Reducing the effects of the noise uncertainty in energy detectors for cognitive radio networks

Abstract

SUMMARYBecause of its ease of implementation and minimum requirements about the primary signals' information, energy detection is broadly considered for signal detection in spectrum sensing algorithms. However, the noise uncertainty phenomenon, caused by the random variations in the noise power, degrades the performance of an energy detector, particularly when the signal-to-noise ratio (SNR) is low. In this work, we propose to reduce the negative effects of the noise uncertainty in the performance of an energy detector by dynamically adapting its detection threshold to the noise conditions experienced at each sensing epoch. The noise power is estimated from the received signal samples using an algorithm based on a high-pass filters bank and median filtering. With our proposal, it is possible to maintain a constant and low false alarm rate in the presence of noise uncertainty, without increasing the probability of misdetection, even in the low SNR regime, and without increasing the number of samples considered for spectrum sensing.

Robust spectrum sensing under noise uncertainty for spectrum sharing

“…Here, T is calculated for the desired FAR according to various forms of the GOS algorithm from (9) , as shown in Table 1. GOS (6) indicates that the sixth smallest sample is selected as the background power level and GOS(1, 2, …, 8) is defined as the summation of the background power level for all samples in the window in Figure 2. Figure 3 shows the detection probabilities of various types of GOS detectors in homogeneous environments. As expected, the GOS(1, 2, …, 8) detector achieves the best performance among the four types of GOS detectors, whereas the GOS(6) detector shows the least performance because the detection performance in homogeneous situations improves as the number of noise samples increases.…”

“…A number of spectrum sensing algorithms have been studied and can be classified into two categories. The first category requires a priori knowledge about the PU signal, such as a cross-correlation scheme [1] and cyclostationary-based detection [2], whereas the second category does not require any information of the PU, such as the eigenvalue decomposition detector [3] and energy detector (ED) [4][5][6][7][8]. On the other hand, the eigenvalue decomposition detector has significant complexity, and cannot achieve a desired performance in terms of the very small adjacent channel interference (ACI) [9].An ED is generally used because it does not require a priori knowledge of the PU signals.…”

“…It provides good detection performance when there is no noise uncertainty. However, an ED is very sensitive to noise uncertainty [4][5][6][7][8]. In practice, the measured data may experience multipath fading and can )].Spectrum sensing plays an important role in spectrum sharing.…”

Interference analysis of cognitive radio networks

“…We first model the distance relations between a PT, a PR, and a group of CRs. Note that the sensing accuracy of energy detector can be improved (e.g., [13]) at the expense of increased complexity. The power decay with respect to propagation distance is modeled using the two-ray ground reflection path loss model.…”

“…In this paper, I aggr is expressed as a function of both N and SNR ε . Note that the sensing accuracy of energy detector can be improved (e.g., [13]) at the expense of increased complexity.…”