Using WiFi in yet another context gave us the idea that given the enormous shipped volume of WiFi chips and the resulting cost reduction, and the huge knowledge base for embedding WiFi, it may well become much more of a generally applicable wireless system, like Ethernet has become for wired systems. Just as Ethernet continually increased its speed to 10Gbps and beyond, used for both wired and optical systems over increasing distances, so is WiFi providing wireless broadband access well beyond its original context of short range (home and office) networks. Long distance rural systems and meshes come to mind.

Even more interesting is the reality that in the off-load application, WiFi is also a “disruptive” technology to the backhaul business. Through off-load, WiFi will reduce traffic demand on backhaul capacity just as new, high density 60GHz and E-Band backhaul systems start deploying. This will emerge because even though WiFi has lower performance (bandwidth), it has very much lower costs. Thus, it is disruptive in the sense defined by Harvard Prof. Clayton Christiansen who first analyzed the disruptive phenomenon. Disruptive technologies provide good-enough quality at really low prices. MP3 immediately exemplifies this idea.

In this regard, around 1993 I had a very odd conversation with a friend who had already been an IEEE Fellow for many years. He was a major force in coding theory and had made serious money in a startup that went IPO. He had also written a few books on communications theory, including a giant tome on spread spectrrum. He said he thought most research topics in comm theory were at their limits. The only real problems, he continued, were implementation. (Of course, he failed to predict Turbo Codes, or even MIMO…but who’s counting?)

I was very surprised to hear this from such a creative guy. And I even thought about how it was once said (around 1900) that the patent office wasn’t needed anymore. So consider me a total optimist about creative research. My optimism has just been boosted by a recent paper discussing the death of PHY research!

The article reviews the state of comm theory and shows how in almost every aspect, the theoretical limits defined by Shannon have been achieved, at least for single cell systems. The authors (see IEEE reference below), claim only these fields need more single cell PHY research:

• Design of Short Codes
• New Coding Paradigms (beyond LDPC and turbo….e.g., fountain codes)
• Implementation Impairments (e.g., nonlinearities in power amplifiers, phase noise,…)

The authors go on to list important system problems clearly needing basic PHY research, especially for multi-cell systems. For example, Relaying and Femtocell architectures. But now, they say, it seems clear that research about single cell PHY layer behavior is so well understood that research may be un-needed, except in certain narrow areas as named above.

However, to add to the research potential of the PHY, consider this theoretical problem for single cell PHYs that has never been formally solved:

Find the optimum Bayes receiver to minimize BER, subject to an implementation complexity constraint. (Recall that the optimum Bayes maximum likelihood receiver minimizes BER without constraints).

So, maybe there is still some theoretical life left in the PHY!

REFERENCE:

M. Dohler, R. W. Heath, A. Lozano, C. B. Papadias, R. A. Valenzuela, “Is the PHY Layer Dead?”, IEEE Communications Magazine, APRIL 2011, v.49, no.4, pp159-165

Remember the wild days when “expert systems (ESs)” would replace your family physician or run a nuclear power plant? Remember how Campbell soup made a big ,splash by describing how ES was replacing the firm’s 80 year old uber-soup taster? Well I admit, I remember. I remember the “artificial Intelligence Company”.Where is the AI company now? Long gone, faded into general engineering practice.

I started thinking about all this a while back, but at our last WCA panel on “Commercial Status of CR and SDR”, on 11-9-10, it became obvious. There is no commercial system that declares itself as CR or SDR, but these technology methods and analyses are used in all the major standards!

By the way, there also seems to be negligible work going on, at least to public knowledge, on IEEE 802.22, which is the famous “white space” system.

So, in true engineering empirical fashion, we did an experiment-on Google Scholar. We examined four important technologies (CR, MIMO, OFDM, Turbo Codes), all of which have had their  major work within the last 20 years. The idea was to see how rapidly the number of articles and patents is growing, and compare the results. By getting a reasonable idea of the number of articles and patents generated by year, we can get an idea of the activity levels and see the growth rates.

Of course, some of these ideas have been worked on much longer than others, so the total number of articles and patents cannot be fairly compared. Here’s a rough indication of the age of these technologies. It will important to keep this in mind below.

CR emerged with Joe Mitola’s Ph.D. thesis in 1999 [see, e.g., G. Maguire and J. Mitola, “Cognitive Radio: Making PCS Personal”, IEEE PCS Magazine, August 99].

Turbo Codes were first described in a paper:  C. Berrou, et. al., “Near Shannon Limit Error Correcting Coding and Decoding: Turbo Codes”, Proc. IEEE Intl. Conf. Commun., pp 54-58, May, 1993

MIMO gained major attention with a paper by G. J. Foschini, “Layered Space-Time Architecture for Wireless Communication in Fading Environments when Using Multiple Antennas”, Bell System Technical Journal, pp41-59, Autumn, 1996

OFDM, the core of WiMax and LTE systems, is  an old technology, but it got a major boost with J.S. Chow, J.C. Tu, J. M. Cioffi, “A Discrete Multitone Transceiver System for HDSL Applications,”  IEEE  J. Sel. Areas Commun., pp 895-908, August 1991. Prof. Cioffi later formed a company Amati Communications which was acquired by Texas Instruments for around $1B in the 1990s.

Google Scholar provides, for a given search phrase, the number of articles and patents found since a year Y designated by the searcher. The earliest Y you could use is 1991. In our study, we simply recorded the number of articles and patents found now by Google Scholar since each year starting with 2002, which implicitly includes all history for that search. By tabulating these search results, we deduced the approximate number of articles and patents added each year for each technology. We note that this addition by year sometimes is negative, probably indicating fewer articles found in that year. But the variations due to negative increases were small.

On the other hand, it’s obvious we cannot compare the numbers found per year, since the scales are so different (ranging from 100s to thousands of articles and patents found). So we just normalized the search results by the number in our starting year for display. The normalization enables us to graphically compare the publications in the four technology ares and see which is growing fastest.

The raw search data and a graphic showing the rapid growth rate in CR research are shown below:

Raw Search Data

Growth Rates of Major Technology Research

The number N shown on the graph represents what Google calls the total number of articles and search results found “anytime” for the search phrase. This reflects the size of the knowledge base written about each technology. The size of CR knowledge is relatively small, but is growing incredibly fast.

It surely seems that CR research is rapidly growing compared to the three other more established, but also hugely important fields.

Download “Why is MIMO the Next Big Thing?“