Deep radio signal clustering with interpretability analysis based on saliency map

Frame interpolation, which generates an intermediate frame given adjacent ones, finds various applications such as frame rate up-conversion, video compression, and video streaming. Instead of using complex network models and additional data involved in the state-of-the-art frame interpolation methods, this paper proposes an approach based on an end-to-end generative adversarial network. A combined loss function is employed, which jointly considers the adversarial loss (difference between data models), reconstruction loss, and motion blur degradation. The objective image quality metric values reach a PSNR of 29.22 dB and SSIM of 0.835 on the UCF101 dataset, similar to those of the state-of-the-art approach. The good visual quality is notably achieved by approximately one-fifth computational time, which entails possible real-time frame rate up-conversion. The interpolated output can be further improved by a GAN based refinement network that better maintains motion and color by image-to-image translation.

As a powerful approach for exploratory data analysis, unsupervised clustering is a fundamental task in computer vision and pattern recognition. Many clustering algorithms have been developed, but most of them perform unsatisfactorily on the data with complex structures. Recently, adversarial autoencoder (AE) (AAE) shows effectiveness on tackling such data by combining AE and adversarial training, but it cannot effectively extract classification information from the unlabeled data. In this brief, we propose dual AAE (Dual-AAE) which simultaneously maximizes the likelihood function and mutual information between observed examples and a subset of latent variables. By performing variational inference on the objective function of Dual-AAE, we derive a new reconstruction loss which can be optimized by training a pair of AEs. Moreover, to avoid mode collapse, we introduce the clustering regularization term for the category variable. Experiments on four benchmarks show that Dual-AAE achieves superior performance over state-of-the-art clustering methods. In addition, by adding a reject option, the clustering accuracy of Dual-AAE can reach that of supervised CNN algorithms. Dual-AAE can also be used for disentangling style and content of images without using supervised information.

Recognizing Human action in video is an very active research topic. There are a growing variety of human action datasets with different video length, different practitioners. Make human action recognition becomes a very difficult topic. A majority researchers solve the problem by extracting key frames from the videos. Most paper use feature Clustering methods to extract key frames in videos. On one hand, the large variety of visual content in videos make handcraft feature isn't effective enough, since there are no fixed descriptors can describe all video cases. On the other hand, traditional clustering algorithms are easily influenced by the choice of initial clustering centers. An Unsupervised feature learning and clustering method for key frame extraction is proposed in this paper, which can be used for human action recognition. Stacked auto-encoder(SAE) is trained using videos from 10 different human actions, SAE is used as a feature extractor to learn features representing human actions. Affinity Propagation Clustering algorithm is used to select key frames from video sequences. We use a variety of videos to do the experiments. Experiments demonstrate that our method can be effectively summarizing video shots considering different human actions.

The advancements in technology have made it possible to automatically record and store large amount of data, which has resulted in a need for development and application of efficient data analysis techniques. Unsupervised data clustering methods have proven to be capable of extracting useful information from various types and sizes of datasets. This paper investigates the performance of the standard agglomerative hierarchical clustering algorithm using two time series datasets from electric power system and neuroscience area. The main steps in clustering procedure are presented in detail. Results show that the effectiveness of the clustering algorithm is affected to a large extent by the main characteristics of the clustering data and algorithm’s parameters.

Seismic facies analysis aims to identify clusters (groups) of similar seismic trace shapes, where each cluster can be considered to represent variability in lithology, rock properties, and/or fluid content of the strata being imaged. Unfortunately, it is not always clear whether the seismic data has a natural clustering structure. Cluster analysis consists of a family of approaches that have significant potential for classifying seismic trace shapes into meaningful clusters. The clustering can be performed using a supervised process (assigning a pattern to a predefined cluster) or an unsupervised process (partitioning a collection of patterns into groups without predefined clusters). We evaluate and compare different unsupervised clustering algorithms (e.g., partition, hierarchical, probabilistic, and soft competitive models) for pattern recognition based entirely on the characteristics of the seismic response. From validation results on simple data sets, we demonstrate that a self-organizing maps algorithm implemented in a visual data-mining approach outperforms all other clustering algorithms for interpreting the cluster structure. We apply this approach to 2D seismic models generated using a discrete, known number of different stratigraphic geometries. The visual strategy recovers the correct number of end-member seismic facies in the model tests, showing that it is suitable for pattern recognition in highly correlated and continuous seismic data.

Currently, strength prediction based on damage pattern recognition of fibre-reinforced composite materials mainly relies on unsupervised clustering methods, which suffer from relatively limited evaluation metrics for clustering outcomes. This study proposes a strength prediction approach for fibre-reinforced composites based on acoustic emission and supervised damage pattern recognition. Firstly, three types of acoustic emission tests were designed to capture the characteristic acoustic emission signals of pure damage modes (fibre fracture, matrix cracking, interface debonding). These signal features were utilised to construct corresponding damage pattern datasets and train a supervised model, achieving a cross-validation accuracy of 92.94%. Subsequently, based on the trained model, the acoustic emission events collected during the tensile failure process of fibre-reinforced composite were distinguished, and the residual strength was predicted using a time series prediction model, yielding performance of MAE = 0.97, RMSE = 1.59, and R 2 = 97.58%. Finally, the prediction results were compared with those derived from unsupervised clustering. The findings demonstrate that the supervised recognition-based approach delivers superior predictive outcomes. This indicates that the supervised identification method offers greater advantages in damage pattern recognition for fibre-reinforced composites. Consequently, this provides a novel approach for structural health monitoring of composite materials.

Intelligence is the trend of industrial development in today’s world, which will inevitably affect the nuclear power field, thus promoting nuclear power industry from digital to intelligent forward.5G and Wi-Fi6, as a new generation of wireless communication technologies, have important advantages in terms of technical performance and security due to the characteristics of dense personnel, wide equipment distribution, complex facility environment and special materials used in nuclear power scenarios. Considering the high requirement of electromagnetic compatibility in nuclear power environment, it is urgent to further study the electromagnetic interference and protection requirements of 5G, Wi-Fi6 and other new technologies on nuclear power unit equipment in nuclear power environment. In this paper, combined with relevant national document standards and industry standards, we will sort out the requirements of Immunity about Radio Frequency Radiation (RFR) for the important equipment in nuclear power plants, and use signal generators to simulate new radio signals such as 5G and Wi-Fi6 to test the immunity of nuclear power instruments and meters. Combined the results of test, we will study the Electro Magnetic Compatibility (EMC) between new wireless technologies and nuclear power plant equipment from different factors, such as Power, bandwidth, modulation, distance, direction. And then, we will summarize the protection requirements about nuclear power equipment.

Modeling visual saliency map of an image provides important information for image semantic understanding in many applications. Most existing computational visual saliency models follow a bottom-up framework that generates independent saliency map in each selected visual feature space and combines them in a proper way. Two big challenges to be addressed explicitly in these methods are (1) which features should be extracted for all pixels of the input image and (2) how to dynamically determine importance of the saliency map generated in each feature space. In order to address these problems, we present a novel saliency map computational model based on tensor decomposition and reconstruction. Tensor representation and analysis not only explicitly represent image's color values but also imply two important relationships inherent to color image. One is reflecting spatial correlations between pixels and the other one is representing interplay between color channels. Therefore, saliency map generator based on the proposed model can adaptively find the most suitable features and their combinational coefficients for each pixel. Experiments on a synthetic image set and a real image set show that our method is superior or comparable to other prevailing saliency map models.

This paper considers one of the problematic issues of creating radio systems based on cognitive radio technology, viz., automatic recognition of the digital-modulation formats of radio signals. In accordance with the recommendations of the E2R and the European Telecommunications Standards Institute (ETSI) consortium, cognitive radio systems have the ability to modulate/demodulate signals in all frequency bands and in all modes of modulation. This process should be performed automatically, according to the current technical capabilities of the available communication system, the requirements for the quality of communication, and different external conditions. This article provides an analysis of the promising methods of automatic recognition of digitally modulated radio signal formats, viz., using the shape of the phase constellation, using the distribution difference of instantaneous phases, and using high-order cumulants. According to the results of the analysis, we propose methods of recognition that are based on cumulant analysis for cognitive radio systems. It is proposed that the decision-making device be an artificial neural network.

The Costas loop can be considered an extension of the phase-locked loop (PLL). The PLL, invented 1932 by French engineer Henri de Belleszice, was first used for the synchronous reception of amplitude modulated radio signals with carrier (c). If the modulation index is less than 100%, the modulated signal is always in phase with the carrier signal (c), irrespective of the polarity of the modulating signal. The PLL then reconstructs a local carrier (loc) that is in phase the carrier (c) of the AM signal. The modulating signal is recovered by multiplying the AM signal with the local carrier (c) with subsequent lowpass filtering. This method of synchronous reception fails, however, when an AM signal with suppressed carrier is transmitted. Here the modulated signal is in phase with the (suppressed) carrier when the modulating signal is positive, but it is in antiphase with the carrier when the modulating signal is negative. When this AM signal is fed to the input of a PLL, it would create a local carrier (loc) that is in phase with the carrier when the modulating signal is positive, but in antiphase with the carrier when the modulating signal is negative. Recovery of the modulating signal is therefore not possible. The situation would be the same when the transmitted signal is a Binary Phase Shift Keying (BPSK) signal, a Quadrature Phase Shift Keying (QPSK) signal, or a Quadrature Amplitude Modulation (QAM) signal. This problem has been solved by the invention of the Costas loop by American engineer James P. Costas in 1956. The Costas loop is able to lock onto a modulated signal also in cases where the carrier is suppressed. When we compare the topology of a Costas loop with that of a PLL, we recognize that the PLL is built up from a cascade of three circuits: (1) a phase detector, (2) a loop filter, and (3) a voltage-controlled oscillator (VCO). There is a feedback path from the output of the VCO to the input of the phase detector. In case of the PLL, the input signal is propagated through this cascade of three blocks. There is only one signal path in the PLL. In contrast, there are two signal paths in the Costas loop: (1) a so-called “I branch” (in phase), (2) a so-called “Q branch” (quadrature phase). It is shown that with that configuration the loop is able to create a local carrier that is correctly locked to the carrier of the received signal even in cases when the carrier is suppressed. The invention of the Costas loop had a dramatic impact on any kind of communication receivers; one or more Costas loops are found in TV, radio, satellite communications, mobile phones, modems, etc.

Currently, breast cancer is one of the most common cancers among women. To aid clinicians in diagnosis, lesion regions in mammography pictures can be segmented using an artificial intelligence system. This has significant clinical implications. Clustering algorithms, as unsupervised models, are widely used in medical image segmentation. However, due to the different sizes and shapes of lesions in mammography images and the low contrast between lesion areas and the surrounding pixels, it is difficult to use traditional unsupervised clustering methods for image segmentation. In this study, we try to apply the semisupervised fuzzy clustering algorithm to lesion segmentation in mammography molybdenum target images and propose semisupervised fuzzy clustering based on the cluster centres of labelled samples (called SFCM_V, where V stands for cluster centre). The algorithm refers to the cluster centre of the labelled sample dataset during the clustering process and uses the information of the labelled samples to guide the unlabelled samples during clustering to improve the clustering performance. We compare the SFCM_V algorithm with the current popular semisupervised clustering algorithm and an unsupervised clustering algorithm and perform experiments on real patient mammogram images using DICE and IoU as evaluation metrics; SFCM_V has the highest evaluation metric coefficient. Experiments demonstrate that SFCM_V has higher segmentation accuracy not only for larger lesion regions, such as tumours, but also for smaller lesion regions, such as calcified spots, compared with existing clustering algorithms.

Волоконно-ефірні технології зможуть стати перспективним рішенням для реалізації телекомунікаційних бездротових систем в міліметровому діапазоні хвиль. Фотонні методи формування радіосигналів та волоконно-ефірна архітектура змужуть забезпечити високу пропускну спроможність мереж за рахунок широкої смуги обробки до 10 ГГц, спектрального мультиплексування та використання форматів модуляції високого порядку. У статті проаналізовані перспективи і переваги реалізації бездротових телекомунікаційних систем з використанням волоконно-ефірних технологій в міліметровому діапазоні хвиль. Показані основні принципи фотонних методів формування та модуляції радіосигналів міліметрового діапазону, архітектури гібридної волоконно-ефірної мережі. Проаналізовано складові шуму, дискретність модуляції амплітуди і фази для квадратурного модульованого радіосигналу з використанням фотонних методів понижуючого перетворення частоти.

The recent improvement of wireless communication networks has made possible for using such networks at the lowest levels of factory automation systems, which typically imposes severe requirements in terms of real-time communication, reliability, and dependability. However, industrial plants represent a hostile environment due to the presence of different noise sources. The presence of these noises can easily inflict damages on the wireless information signals and creates erroneous interruptions on the data transmitted. Wireless Fieldbus can be one of solution where real time communication between controllers and sensor/actuators at the field level of factory automation. Wireless Fieldbus technologies have the potential capability and significant benefits in industrial application. In factory automation, the cost and time needed for the installation and maintenance of the large number of cables can be reduced by wireless technology. This is important in harsh environment where vibrations, high temperature, or moving parts exist that could potentially damage any sort of cabling. The objective of this chapter lies on the feasibility analysis on usage Wireless Fieldbus technologies and comparison of recent industrial wireless technology. For this, we investigate the fundamental problem of Wireless Fieldbus and details of each wireless technology at factory automation.

The introduction of wireless communication technology in nuclear power plants faces challenges, although it can improve productivity and save cable costs. Firstly, wireless communication signals may cause electromagnetic noise interference to existing nuclear power equipment, and radio wave monitoring can lead to information leakage. Secondly, nuclear power plant equipment also generates noise in the working frequency band of wireless equipment, which affects its communication quality. Based on existing standards and test results, this paper discusses strategy about how nuclear power equipment and wireless technologies can coexist, and proposes recommendations to avoid mutual interference.

Recently, the use of unmanned aerial vehicles for both reconnaissance and fire strikes has been increasing. The control of such vehicles and the transmission of telemetry information are carried out using a combined command and telemetry radio link and frequency hopping spread spectrum technology. In such conditions, the task of developing automatic algorithms for detecting and separating radio links for portable radio monitoring devices becomes especially relevant. The aim of this work is to extend the capabilities of portable radio monitoring devices to detect unmanned aerial vehicles’ radio signals in sophisticated electronic environment by automating signal processing. Formally, the research task is to detect and separate several frequency hopping spread spectrum signals in same frequency band. To solve this task, it is first necessary to detect the frequency channels and estimate frequency and, if necessary, temporal and modulation parameters of the signals in these channels. Estimates of signal parameters values are recorded in associative arrays. The paper proposes a scheme of the algorithm for automatic detection and separation of signals with frequency hopping spread spectrum, as well as the structure of the array with the results of frequency band analysis. In the study of developed algorithm, we considered the practically important option of separating radio links with frequency hopping spread spectrum by the spectral width of the frequency element and the value of the signal-to-noise ratio in the frequency channel, since these features can be quite easily distinguished in the frequency domain. The radio links separation was performed by cluster analysis of scatter plots of parameter pairs estimates using a Gaussian mixture model. The developed algorithm can be implemented in existing and prospective portable radio monitoring devices to automatically detect and determine the number of unmanned aerial vehicles by their radio links signals, in particular, under conditions of a priori uncertainty about the values of signal parameters.