XXXI SIMPÓSIO BRASILEIRO DE TELECOMUNICAÇÕES – SBrT2013, 1-4 DE SETEMBRO DE 2013, FORTALEZA, CE Performance Analysis of Spectrum Sensing Techniques in Nakagami and Rice Fading Channels Ricardo A. da S. Júnior, Rausley A. A. de Souza and Dayan A. Guimarães Abstract— This paper aims at investigating the performance of four eigenvalue-based techniques for centralized data-fusion cooperative spectrum sensing in cognitive radio networks over flat Nakagami-m and Rice fading channels. The detection techniques are the generalized likelihood ratio test (GLRT), the maximum-minimum eigenvalue detection (MMED), the maximum eigenvalue detection (MED), and the energy detection (ED). In the case of Nakagami-m, arbitrary fading and phase parameters were assumed, and so was with the Rice parameter in the case of the Rician model. Index Terms— cognitive radio, cooperative eigenvalue spectrum sensing, Nakagami, Rice. I. INTRODUCTION Nowadays, there is a growing demand for effective use of spectrum and spectral efficient management strategies in the context of fast developing wireless communications systems. The spectrum resources have become scarce, and at the same time there is an increasing demand for better quality of service, as well as higher transmission rates. Nevertheless, in fact there is an artificial scarcity of spectrum, since there are bands that are not actually used during all time in a given region [1]. The cognitive radio (CR) concept [2] can be applied to this context, aims at using the electromagnetic spectrum more efficiently. A CR system uses advanced techniques that optimize the occupation of the bands, and spectrum sensing techniques to find the so-called spectral opportunities within bands of interest in a given area and in a given time. Thus, a CR system makes it possible to use the available spectrum in temporal, spatial and frequency dimensions, without causing interference to licensed systems. Nonetheless, the scenarios of spectral occupancy differ depending of several factors, such as channel conditions, location and prevailing political control of spectrum usage. This implies greater system complexity, since the cognitive cycle of the CR concept includes a step for learning the channel [2]-[3]. Hence, the behavior of the channel, or more precisely the channel model, influences the operation and performance of a CR. Then, evaluating the performance of a CR system under different channel models is of paramount importance. Moreover, the choice of the spectrum sensing technique will also influence the detection performance, depending on the cognitive network architecture and the conditions of the channel. Many detection techniques for spectrum sensing have been proposed so far, e.g. the matched filter, the cyclostationary and the energy detection [4]-[5]. Among the latest ones are those based on the eigenvalues of the received signal covariance matrix; see [6]-[8] and references therein. These techniques have received a lot of attention mainly because they do not require prior information on the transmitted signal, and, in contrast to the energy detection, some eigenvalue-based schemes do not need to know the noise variance either. No matter the sensing technique adopted, the detection performance depends on the reception conditions of the CRs, and therefore on the propagation environment. For example, in [9] comparisons were made among different models for the energy detector under conditions of additive white Gaussian noise (AWGN) and Rayleigh fading channels. It has been shown that the problem of energy detection lies in the uncertainty of estimating the noise power, which degrades the detection performance [10]-[12]. In [13] the authors analyze the probability of miss detection of the energy detector under Nakagami fading channels. Recently, in [14] the authors presented a new implementation-oriented model in which typical signal processing tasks of a direct-conversion CR receiver were taken into account considering the Rayleigh fading channel. The aim of this paper is to present the analysis of the spectrum sensing performance under two important channel models: Nakagami-m [15] (with arbitrary phase and fading parameters) and Rice [16] (with arbitrary Rice parameter). The Nakagami distribution can be parameterized to model various fading conditions such as Rayleigh and Rice. This means that it is possible to control the severity of the Nakagami fading by making this distribution to fit more appropriately into real scenarios with multipath propagation [13]. The Nakagami-m and Rice distributions, which are general, flexible, and easily tractable mathematically, have also been proved useful in practice [17]-[18]. In what concerns the detection technique, we consider the eigenvalue-based generalized likelihood ratio test (GLRT); the maximum-minimum eigenvalue detection (MMED), also known as the eigenvalue ratio detection (ERD); the maximum eigenvalue detection (MED), also known as Roy’s largest root test (RLRT); and the energy detection (ED), applied to a centralized data-fusion cooperative spectrum sensing scheme. ED is not an exclusively eigenvalue-based detection technique, but it can be implemented using eigenvalue information. It has been included in the present investigation for the sake of completeness, also giving support to a broader pool of comparisons. The remainder of this paper is structured as follows. Section II presents the system model for the eigenvalue-based sensing technique and the fading channels models. Section III reports simulation results and discussions concerning the influence of Ricardo A. da S. Júnior, Rausley A. A. de Souza and Dayan A. Guimarães. National Institute of Telecommunications (Inatel), PO Box 05, 37540-000 Santa Rita do Sapucaí - MG - Brazil, E-mails: ricardojunior@inatel.br, rausley@inatel.br, dayan@inatel.br. This work was partially supported by Fapemig (TEC - APQ-01255-12).