Due to the rapid growth of new wireless communication services and applications, much attention has been directed to frequency spectrum resources. Considering the limited radio spectrum, supporting the demand for higher capacity and higher data rates is a challenging task that requires innovative technologies capable of providing new ways of exploiting the available radio spectrum. Cognitive radio (CR), which is among the core prominent technologies for the next generation of wireless communication systems, has received increasing attention and is considered a promising solution to the spectral crowding problem by introducing the notion of opportunistic spectrum usage. Spectrum sensing, which enables CRs to identify spectral holes, is a critical component in CR technology. Furthermore, improving the efficiency of the radio spectrum use through spectrum sensing and dynamic spectrum access (DSA) is one of the emerging trends. In this thesis, we focus on enhanced spectrum sensing techniques that provide performance gains with reduced computational complexity for realistic waveforms considering radio frequency (RF) impairments, such as noise uncertainty and power amplifier (PA) non-linearities.
The first area of study is efficient energy detection (ED) methods for spectrum sensing under non-ﬂat spectral characteristics, which deals with relatively simple methods for improving the detection performance. In realistic communication scenarios, the spectrum of the primary user (PU) is non-ﬂat due to non-ideal frequency responses of the devices and frequency selective channel conditions. Weighting process with fast Fourier transform (FFT) and analysis filter bank (AFB) based multi-band sensing techniques are proposed for overcoming the challenge of non-ﬂat characteristics. Furthermore, a sliding window based spectrum sensing approach is addressed to detect a re-appearing PU that is absent in one time and present in other time. Finally, the area under the receiver operating characteristics curve (AUC) is considered as a single-parameter performance metric and is derived for all the considered scenarios.
The second area of study is reduced complexity energy and eigenvalue based spectrum sensing techniques utilizing frequency selectivity. More specifically, novel spectrum sensing techniques, which have relatively low computational complexity and are capable of providing accurate and robust performance in low signal-to-noise ratio (SNR) with noise uncertainty, as well as in the presence of frequency selectivity, are proposed. Closed-form expressions are derived for the corresponding probability of false alarm and probability of detection under frequency selectivity due the primary signal spectrum and/or the transmission channel. The offered results indicate that the proposed methods provide quite significant saving in complexity, e.g., 78% reduction in the studied example case, whereas their detection performance is improved both in the low SNR and under noise uncertainty.
Finally, a new combined spectrum sensing and resource allocation approach for multicarrier radio systems is proposed. The main contribution of this study is the evaluation of the CR performance when using wideband spectrum sensing methods in combination with water-filling and power interference (PI) based resource allocation algorithms in realistic CR scenarios. Different waveforms, such as cyclic preﬁx based orthogonal frequency division multiplexing (CP-OFDM), enhanced orthogonal frequency division multiplexing (E-OFDM) and filter bank based multicarrier (FBMC), are considered with PA nonlinearity type RF impairments to see the effects of spectral leakage on the spectrum sensing and resource allocation performance. It is shown that AFB based spectrum sensing techniques and FBMC waveforms with excellent spectral containment properties have clearly better performance compared to the traditional FFT based spectrum sensing techniques with the CP-OFDM.
Overall, the investigations in this thesis provide novel spectrum sensing techniques for overcoming the challenge of noise uncertainty with reduced computational complexity. The proposed methods are evaluated under realistic signal models.
|Translated title of the contribution||Enhanced Spectrum Sensing Techniques for Cognitive Radio Systems|
|Publisher||Tampere University of Technology|
|Number of pages||86|
|Publication status||Published - 13 Feb 2015|
|Publication type||G5 Doctoral dissertation (article)|
|Name||Tampere University of Technology. Publication|
|Publisher||Tampere University of Technology|