Skip to main content
European Commission logo print header

Cooperative Spectrum Sensing Algorithms for Cognitive Radio Networks

Final Report Summary - COSSAR (Cooperative Spectrum Sensing Algorithms for Cognitive Radio Networks)

In most countries, all frequencies have been completely allocated to specific uses. Figure 1 (a) shows the Federal Communication Commissions (FCC) frequency allocation chart that indicates multiple allocations over essentially all of the frequency bands. Cognitive radio, a paradigm originated by Mitola, has emerged as a promising technology for maximizing the utilization of the limited bandwidth while accommodating the increasing amount of services and applications in wireless networks.

In cognitive radio networks (CRNs), the first cognitive task preceding any form of dynamic spectrum management is the sensing and identification of spectrum holes in wireless environments as depicted in Fig. 2 (a). Thus, spectrum sensing (SS) and subsequently efficient Primary User (PU) detection are the most essential enabling functionality for CRNs to detect the spectrum holes and opportunistically use the under utilized frequency bands without causing harmful interference to legacy networks, i.e. PUs. SS should also monitor the activation of PUs in order for the secondary users to vacate the occupied spectrum segments. However, it is difficult for a CR to capture such information instantaneously due to the absence of cooperation between the primary and secondary users. Recently, the vast majority of leading researchers are putting their efforts into developing accurate cooperative SS (CSS) models and deriving efficient PU detectors under developed models in order to address this major bottleneck in CRNs.

“Cooperative Spectrum Sensing Algorithms for Cognitive Radio Networks”, i.e. “COSSAR” project addressed the major drawbacks of the conventional CSS models and techniques. The project, incorporating concerns of practical applications, such as detection algorithms that are able to operate in various environments and a comprehensive communication model that represents the reality, produced a comprehensive view of the mechanics and capabilities of CSS and CRNs.

Specifically, the drawbacks of the conventional CSS have been addressed by accomplishing the following scientific objectives i) Robust Spectrum Detection Algorithms, ii) Comprehensive and Realistic Spectrum Sensing Models, iii) In-Network Cooperative Spectrum Sensing.

The main outcome of the project has been the development of Distributed Detection Algorithm.

In general terms, this project was based on the premise that, to improve the detection performance, one needs to improve the quality of the estimate of the signals. To this end, adaptive set-theoretic filters have been proposed. In this paradigm, from the measurements every node builds closed convex sets that are likely to contain the estimates of the signal. Nodes are also free to build additional sets to incorporate a priori knowledge of the signal. It has to be noted that the performance of set-theoretic adaptive filters greatly depends on the choice of the sets. To keep the computational complexity low, sets have to be devised onto which the projections are easy to compute.

In the first step, measurements are used to improve the estimates. In the second step prior knowledge to improve the estimates is used. In the third step it is tried to reach consensus to improve the estimates. In the first two steps, there is no information exchange between two nodes that are not directly connected. To further improve the detection performance, one can diffuse information throughout the network by trying to make nodes reach consensus on their estimates. The third step of the algorithm is thus an information diffusion operation performed by one iteration of a consensus algorithm.

Empirical evaluations have been performed. A system consisting of e.g. 20 nodes has been simulated. The nodes were distributed uniformly at random in a unit grid. With this communication range, the network is fairly sparse when strongly connected. Networks which are not strongly connected are discarded because evaluation of the performance of the algorithm is of interest where all N nodes cooperatively detect the presence of the primary user. The network is time-invariant during each run of the simulation.
The proposed method is compared to that in (Cattivelli), which is known to provide low mean-square error at steady state, with step size 0.06 and detection threshold 0.08.

Both compared algorithms use as the performance metric the probability of detection and the probability of false alarm in the nodes with the worst performance. The empirical probabilities of false alarm and detection are obtained from the results of 100 000 runs of the simulation, each of which is using a different network topology and noise variances. Fig. 3 shows the results obtained with the described settings. It can be seen from the figure that, at steady state, the proposed method outperforms the LMS algorithm in both the probability of false alarm and the probability of misdetection. The reason is that the project’s approach was able to increase the reliability of the sets by effectively decreasing the variance of noise, and also recalling that the distributed LMS algorithm is similar to a particular case of this approach with m=1 and without the projection onto C. An additional important advantage of the proposed method over the distributed LMS algorithm is that the estimates of the signal belong to the set C when nodes exchange estimates. This property decreases the number of scalars exchanged via wireless links from M to 1. As a result, in practice the proposed algorithm can also be more energy efficient than the original LMS method because most energy is typically spent on data transmission rather than on local computation.

In addition to the new technical and scientific skills that the Marie Curie Fellow Researcher obtained during the execution of this project, he also got a chance to enhance his personal and management skills. This project included training activities to develop complementary skills (proposal preparation to request funding, project management, tasks coordination, technical staff supervision, etc.).

Consequently, ability to view problems from a wider perspective has been nourished and the Fellow Researcher attained a high level of professional maturity and independence thanks to a complete set of technical tools and inter-personal skills gathered during the implementation of this project. The Fellow Researcher was encouraged to actively pursue to acquire additional funding and build his own group. He fully exploited this opportunity by setting up his own team at Wroclaw Research Center EIT+ in which the University of Wroclaw had its shares and that was an endeavor promoted by the local government, local academia and financed from the structural funds.

The fundamental complementary research and management skills that the Fellow Researcher acquired have immensely contributed to his long-term career goals which is i) to step up in academic degree and ii) to become competent and skillful in management of large-scale R&D projects. COSSAR has contributed immensely towards fulfillment of the first objective by placing the Fellow Researcher in the research environment, i.e. the host institution where he had an opportunity to pursue habilitation with his research. COSSAR has also paved the way for the Fellow Researcher to obtain appropriate skills and know-how in managing large-scale projects. The Fellow Researcher has been encouraged strongly to organize consortia in Poland and apply for research funding from the Polish Innovative Economy Program, where the level of funding has been very often an order of magnitude greater than in regular FP7 projects. The Fellow Researcher has leveraged on COSSAR and his past experience, know-how and contacts to acquire funding from this source. He has been successful and it has boosted his career and equipped him with skills to lead complex R&D projects. He has also taken active part in the COST IC0902 action and in this way leveraged on the COSSAR results to enhance his presence on the European cognitive radio stage.