48 research outputs found
Spectrum sensing assisted by windowing for fast time-varying channel
No full textInternational audienceIn this paper, we introduce new totally blind spectrum sensing (SS) algorithms, for fast time-varying channel, based on eigenvalue decomposition (EVD) of the covariance matrix of the received signal. The new scheme is based on the sliding window whose the size depends on the coherence time of the channel. First, we evaluate the impact of the mobility on the detection performance. Then, by applying EVD in each window, we focus our study on the maximal estimated largest eigenvalue (MELE). We provide simulation results in order to validate the proposed theoretical expression of the probability density function of the MELE. Finally, simulation results illustrate the performance of the contributions and are compared to other SS methods
Detection of Primary User assisted by Machine Learning over Multipath Channels
No full textInternational audienceIn this paper, we provide new blind spectrum sensing (SS) methods based on a machine learning (ML) model to overcome the effects of multipath channels. We introduce three ML methods in order to improve the detection of the primary user (PU) in a cognitive radio network in severe multipath environment. The ML approaches proposed here are the Naive Bayes Classifier (NBC), the Support Vector Machine (SVM) and the Artificial neural network (ANN). The non-linear separation of the training samples provided by ML features is a good alternative to avoid miss-detection or false alarm obtained with classical fixed threshold detection. Simulations shows that the proposed algorithms outperform classical non-cooperative SS algorithms based on the eigenvalues of the covariance matrix of the received signal. The proposed SS detectors based on ML algorithms are concluded to be good candidates for PU detection over multipath channels as in indoor scenarios, especially in low signal to noise ratios (SNR)
Potential Application of Neuron with Multi Dendrites in Medicines
Get PDFA neural networks based on neurons with multi dendrites employing activation function with variable structure is introduced. No prior work, however, has discovered the impact of dendrites position on heavies’ information to propagate. Given the frequent observations of correlations in the dendrites structure, and the importance of the spike propagation and burst problems, this is a significant gap in our knowledge. To fill that gap, we consider a model neuron with multi dendrites. In this paper, we discover that Hopfield’s type of neural network can generate multi dynamic behaviors. Our contribution in this paper is to modify Hopfield’s neural network in paper [5] by changing position sign of differential system equation in line two and we add a parameter of P, which has an interesting effect on the output response, we obtain four-differential system equation, which means obtaining four models of neurons. Maybe, the results reached in this work shed some light on Henry Poincare’s conjuncture. Each neuron has its dynamic behavior which has never been reported before; one of them has a great similarity with real biological neuron signal. Then we notice that the dynamic behavioral response contains impact of neuron such as, signals of neurons with different directions, bounded regions with different segments and separation regions. Finally, the numerical examples will illustrate the effectiveness of this work
Blind Spectrum Sensing Using Extreme Eigenvalues for Cognitive Radio Networks
No full textInternational audienc
Author-topic based representation of call-center conversations
No full textInternational audiencePerformance of Automatic Speech Recognition (ASR) systems drops dramatically when transcribing conversations recorded in noisy conditions. Speech analytics suffer from this poor automatic transcription quality. To tackle this difficulty , a solution consists in mapping transcriptions into a space of hidden topics. This abstract representation allows to substantiate the drawbacks of the ASR process. The well-known and commonly used one is the topic-based representation from a Latent Dirichlet Allocation (LDA). Several studies demonstrate the effectiveness and reliability of this high-level representation. During the LDA learning process, distribution of words into each topic is estimated automatically. Nonetheless, in the context of a classification task, no consideration is made for the targeted classes. Thus, if the targeted application is to find out the main theme related to a dialogue, this information should be taken into consideration. In this paper, we propose to compare a classical topic-based representation of a dialogue, with a new one based not only on the dialogue content itself (words), but also on the theme related to the dialogue. This original representation is based on the author-topic (AT) model. The effectiveness of the proposed representation is evaluated on a classification task from automatic dialogue transcriptions between an agent and a customer of the Paris Transportation Company. Experiments confirmed that this author-topic model approach outperforms by far the classical topic representation, with a substantial gain of more than 7% in terms of correctly labeled conversations
SVM Assisted Primary User-Detection for Non-Cooperative Cognitive Radio Networks
No full textInternational audienceThis paper presents a new blind spectrum sensing (SS) algorithm based on a machine learning model: the radial basis function support-vector machines (RBF-SVM). As features, the introduced approach uses statistical tests that are based on the eigenvalues of the received signals covariance matrix. Since the decision on the frequency resource occupancy is in fact an issue of labeling binary data, SVM is intended as a potential technique for SS paradigm. The flexibility of SVM for linearly non-separable and high dimensional data makes it a good candidate for our issue, particularly that we consider low signal to noise ratios (SNR). Computer simulations shows that the proposal outperforms classical non-cooperative SS algorithms. © 2020 IEEE
