the “open access” philosophy, which means that all the information is free (but belonging to an author that has to receive credit for his/her work), and intended to spur collaboration and further contribution. Besides the learning material, this site is also interesting in how the Internet is used intelligently to promote collaboration as well as the use of new tools and technologies (such as MathML).