Search

In order to realize smart factory which is appeared in the concepts of Society 5.0 and Industry 4.0, wireless Internet of Things (IoT) becomes a key technology. Inside of the factory, there are heavy DATA traffics exist between moving/movable machine tools/robots/automated guided vehicle (AGV) with wireless access point (AP) and high speed moving/rotating objects/materials with wireless sensor node (SN). Currently, wireless systems used for wireless IoT are limited such as wireless-LAN/Bluetooth/RFID. Since these system’s frequency bands are limited and narrow, the interference between these systems becomes severe. In this paper, we propose a simplified beam forming wireless IoT communication using novel Wi-Fi backscatter system to notify the existence of SN who wants to send its DATA to AP. For Wi-Fi AP with beamforming function, knowing the numbers of existing SNs in each beam area will be a help for efficient beam control. This paper describes the Wi-Fi backscatter system in which the SNs backscatter the Wi-Fi signal transmitted from the AP then the AP detects the backscatter signals from SNs and know the existing SN’s number in its beam area. Since this backscatter system uses Wi-Fi modulated signal, the separation of SNs’ backscatter signals from AP’s transmitting Wi-Fi signal becomes complicated. This paper also describes the transceiver architecture of AP which reduces the complexity of signal processing and improves the dynamic range of reception.

Being able to learn from complex data with phase information is imperative for many signal processing applications. Today’s real-valued deep neural networks (DNNs) have shown efficiency in latent information analysis but fall short when applied to the complex domain. Deep complex networks (DCN), in contrast, can learn from complex data, but have high computational costs; therefore, they cannot satisfy the instant decision-making requirements of many deployable systems dealing with short observations or short signal bursts. Recent, Binarized Complex Neural Network (BCNN), which integrates DCNs with binarized neural networks (BNN), shows great potential in classifying complex data in real-time. In this paper, we propose a structural pruning based accelerator of BCNN, which is able to provide more than 5000 frames/s inference throughput on edge devices. The high performance comes from both the algorithm and hardware sides. On the algorithm side, we conduct structural pruning to the original BCNN models and obtain 20 × pruning rates with negligible accuracy loss; on the hardware side, we propose a novel 2D convolution operation accelerator for the binary complex neural network. Experimental results show that the proposed design works with over 90% utilization and is able to achieve the inference throughput of 5882 frames/s and 4938 frames/s for complex NIN-Net and ResNet-18 using CIFAR-10 dataset and Alveo U280 Board.

The spectrum underlay concept promises enhanced spectrum utilization without disturbing legacy / licensed or scientific primary users, so long as their interference constraints can be met. Existing underlay schemes assume that both the primary signal to secondary interference plus noise ratio, and the secondary signal to primary interference plus noise ratio can be high enough at the primary and secondary receiver, respectively. Even if the cross-network channel state information is available at the secondary users, these two conflicting requirements are hard to achieve simultaneously in practice. This work proposes a practical data-driven approach that allows a pair of secondary users to reliably communicate in underlay mode while keeping the interference at the primary receiver close to its noise floor. The secondary transmitter merely has to transmit its signal twice, at very low power - above the noise floor, but well below the primary’s interference. It is shown here that reliable detection of the secondary signal is possible via canonical correlation analysis (CCA). Theoretical and experimental results reveal the remarkable detection performance of the proposed CCA-based approach, which does not require any cross-network coordination, or even channel state information.

In this paper we analyze the extreme and real efficiency of spatial signal processing in presence of background noise interference when the receivers with non-identical frequency response and phase response are placed at various output blocks of active electronic scanning array: at the antenna elements, at the subarrays (group of antenna elements) and at the main and auxiliary (compensating) channels of adaptive array. After that, we formulate practical recommendations on constructing a system of compensation modules for the airspace surveillance radars.

Photonic signal processing offers a new powerful paradigm for processing high speed signals, due to its inherent advantages of wide bandwidth and immunity to electromagnetic interference. Recent trends are focused on the integration of photonics on platforms that co-exist with electronics. Microwave photonic advances including broadband tunable microwave photonic filters, dispersion-free microwave-optical links, multi-function processors, and high-resolution photonic sensing systems that provide new capabilities for high-performance signal processing, are presented.

We use a new class of optomechanical waveguides to we create strong stimulated Brillouin scattering in silicon waveguides. Harnessing these interactions, we create Brillouin-based narrow-band RF-photonic filters, optical amplifiers, and lasers in silicon photonics.

We present an information-theoretic approach to discrete signal processing called information-optimum signal processing for the example of a wireless receiver comprising LDPC decoding, channel estimation and detection. All operations are replaced by simple lookup tables and all messages which are exchanged between detection stages are unsigned integers. The lookup tables are designed offline using the information bottleneck concept of preserving relevant information. We show that the approach allows for simple signal processing with coarse quantization while achieving virtually the same performance as conventional signal processing approaches with high resolution, e.g. double precision. The contribution of the paper is a tutorial style explanation of the concept as well as an exemplary survey of applications and performance results which illustrate the potential of the concept.

Paper is a very interesting substrate material for sensing systems due to its porous nature, low cost, light weight and biodegradability. Low-cost sensing platforms can be created by combining paper substrates with low-cost printed fabrication methods. Here, recent progress is reviewed in the areas that will be needed to create powerful paper-based sensing systems: pump-free microfluidics to manipulate fluids of interest, biosensors to detect analytes, printed microelectronics for signal processing and novel methods of deploying paper-based systems such as printed gliders.

The paper presents a method to classify electromyographic (EMG) signals according to the postures of rock-paper-scissors by using multi-layer perceptrons (MLPs). The EMGs are first applied to He-Zajac-Levine bilinear activation model and then the output of model is utilized to be inputs of the MLPs. Cross validation method is used to evaluate the classification performance of MLPs and its outcome also shows that accuracy of the proposed method is over 97%.

The vital element of the DSP processor is Multiplier unit. The main objectives of the DSP processor are speed, power, delay and area. These goals have been realized with fixed-width multiplier whose output bits equal to input bits. The fixed-width multiplier implemented DSP processors can be applied for audio signal processing, video signal processing, VLSI signal processing, speech recognition, digital communication, medical imaging, MRI, MP3 and so on. Many researchers are optimizing, the performance of the multiplication process. In this review paper, the technologies to achieve the objectives of the DSP processor have been studied. And also the most recent developments in the multiplier circuit have been discussed. In this paper, first, the brief background of the fixed-width multiplier is outlined. Then, several multiplier architectures proposed for MAC (multiplier-accumulator) presented, narrating their functioning principles and key features. To provide a perception into future research directions, open research issues are discussed at the completion of this paper.

We describe recent progress in integrated microwave photonics in wideband signal processing applications with a focus on the key signal processing building blocks, the realization of monolithic integration, and cascaded photonic signal processing for analog RF photonic links.

In this work, a detailed experimental characterization of GPRS and 3G data transmission energy consumption, in a custom Remote Telemetry Unit (RTU) is presented. This comparison results valuable for the design of low data-rate, battery-powered, RTUs until reliable and wide spread IoT networks specifically developed for this type of application in suburban areas are available (probably not before 2 years). The proposed RTU is aimed at illegal logging detection in protected forests, by acoustic signal processing. The RTU sends daily data and alarms to a host server through the GPRS or 3G network and the Internet, and is powered for 3 years from a single 2200mAh primary battery.

Proper operation of the dry end of a paper machine requires some type of sheet tension control. Present systems use either indirect-speed-control-based schemes, or sensor-based direct tension control utilizing load cells or swing rolls. The indirect schemes are inadequate for most newer machines, and both types of the commonly used sensor-based schemes are invasive, expensive, require either special, or extra rolls, and add complexity to the sheet runs. This paper proposes and evaluates sheet tension estimation using only conventional motor and control signals, present in all newer drive controllers, integrated into a process model observer. This observer-based approach eliminates the need for the invasive tension transducers, and may also be used to replace the inaccurate indirect speed control schemes.

Maintaining low and stable electrode contact impedances is critical for obtaining high quality signals in out-of-the-lab EEG units. Current EEG units measure the impedance of the electrode contacts by injecting an out-of-band, typically 1 kHz, current into the head. This high frequency component is easily isolated from the true EEG to avoid introducing artifacts, but does not give direct information on the contact impedance at the wanted cortical frequencies, typically at 5–30 Hz. This paper investigates two techniques for allowing simultaneous impedance measurements at 5–30 Hz for the first time. One method uses digital processing for removing the EEG artifact that continuous in-band impedance monitoring produces. The other uses a new 36 nW notch filter for removing the interference. Both are shown to allow impedance monitoring at 5–30 Hz while leaving minimal residual artifacts in the collected EEG traces.

All the real time signals are non-periodic and non-stationary, which gives more information for any signal processing techniques. Electrocardiogram (ECG) signal is an example of real time signal. ECG signal gives electrical activities that are useful information about the functioning of heart. ECG signals helps in diagnosing the complex cardiac diseases. ECG signals are low frequency signals and contain lot of clinical information. The important characteristic information called features are extracted from ECG signal and used for medical diagnosis. The denoising of signal becomes important stage in any signal processing technique and lot of scope involved in computer aided diagnosis of heart. Many research work focusing on denoising the signal for extracting important features like Extended Kalman Filter, Wavelet Transformation and Singular Value Decomposition (SVD). The denoising signal processing techniques evaluated using mean square error and signal to noise ratio. In addition, ECG signal is used for person authentication. The proposed paper focuses on the study of different techniques of denoising and an identification of person through an ECG signal.

In this paper a laser scanner based approach for simultaneous detection and tracking of people in an indoor environment is presented. The operation of an autonomous truck, for transporting paper reels in a dynamic environment shared with humans, is considered as the application setting for this work. Here, a human leg detection procedure and an Extended Kalman Filter (EKF) based tracking method are employed for real-time performance. Several experiments with different data sets collected from an autonomous forklift truck in a paper mill warehouse have been performed in an offline situation. The results show how the system is able to detect and track multiple moving people.

This paper presents a signal processing engine aiming to provide the brush function for electronic papers. The engine contains four stages: (a) screening the detected pixel coordinates, (b) interpolating pixels, (c) image halftoning for generating more visual gray levels, and (d) driving the pixels to the target states. The experimental results show that the engine can complete the pixel interpolation as well as produce the gray-tone traces in order to meet the real-time requirement.

As digital technologies advance, it is very easy to forge a paper-based certificate or images to deceive people, and the counterfeits look almost the same as the originals. Because such forgery causes tremendous losses to the society, the demand for automatic and on-line authentication of paper-based images has been increasing. To solve such a problem, watermarking techniques against the “print-and-scan” attack have been studied extensively. In this paper, a print-and-scan resilient watermarking technique operating in the discrete Fourier transform (DFT) domain is proposed. By embedding a set of peaks as a watermark into the selected DFT coefficients, the watermark is robust against the print-and-scan attack. The experimental results demonstrate that the proposed algorithm can correctly detect the embedded watermark from a printed-and-scanned image. Such a system is thus very useful to prevent forgery by providing on-line, real-time, and automatic authentication of paper-based certificates.

This paper presents an application that is designed to control the process of paper manufacturing in order to obtain a constant thickness for the resulting paper. The application teaches a system to control a thickness regulating member used in the manufacture of sheet products. The position of the slice lip is controlled by a number of actuators positioned along the slice lip and connected to it and to the head box. The system is composed of a Siemens S7–300 PLC connected to a Daytronic equipment. Also a TP270 operating panel is connected in order to allow users to view the state of the entire system and also set the position of the actuators.

Neural networks (NNs) have proven to be a very powerful tool both for one-dimensional (1D) and two-dimensional (2D) direction of arrival (DOA) estimation. By avoiding complex and time-consuming mathematical calculations, NNs estimate DOAs almost instantaneously. This feature makes them very convenient for real-time applications. Further, unlike the well known MUSIC algorithm, neural network-based models provide accurate directions without additional calibration procedure of antenna array and a priori knowledge of the number of sources. In this review paper, the results achieved by the research group at the Faculty of Electronic Engineering in Nis are presented. The problem of DOA estimation of narrowband signals impinging upon different configurations of antenna arrays is addressed. Both Multi-Layer Perceptron (MLP) and Radial Basis Function (RBF) neural networks are considered, and their advantages and disadvantages are discussed. To improve the resolution of DOA estimates, sectorization model is introduced. As shown in this work, neural network-based models demonstrate high-resolution localization capabilities and much better efficiency than the MUSIC.

The problem that paper breaking can be caused by many factors in papermaking process, the dynamic transition of paper breaking must be reflected to the change of some random variables including speed of paper machines, consistency of pulp, quantitative and moisture of paper, which is called paper breaking parameters, by way of non-stationary random signals. The non-stationarity of these random signals will boost up rapidly with paper breaking process. According to the phenomenon a new method predicting paper breaking based on analyzing con-stationary property of random signals of paper breaking parameters is presented here. During the prediction process, the non-stationarizing amplitude of these random signals are estimated by the distance of 1.5 dimension cepstrum, and a paper breaking predictor is formed by proportional-integral-derivative algorithm of 1.5 dimension cepstrum of paper breaking random signals on line so that paper breaking can be judged in time from variation of 1.5 dimension cepstrum. The correctness and the practicality of this method are shown by application results.

Following an “unwritten tradition” this paper offers a review of another decade of achievements in the area of Spectral Techniques. Topics like transforms of matrix-valued functions, transforms based on non-Abelian groups, latest advancements related to Linear Independent transforms, a space-efficient algorithm to calculate spectral transforms, will be discussed. A special mention will be given to other areas, like signal processing, and coding theory, where Spectral Techniques have been (and continue to be) used, even though with possibly different names.

The paper presents embedded system design project based on RFID (Radio-Frequency Identification) securities. The project adopts RFID to complete reading and writing tags, and embedded system to implement system control, communication, signal processing, showing results, human-computer interaction, and so on. Module of RFID is made of MFCR500 of Philips, AT89S52 of Atmel, and integrated antenna. The embedded system is made of S3C2410 as kernel chip from Samsung. The whole system integrates portable securities, communication, and distribution management.

This research aims at establishing criteria for the automatic identification of defects in paper/oil insulation systems. Statistical analysis of partial discharge (PD) data collected in a reference database (built up from tests on artificial defects: internal PD, surface PD, corona and discharges due to bubbles in oil) is carried out to highlight the best identification markers. The development of an automatic identification system based on fuzzy logic is presented. Eventually, applications of identification are discussed.

In recent years, the author has been involved in starting up a high-tech company and bringing both the product and the company itself to a successful fruition. The paper describes the author's experiences from this venture. The product in question was a VDSL modem, delivering about 15 megabits-per-second over unshielded telephone wires in full duplex. The paper reviews VDSL technology, its capabilities and limitations, and puts it in the context of the "broadband to the home" trend. The steps involved in turning a paper-concept into a real working product are described, as well as the main challenges and obstacles. The possible future of copper-based broadband technology is discussed and what may come after it. Finally, the author reflects on possible implications of his industry experience on teaching digital signal processing in the future.

The implementation challenge for new low-cost low-power wireless modem transceivers has continuously been growing with increased modem performance, bandwidth, and carrier frequency. Up to now we have been designing transceivers in a way that we are able to keep the analog (RF) problem domain widely separated from the digital signal processing design. However, with today's deep sub-micron technology, analog impairments-dirt effects-are reaching a new problem level which requires a paradigm shift in the design of transceivers. Examples of these impairments are phase noise, non-linearities, IQ imbalance, ADC impairments, etc. In the world of dirty RF we assume to design digital signal processing such that we can cope with a new level of impairments, allowing lee-way in the requirements set on future RF sub-systems. This paper gives an overview of the topic and presents analytical evaluations of the performance losses due to RF impairments as well as algorithms that allow to live with imperfect RF by compensating the resulting error effects using digital baseband processing

The paper describes the analysis of paper machine process data using discrete wavelet transforms. The techniques have been adapted from a general signal analysis theory. The authors previously showed (1996) that wavelets are an effective representation for the detection of basis weight and moisture process variations in noisy data and lead to improved estimation and visualization of the machine direction and cross machine variations. This paper discusses data storage using the wavelet representation, and shows that the method also allows significant compression of the scanned data without diminishing the accuracy with which profiles can be reconstructed. It is shown that the compression method can be embedded into the estimation algorithm, producing excellent results without major expense in computation time. The ability to reduce data storage requirements is of increasing importance in mill-wide process monitoring systems and quality assurance.

Microwave systems continue to benefit from technological and system advances that provide for new levels of performance, form factor and lowered cost. In the past, these advances were primarily associated with new materials systems or device types (e.q. GaAs, GaN, InP, MEMS, HBTs, PHEMTs, etc.). Although improvements continue in these and other new areas, system designers are increasingly looking for areas where advances in CMOS digital IC technology can improve the performance of microwave systems. The ability to realize millions of logic gates in an area the size of a bond pad provides the signal processing capability required for highly adaptive microwave systems, which monitor their environment and adapt their matching, tuning, bias, or configuration, to realize the best possible performance. These adaptive microwave systems will find wide application in areas as diverse as active phased- array radars, wireless local area networks and even the ubiquitous cellular telephone. This paper will summarize the device and signal processing requirements of these new adaptive systems of the future, and highlight some of the research areas required for further development of the field.

The automatic system makes use of mostly generic components like the PC and a desktop colour scanner to perform electronic data capture. Signal processing modules were added to extract relevant statistical and diagnostic information for rapid data abstraction and indexing. The system accepts standard ECG paper records and produces three data files of about 35 KBytes in total consisting of a compressed (lossless) copy of the scanned image, the digitised (recovered) values of the individual leads and a set of features describing the ECG. At each stage of the development, exhaustive tests were carried out to ensure maximum accuracy and speed. The entire operation from scanning to segmentation and signal recovery took less than 2 minutes per record. The system has been installed at the National University Hospital and it incorporates a telemedicine module to classify incoming ECG queries.