Impacto dos erros de estimação de canal na fusão de decisões com pré-distorção aplicada ao sensoriamento espectral cooperativo

Abstract

The concept of cognitive radio, CR emerged with the objective of relieving the scarcity and under-utilization of the radio frequency. Fundamentally, the reconfigurability and the ability of the spectrum opportunistic utilization through spectrum sensing allow CRs to monitor and locate frequency bands that are not being occupied by the users that hold the license to use these bands, called primary users (PU) at any moment and location. The spectrum sensing performance may be affected due to various factors inherent in wireless communication systems such as thermal noise, hidden terminal, multipath fading and blocking or shadowing. One way of easing the effect of these factors is the use of cooperative spectral sensing (cooperative spectrum sensing (CSS). CSS takes advantage of the spatial diversity promoted by several geographically spaced CRs to reduce uncertainty in local decision-making relative to the state of the sensed channel. Nevertheless, in traditional CSS schemes the need for spectral resource is raised due to the increase in the number of CRs in the orthogonal transmission techniques used to provide channels dedicated to these CRs. Recently, a CSS scheme has been proposed where all the CRs in cooperation send simultaneously their individual decisions to the fusion center (FC) using the same frequency of the control channel. Due to this, the spectral efficiency is increased in the use of the control channel. However, as a consequence, the signals received by the CF add up in a noncoherent way due to the intentional interference produced by this scheme. However, even with this interference, the global decision process on the occupation state of the sensed channel can still be performed. In order to improve the performance of this system, a CSS technique was proposed, where the individual decisions of the MCRs are pre-distorted before being sent to the FC simultaneously and at the same frequency, in order to pre-compensate channel gains of control. Due to this, the spectral efficiency of the above mentioned scheme is maintained, in addition to providing better performance in spectral sensing. However, no research was done about the impact of control channel estimation errors in this technique. This thesis analyzes the performance of the CSS technique, with pre-compensation, under the influence of errors in the estimate of the channel gains, which degrade the pre-compensation. In order to assign the analysis of a practical appeal, the estimation errors are modeled in agreement with the linear estimation method under the criterion of minimum mean square error (LMMSE) and with the sample average estimator (SA), allowing an adequate tying between the degree of inaccuracy of the estimation and the signal-to-noise ratio at the input of the receiver where the estimation is performed.