24 Mar 2010

Future of Cognitive Radio - an interview with Dr Joseph Mitola

Our editor, Ian Poole, talks to Dr Joseph Mitola III about his input to Cognitive Radio and how he sees its development in the future.

The name of Dr Joseph Mitola III is virtually synonymous with Cognitive Radio - an area of technology that is rapidly growing in importance with the need to utilise the available spectrum more efficiently and the need to provide more effective communications.

Joe has long been interested in radio, and in particular software defined radio technology. He coined the term "Cognitive Radio" in an email exchange between himself, Jens Zander and Gerald Q. ("Chip") Maguire of KTH and first presented publicly in his Licentiate lecture at KTH, The Royal Institute of Technology in 1998. Later in his doctoral dissertation, Joe developed the idea further describing Cognitive Radio as: "The point in which wireless personal digital assistants (PDAs) and the related networks are sufficiently computationally intelligent about radio resources and related computer-to-computer communications to detect user communications needs as a function of use context, and to provide radio resources and wireless services most appropriate to those needs."

Having had an interest in radio since the age of 11 when he built his first radio, he moved on to take his interest further as a career, making significant advances in the field of the Software Defined Radio, this provided the basis for his development of the idea of Cognitive Radio.

We have heard that you coined the name Cognitive Radio while reading for your Licentiate at KTH, but how did you actually come to develop the idea for Cognitive Radio - can you tell us a little more about the background?

When speaking at an IEEE military communications conference (MILCOM) in 1997 where I was complaining about how complicated software radios were becoming with the 42 air interfaces planned for the US Joint Tactical Radio System (JTRS) program, suggesting we need a user interface with some artificial intelligence (AI) in there for the user or suffer the "blinking VCR" syndrome where video cassette recorders had become so complicated that nobody could figure out how to set the time, so finally you just slide a potted plant in front of the blinking display. Chip introduced himself to me after the talk since he agreed with me and his lab at KTH in fact was inserting AI into radio. Nearly by accident a few months later, I visited Chip at KTH and when he invited me to do a post-doc with him, I said I'd love to but I had no PhD. "We can fix that" he said with a big smile. Three years of "post-doc research" later KTH awarded me the PhD in teleinformatics for cognitive radio.

What are the main ways in which Cognitive Radio is currently being used and what are the main benefits you see?

The US Defence Information Systems Agency (DISA) is very lucky to have Ms. Paige Atkins leading their Defence Spectrum Organization (DSO). Under her direction, DoD is moving into the age of dynamic spectrum access (DSA), which is the key near-term contribution of cognitive radio. Industry leaders like Nokia have adopted cognitive radio within allocated spectrum bands to more effectively manage heterogeneous spectrum and femtocell dynamics, moving network intelligence to the radios to reduce network overhead while substantially improving user experience for the more innovative integrated telecommunications (cable/ fibre TV, telephone, Internet Service, and wireless) service providers. Soon, public safety will benefit from a combination of DSA and secure heterogeneous network management.

Initially, the main focus on Cognitive Radio (CR) was for it to adapt itself to provide the optimum communications channel, now there is a focus on using Cognitive Radio to enable much more efficient use of the spectrum. Can you tell us more about this?

Well, my own CR focus was on enhancing the user experience at low cost via machine learning embedded into the PDA. My first IEEE paper on CR showed how to use this embedded computational intelligence for short term opportunistic spectrum rentals with etiquette to legacy users whose radios were not cognitive. Regulatory time scales have limited the deployment of this technology, but the TV white-space phenomenon in the US is moving this kind of DSA forward now. Conventional "lanes in the road" spectrum regulation leaves many space-time holes in spectrum, particularly in urban settings where buildings cause tower blind spots that can be employed for short range low power DSA causing zero interference to the primary spectrum user. The IEEE 1900.5 spectrum use policy language group now is including spatial maps that hopefully will evolve to context-aware space-time maps with machine learning to reduce the time and effort to keep these maps current. For example, sporting events and emergencies distort P1900.5 radio environment maps, but the policy community has not yet reflected this level of adaptability. Crawl walk run is the process and we are beginning to walk but not yet run.

One of the key issues with many Cognitive Radio applications is that of spectrum sensing. What are the key challenges you see regarding that?

There are no key challenges in spectrum sensing at the physical layer. Nearly all commercial and public safety radios sense channel occupancy within their allocated spectrum bands, and this is sufficient for DSA, so sensing per se is not the issue. However, making sense out of what is sensed is another story. Although today most mobile radios include GPS or some other location estimation technology and all include clocks, relatively few such radios or networks integrate location and time with spectrum sensing to learn space-time maps. Professor Hong Man for example, is doing that at Stevens Institute of Technology with a type of machine learning called reinforcement learning, and we are hybridizing this with other learning and inference techniques to address the problems of unusual events such as emergency situations.

Conventional cellular networks keep track of channel occupancy by tower, but with 4G rates of > 100 Mbps, these networks will have to develop much finer scale maps of space-time occupancy and user movement within each MIMO footprint and among MIMO towers and femtocells. So the integration of space-time-spectrum sensor data over time into user behaviour trajectories at low cost is the key challenge that I see in the near term.

What are the key technology enablers that are allowing the development of CR to move forwards at the moment?

The DSA community are moving forward on multiple fronts at once. Professor Yingying Chen of Stevens developed some of the basic science of spatial reasoning at Rutgers ORBIT facility, so at Stevens she is enhancing the science into engineering location awareness in real world systems which are substantially more complex and dynamic than the relatively pristine academic laboratories like ORBIT. The costs continue to drop for RF hardware with useful bandwidth that is 10% of carrier frequency.

This enables greater use of DSA across adjacent bands to leverage the physics of radio devices for greater spectrum efficiency. Heterogeneous networks also are developing quickly with a Stevens spin-out company able to aggregate the network capacity of four bands at once (e.g. WiFi, GSM, CDMA, PCS) at the network layer. Each wireless band/mode carries typical traffic, but the user gets the experience of the best of the bands or of all four at once if all four happen to be available. The enhanced user experience of cognitive band aggregation will create competitive pressures in the marketplace that will accelerate CR deployments.

What do you feel are the key issues that currently need to be overcome with Cognitive Radio at the moment?

Since major information systems market players have discovered the potential value of TV white space, the major issues have to do with market structure and not technology per se. The FCC spectrum occupancy database ruling removed regulatory barriers, so the key issues now have to do with cognitive networks, particularly of the integration of space-time-RF databases into cognitive networks. Stevens professor: R. Chandramouli (Mouli) founded and chairs the IEEE Technical Committee on Cognitive Networks (TCCN), and this is where the key network research is taking place. Security is the biggest issue after that, I believe. Although my fundamental research provides the theory on how a cognitive radio can be inherently secure, including smart enough to examine uploads to accept trustable updates and to reject certified images that are potentially malicious, the foundations have yet to be embraced by the SDR-CR community. This is an issue of vertical market integration across IT infrastructure, computing, and communications, so it is not purely an SDR-CR issue.

What new applications are being investigated for CR - what might we see in the very near future?

There is a long list of high value use cases in my second book, Cognitive Radio Architecture (Wiley 06) that have yet to be realized in practice. In that book, I looked ahead ten years from 2005 to the possibility of self-extending machine learning. Starting with fundamental results from the breakthroughs of Doug Lenat, of heuristic discovery, and Randy Davis, of learning of new concepts mediated by database schema, I develop the basic theory of how cognitive radios and cognitive wireless networks (CWNs) can learn to extend themselves and to autonomously tailor services for the user.

Today, there are about 3,000 parameters in a CDMA network to be adjusted in order to control network behaviour. There will be hundreds of such parameters for practical urban cognitive radio networks, where the automatic tuning of these parameters by the CRs and CWNs will reduce cost and complexity to deploy CWNs. Embedding user-awareness and machine learning will complete the transition through radio-centric learning to the CR's mass customization for enhanced user experience as developed in the book.

Are there any technologies that need to be developed further - their performance improved - to enable these new applications to be realised?

The basic technology of exascale networks is only embryonic today. The Internet, the largest ant, bacteria, and algae populations, and the human central nervous system (CNS) are the significant exascale networks in existence today. The theory of such networks has not been ignored, but our understanding of such complex adaptive systems (CAS) like ant colonies remains embryonic. Our knowledge of nano-scale phenomena in the CNS such as molecular signalling pathways is growing very rapidly, but CAS research has not translated into new systems engineering methods, processes, or tools to address exascale networks, particularly not trustable or militarily secure exascale networks. So I'm leading an interdisciplinary team to create the academic foundations of self-aware exascale systems that learn, but that also remain stable and secure.

Looking further into the future, where do you feel Cognitive Radio technology might ultimately lead?

I think cognitive radio already is going the way of the buggy whip. We still have buggy whips, but they do not have a significant role in transportation systems. Similarly, a decade ago when I formulated and published cognitive radio, that question really was important. During this past decade cognitive radio morphed into DSA, adaptive networks, and heterogeneous networks as communications enablers.

At the same time, user awareness, context aware services, and malleable user interfaces are embracing my vision of embedded cognition supporting the user, simplifying the user's life, embracing communications and computing, fixed and mobile, with fibre cores extending to wearable fashion statements. So as cognition becomes ubiquitous and integrated into everything from home appliances to automobiles, the excitement will morph from cognitive radio to the sentient spaces that such smart devices enable. Those who keep thinking of cognitive radio as if it were an end in itself will go the way of the buggy whip.

What do you feel the importance of CR will be for the future?

When it was clear that SDR needed embedded intelligence, I formulated CR and shared that vision. In one of those experiences where lightning is not supposed to strike the same place twice, the community of radio engineers really resonated with that vision. If CR launches the community into sentient spaces such as homes that are smart enough to extend quality of life well into ageing, then that would be great.

Again, see my Cognitive Radio Architecture book for my vision of sentient spaces, and other examples, e.g. for CRs to team up to help a single Mom take care of her two year old. Are workplaces going to become more sentient because of cognitive radio? In part, but not via DSA, that's for sure. How is a cognitive radio supposed to authenticate a user? As part of a sentient space, it's easy. By itself, it's inconvenient at best. So I hope the CR community transitions to interdisciplinary collaboration, augmenting core SDR and CR value propositions to become the major innovators for the future.

Page 1 of 1

About the author

Dr Joseph Mitola is a key figure in the arena of cognitive radio. He is a Fellow of the IEEE, Distinguished Professor of systems engineering and of engineering and science and is Vice President for the Research Enterprise at Stevens Institute of Technology. Previously, he was consulting scientist at Mitre Corporation.

Most popular articles in RF topics

  • Measuring Noise Figure on RF systems
  • Locating Passive Intermodulation, PIM Faults
  • Open Source Comes to Software Defined Radio SDR
  • Design of 15 GHz GaN MMIC Power Amplifiers
  • Integrated PLLs and VCOs for Wireless Applications
  • Share this page

    Want more like this? Register for our newsletter

    e-paper service refill button Scott Soong | Pervasive Displays
    Machine maintenance made easy with e-paper displays
    Do you wake up craving a caffeine hit to boost you through the morning? And do you rely on coffee to get you through the day?
    Online - Effective Spectrum Analyzer Measurements
    Learn how to make spectrum analyzer measurements at RF and microwave frequencies

    More training courses

    Low Loss Dynamic Compression of CPRI Baseband Data
    Read this paper from Altera that describes a method of using Mu-Law compression for Gaussian-like waveforms providing an efficient methodology.

    More whitepapers

    Radio-Electronics.com is operated and owned by Adrio Communications Ltd and edited by Ian Poole. All information is © Adrio Communications Ltd and may not be copied except for individual personal use. This includes copying material in whatever form into website pages. While every effort is made to ensure the accuracy of the information on Radio-Electronics.com, no liability is accepted for any consequences of using it. This site uses cookies. By using this site, these terms including the use of cookies are accepted. More explanation can be found in our Privacy Policy