Survey of surveys on spectrum sensing for Cognitive Radio

Several reviews on spectrum sensing techniques for cognitive radio have been published. In this post I comment three recent surveys that focus on different aspects of the problem of primary user detection in cognitive radio environments, finding that some important (at least to me) questions are surprisingly left out.

[Y09] offers a high level overview of spectrum sensing for cognitive radio describing the main challenges (such as the hidden node problem, detection of spread spectrum users...) and classifying the different sensing methods as shown in the following figure:


We can see that while they maintain the classical division into Energy detection, Cyclostationarity detection and Matched filter detectors; they also introduce the concepts of Waveform-based sensing and Radio identification.

Waveform-based sensing refers to the detection of known transmitted patterns such as pilots or preambles, and of course increases the accuracy of the detector with a relatively low complexity. On the other hand radio identification refers to using the available a priori knowledge about the transmitter technology we are interested in (such as transmission range, frequency hops, spectral shape...).

The topics covered in [Y09] also include cooperative detection and how to use historic data, such as temporally correlated traffic, to improve detection performance. Additionally this survey includes an interesting part dedicated to how Spectrum Sensing is performed in current wireless standards, including IEEE 802.11k, Bluetooth or IEEE 802.22.


[A09] is a short review focusing on spectral estimation. Different approaches to power spectral density (psd) reconstruction are discussed, including pilot detection, multi taper spectrum estimation or filter banks. Once the psd has been estimated it can be used to detect primary users in cognitive radio systems.


Finally, [Z10] focuses on technical aspects of detection theory applied to cognitive radio environments. This includes using space and time correlation, cyclostationarity detection or cooperative sensing. Other more involved topics are threshold derivation, noise power uncertainty, or what they call robust spectrum sensing. Robust spectrum sensing theory is useful when the a priori knowledge of the noise or signal distribution is limited or imprecise.

The section Future Developments in [Z10] poses some problems related to primary user detection that did not get so much attention from the research community. Here is briefly presented one of the points I missed in all these reviews, namely wideband spectral sensing. Most of the articles cited in these reviews focus on the problem of detecting a single primary user present in the band of interest, however, in general it can be expected that the spectral monitor simultaneously processes multiple channels, and thus more sophisticated detection techniques have to be used (such as, e.g. compressive sensing).


Remember that another way to keep track of the most recent research on spectrum sensing is to take a look to key conferences on cognitive radio, such as the marathonian spectrum sensing sessions at the ICASSP 2010.

[Y09]	T. Yucek and H. Arslan A survey of spectrum sensing algorithms for cognitive radio applications. IEEE Communications Surveys & Tutorials, Volume: 11 Issue:1. First Quarter 2009.
[A09]	D.D. Ariananda, M.K. Lakshmanan and H. Nikoo A survey on spectrum sensing techniques for Cognitive Radio. Second International Workshop on Cognitive Radio and Advanced Spectrum Management, May 2009. CogART 2009.
[Z10]	Y. Zeng, Y. Liang, A. T. Hoang, and R. Zhang A Review on Spectrum Sensing for Cognitive Radio: Challenges and Solutions. EURASIP Journal on Advances in Signal Processing Volume 2010.

Unlicensed access to television broadcasting spectrum in Europe

The previous post about the unlicensed access to TV-band in the US was finished with the question What can Europe learn from this?. I will try here to answer this question by summarizing the current state of the European regulations for unlicensed access to the white spaces in the television band.

The Open Spectrum Alliance is a coalition of companies, organizations and individuals founded in May 2009 in order to push the unlicensed access to the spectral resources. They actively collaborate with the CEPT's Electronic Communications Committee working group designated to study the technical and operational requirements for the operation of cognitive radio systems.

The SE43 working group is currently preparing a document defining the requirements for the operation of cognitive radio systems in the white spaces of the band 470-790 MHz. This draft describes the protection requirements of terrestrial broadcasting (detection thresholds, hidden node margins...) together with the architecture to achieve it (spectral sensing, geolocation database, combination of sensing and geolocation...).

That is, they are working in a document equivalent to the rules approved by the FCC in November 2008 for the regulation of devices using white spaces of the TV band in the USA. While the FCC had been examining this issue for six years prior to the elaboration of the document, the European version could be expected to be ready at the end of the year. That is, the document would be developed in three years in part by using the experience obtained from the american pioneering work.

In order that the european ruleset comes into play further steps are needed due to the more complex hierarchy in the EU. Therefore we could say that the european regulatory process presents a delay of 3 years with respect to the USA's framework for cognitive radio, which already defined proposals for the geolocation databases and cognitive radio data networks tests.

DySPAN coverage: 2 Pragmatic Papers

SpectrumTalk has a new entry about two papers presented at the DySPAN'10 conference. The post 2 Pragmatic Cognitive Radio Papers from DySPAN 2010 comments two different aspects of regulation issues related to Cognitive Radio in order to guarantee primary users protection.

Signal processing with compressed measurements

This post is about two papers related to signal processing in the compressed domain. In the special issue on Compressive Sensing in the Journal of Selected topics in Signal Processing we can find an article with title "Signal Processing With Compressive Measurements" [D10]. The abstract reads:

The recently introduced theory of compressive sensing enables the recovery of sparse or compressible signals from a small set of nonadaptive, linear measurements. If properly chosen, the number of measurements can be much smaller than the number of Nyquist-rate samples. Interestingly, it has been shown that random projections are a near-optimal measurement scheme. This has inspired the design of hardware systems that directly implement random measurement protocols. However, despite the intense focus of the community on signal recovery, many (if not most) signal processing problems do not require full signal recovery. In this paper, we take some first steps in the direction of solving inference problems-such as detection, classification, or estimation-and filtering problems using only compressive measurements and without ever reconstructing the signals involved. We provide theoretical bounds along with experimental results."

A general framework for signal processing of compressed measurements without reconstructing the original signal would be of major importance. However I find that the paper fails to address a key point of the problem: sparsity. If sparsity is ignored the compressed (in the usual setting) measurements reduce to a linearly transformed version of the original signal, what fits into the classical signal processing framework (since the compression matrix is assumed known). [D10] analyzes the performance of different signal processing algorithms (detection, estimation and filtering) from compressed meassurements when the compression matrices are chosen randomly fulfilling a modified RIP property. Since the only reference to the sparsity of the original signal is then through the RIP of the compression matrices higher processing performance may be achieved by reconstructing the signal (remember that RIP is a sufficient but not necessary condition for reconstruction).

A much more involved analysis for the estimation setting was presented in the compressed sensing's day at the ICASSP. In "On unbiased estimation of sparse vectors corrupted by Gaussian noise" [J10] Alexander Jung theoretically analyzes the behavior of the achievable estimation performance in the sparse setting, as opposed to [D10] that only analyzes a quite simple estimation problem from a set of compressed measurements.

[J10] studies the problem of estimating a sparse vector of parameters with unknown support. The interesting result appears when the cardinality of this support is assumed known. In this setting the Cramer Rao Lower Bound (CRB) of the estimate coincides with the minimum square error obtained by an estimator that has a priori knowledge about the positions of the non-zero entries (support) of the vector to estimate [B09]. It is easy to see that this bound is not achievable if this support cannot be determined from the measurements, e.g. in the low SNR regime. The Barankin Bound (BB) is also a lower bound on the unbiased estimation performance and in general it is tighter than the CRB. The interesting part is that it shows a behaviour similar to the Maximum Likelihood (ML) estimate of the parameter vector.


In the Fig. 1 it is shown the behaviour of the different bounds for a given problem instance. We can see that while the CRB is constant even for low SNR values, in this regime the BB rises due to the impossibility of completely determinig the support. We can also see that the ML estimator presents a similar behaviour, with the only difference of the higher SNR threshold where its MSE drops to the CRB.

[D10]	M.A. Davenport, P.T. Boufounos, M.B. Wakin and R.G. Baraniuk Signal Processing With Compressive Measurements. Selected IEEE Journal of Topics in Signal Processing, April 2010.
[J10]	A. Jung, Z. Ben-Haim, F. Hlawatsch, and Y. C. Eldar On unbiased estimation of sparse vectors corrupted by Gaussian noise  in Proc. 2010 IEEE International Conference on Audio, Speech and Signal Processing (ICASSP 2010), Dallas, TX, March 2010.
[B09]	Z. Ben-Haim and Y. C. Eldar The Cramer–Rao bound for sparse estimation   IEEE Trans. Signal Processing, May 2009, submitted.

CR Spam

Many websites attempt to use Google's pagerank algorithm to climb positions in the search results. Most of these backlinks are automatically generated: I recently found a link building website with an article about Cognitive Radio. I specially liked the part: "[Cognitive Radio] senses the spectrum environment, automatically search and use idle Cheap Lacoste Polo Shirts [...]".

Patent: Signal Detection in Cognitive Radio Systems

After writing the post about the companies with most patents on cognitive radio I wondered how innovative these patents could be. To check it I present here one of the lastest patents filed by Motorola related to spectral sensing: US20100081387: Signal detection in cognitive radio systems.


The first claim, the soul of the patent, reads like that:

A method, with a cognitive radio wireless device, for dynamically managing signal detection in a cognitive radio system, the method comprising:

• performing spectrum sensing for a first sensing frame on at least one communication channel;
• receiving, in response to performing spectrum sensing, at least one observed signal on the at least one communication channel;
• and performing a detection decision to determine if the observed signal is one of noise and an active signal associated with an active user, wherein performing the detection decision comprises:
  1. determining an energy estimation .epsilon. associated with the at least one observed signal;
  2. comparing the energy estimation .epsilon. with a current detection threshold, wherein the current detection threshold is one of an arbitrarily defined threshold and a detection threshold based on a previous detection decision for a sensing frame immediately prior to the first sensing frame;
  3. setting, in response to the energy estimation being above the current detection threshold, a first new detection threshold equal to the current detection threshold;
  4. and setting, in response to the energy estimation .epsilon. being below the current detection threshold, a second new detection threshold as a function of the current detection threshold and the energy estimation .epsilon.;
  wherein one of the first and second new detection threshold is used for at least one subsequent detection decision for at least a second sensing frame."

That is, they use energy detection in a given channel by comparing the meassured energy with a threshold that is updated in the case that the channel is decided as vacant.

A patent by definition

must be innovative to the point that it wouldn't be obvious to others."

Then my question is if updating the threshold after non-detection of a primary user is innovative enough in order to justify a patent, or most of these patents are just for the future work of their lawyer's offices.

Performance evaluation in Cognitive Radio systems

While performance evaluation is a key issue to compare and rank different cognitive radio systems, it has received a limited attention by the research community [Z09]. For example when I attended the ICASSP sessions related to cognitive radio I observed the lack of a common framework to rank the different algorithms.

Each author employs different assumptions on the cognitive node a priori knowledge, channel model, front-end characteristics, working environment... In the case of spectral sensing this problem is generally avoided by comparing the proposed algorithms with the very simple energy detector (and of course beating it). Other global algorithms are more complicated to evaluate since even the most simple cognitive network presents a cumbersome number of possible metrics (e.g. total throughput, maximum achievable sum rate at primary or secondary systems, power dissipated at a given node, probability of outage driven by secondary interference, spectral efficiency...).

This problem, probably common to other research areas, has difficult solution until a common framework for testing cognitive radio algorithms is developed. In this context Wireless @ Virginia Tech is developing an open source Cognitive Radio architecture:

The objective of the design is to develop a distributed & modular system that provides portability and interoperability between components developed in different programming languages, across different SDR and hardware platforms. [...] Users of CROSS can focus entirely on one aspect of the cognitive radio radio without developing or modifying components that have no direct relevance to their specific focus of research."

However, from the available documentation, I understand that the physical layer is limited to existing SDR components, and thus it is not useful for experiments that involve, for instance, a crompressive sampling frontend or sophisticated sensing algorithms.

In the same direction another Cognitive Radio Cognitive Network Simulator is being developed by Jing Zhong and Jialiang Li. I say in the same direction because it seems that it is a high level implementation of the cognitive radio network and it does not allow complex physical layer tweaks.

I have a keen interest in these network simulators since I've been recently working with performance evaluation of the game theoretical framework developed in [JVM10]. I found it difficult to determine the most relevant performance metrics.

[Z09]	Y. Zhao, S. Mao, J.O. Neel and J.H. Reed Performance Evaluation of Cognitive Radios: Metrics, Utility Functions, and Methodology. Proceedings of the IEEE, 2009.
[JVM10]	Sudharman K. Jayaweera, Gonzalo Vazquez-Vilar and Carlos Mosquera Dynamic Spectrum Leasing (DSL): A New Paradigm for Spectrum Sharing in Cognitive Radio Networks  accepted for publication in IEEE Transactions on Vehicular Technology, Jan 2010.

DySPAN 2010 coverage

Michael J. Marcus at Marcus Spectrum Solutions LLC is attending the DySPAN 2010 conference in Cognitive Radio and offering some coverage in his blog SpectrumTalk.

In his post Cognitive Radio Conference in Singapore he comments how Singapore is very interested in exploring the various spectrum environments that white space technology could operate in. A set of terms and conditions has been already developed for interested parties in order to guarantee primary users protection.

Patents related to Cognitive Radio

If you wonder who is taking positions in the upcoming fight for the future Cognitive Radio standards just take a look to this chart showing the patents related to cognitive radio approved in the first quarter of 2010 by Tech & IP. The three leading companies in number of patents are Motorola (with 14% of the patents), Samsung (14%) and Qualcomm (8%):

While Motorola is one of the greatest contributors of cognitive radio research (with numerous technical contributions in IEEE DySPAN, European End-to-End Reconfigurability Project, SDR Forum, ...) most of its work was about integrating next generation “cognitive” handsets into current telecommunication infrastructure. On the other hand Samsung actively participated into the development of the elaboration of the IEEE 802.22 proposal, first CR-oriented standard. Its efforts focused mainly on pushing their spectrum sensing technologies. While Qualcomm is yet focused on the development of LTE, it is also taking positions with its patents on cognitive radio.