Is research in Cognitive Radio useless?

It's been a long time since my last post. In my defense I have to say that it was just because I entered the wormhole of "writing a PhD thesis". Now I'm almost finished, and since there are only a few days left before ICASSP 2011 (I will be posting from Prague) I want to catch up with the blog.

A couple of days ago, Volkan, a first year PhD student, wrote a shrewd comment in the blog. And since it raises some interesting issues I decided to answer it in a separate post. Here the comment and my response.

"Hi Gonzalo, I just have found your blog and its really informative and helpful. I study CRs as well, and this is my first time getting introduced to the internals of the wireless communication technologies.

The more I study about the history of wireless standards and its evolving nature, the less I believe in the future of CRs. (No offense! I need to fulfil the requirements of a PhD about CRs as well.) Let me put it this way: There are various technologies (IEEE 802.11, 802.15.1, 802.15.4, 802.16, 802.22, etc.) working on the ISM band in an uncoordinated and non-cooperative manner. Consider a case where there are 3 WiFi APs operating in the ISM band and we all know that they are almost oblivious to each other and incapable of fully utilizing the available ~80MHz of spectrum. You might say that there are various industrial WiFi providers providing a centrally managed network of WiFi APs for such scenario. Yes, there are, but they don't work that well as advertised. Think about what happened in the presentation of iPhone 4 in 2010. ("Because there are 570 Wi-Fi base stations operating in this room. We can't deal with that." -- Steve Jobs) At TechCrunch 2008, RailsConf 2010, Web 2.0 Expo 2010. These giant conferences and public events are equipped with centrally managed and professionally placed >500 WiFi APs, and they just didn't work. On the other hand, take look at GSM. We are not capable of making a stupid WiFi network work in ~80MHz, but a GSM operator is able to serve "millions" of its customers in just ~25MHz.

To sum up, my prevision is that: When we manage to finalize a standard for CRs and in the progress of introducing it into the market, GSM will be serving up to %90 of the internet connectivity market in a much more effective and successful manner. (Please, make me wrong.) It just seems to me that we are trying to clean up the mess of a bunch of uncoordinated big-brother company activities, which will just inevitably get lost in the dusty pages of the history in a near future. Here is another example: For decades people worked on sensing and now FCC tell us that let's forget about that sensing and use a GPS database. Sorry but.. WTF! I really would like to hear your opinions on these matters."

First I would like to thank Volkan for sharing his opinion. In fact, I have to say that I agree with him in several points. For example, I believe that Wifi and probably CR based standards will never be able to completely substitute centralized GSM-like technology (in fact I even believe that GSM is already serving the 90% of the mobile internet connectivity ;) ). And so much effort has been put into developing a true dynamic spectrum access and the first approach is just to check a database with information about vacant channels? Completely agree.

However, I think that Volkan's view about the services a cognitive radio network can provide is quite limited. While Bluetooth/Wifi/Wimax cannot deal with certain situations (as he pointed out), they showed to be a great success in offering certain services that otherwise would not exist. In most of the homes there exists today at least one Wifi hotspot connecting different devices, such as TV, printer, laptop… or when you get into a car equipped with a Bluetooth headset the mobile phone automatically recognizes it and gets connected. My point here is that none of these (and many other) services would be possible by using a standard centralized GSM network. Moreover, while I do not have the actual data I would bet that just Wifi networks worldwide move more information than the sum of all the 2G-3G data. That is the power of decentralized / low range communications. I believe that the availability of additional spectral resources, such as opening certain licensed bands, will create an environment in which other unexpected services can be born.

And why requiring a geolocated database? Let’s think a little bit about the reason for that: just time. The first studies on the spectral efficiency of conventional static access schemes date back before 2002 (see e.g., “Report of the spectrum efficiency working group”, FCC, 2002). After that the FCC showed interest in allowing a dynamic spectrum access in certain licensed bands… and almost 10 years after, this access scheme was not implemented yet. Even worse, other countries catching up the initial advantage of US in this matter. That is, the FCC is now in a hurry to have something in the market as soon as possible, hence the database aided approach.

To sum up, on the one hand I believe that CR will facilitate a series of new services/applications, and on the other, I hope that future versions of CR technology will incorporate sensing as a requirement… even if it is only to give some sense to my past research ;)