REV Journal on Electronics and Communications
Not a member yet
230 research outputs found
Sort by
A Multitask Data-Driven Model for Battery Remaining Useful Life Prediction
Get PDFLithium-ion batteries (LIBs) have recently been used widely in moving devices. Understand status of the batteries can help to predict the failure and improve the effectiveness of using them. There are some lithium-ion information that define the battery health over time. These are state-of-charge (SOC), state-of-health (SOH), and remaining-useful-life (RUL). Normally, a LIB is working under charging and discharging cycles continuously. In this paper, we will focus on the data dependency of different time-slots in a cycle and in a sequence of cycles to retrieve RUL. We leverage multi-channel inputs such as temperature, voltage, current and the nature of peaks cross the cycles to improve our prediction. Comparing to existing methods, the experiments show that we can improve from 0.040 to 0.033 (reduce 17.5%) in RMSE loss, which is significant
An FPGA-based Convolution IP Core for Deep Neural Networks Acceleration
Get PDFThe development of machine learning has made a revolution in various applications such as object detection, image/video recognition, and semantic segmentation. Neural networks, a class of machine learning, play a crucial role in this process because of their remarkable improvement over traditional algorithms. However, neural networks are now going deeper and cost a significant amount of computation operations. Therefore they usually work ineffectively in edge devices that have limited resources and low performance. In this paper, we research a solution to accelerate the neural network inference phase using FPGA-based platforms. We analyze neural network models, their mathematical operations, and the inference phase in various platforms. We also profile the characteristics that affect the performance of neural network inference. Based on the analysis, we propose an architecture to accelerate the convolution operation used in most neural networks and takes up most of the computations in networks in terms of parallelism, data reuse, and memory management. We conduct different experiments to validate the FPGA-based convolution core architecture as well as to compare performance. Experimental results show that the core is platform-independent. The core outperforms a quad-core ARM processor functioning at 1.2 GHz and a 6-core Intel CPU with speed-ups of up to 15.69× and 2.78×, respectivel
Real-Time Face Detection and Human Tracking System on FPGA Cyclone-V
Get PDFFace detection in image sequence (real-time video stream) has been an active research area in the computer vision field in recent years due to its potential applications such as surveillance cameras, human computer interfaces, smart rooms, intelligent robots and biomedical image analysis. Face detection is a process that determines whether an image has a face or not. In this paper, an embedded system for detecting and tracking human faces in real-time video stream implemented on FPGA DE10-Nano is proposed. The system can be divided into two parts: data streaming, data processing. Experimental results show that the system is capable of accurately detecting faces of up to 5 different people at a distance of up to 1.5 meters from the camera, coexisting in the same frame in resolution of 320 × 240 pixels with a detection speed of only several hundred milliseconds prove the feasibility of the system. A comparison with similar existing projects will be discussed for evaluation and conclusion as well
An efficient hardware implementation of Convolutional Neural Network in detect Breast Cancer Histopathology Image
Get PDFThis paper presents our work on evaluating the effectiveness of a novel deep convolutional neural network architecture (CNN) for classifying breast histology images for cancer risk factors as negative or positive. Also, the hardware structure of the proposed model was successfully synthesized and verified. The results indicate that a CNN trained on a small dataset achieved an overall AUC (Area under ROC Curve - ROC is an acronym for receiver operating characteristic) value of 0.922 across a set of 55505 test images. In addition, the time it takes to classify each image is within 3.8 milliseconds instead of a task that even trained pathologists take hours to complete
Network Coding with Multimedia Transmission and Cognitive Networking: An Implementation based on Software-Defined Radio
Get PDFNetwork coding (NC) is considered a breakthrough to improve throughput, robustness, and security of wireless networks. Although the theoretical aspects of NC have been extensively investigated, there have been only few experiments with pure NC schematics. This paper presents an implementation of NC under a two-way relay model and extends it to two non-straightforward scenarios: (i) multimedia transmission with layered coding and multiple-description coding, and (ii) cognitive radio with Vandermonde frequency division multiplexing (VFDM). The implementation is in real time and based on software-defined radio (SDR). The experimental results show that, by combining NC and source coding, we can control the quality of the received multimedia content in an on-demand manner. Whereas in the VFDM-based cognitive radio, the quality of the received content in the primary receiver is low (due to imperfect channel estimation) yet retrievable. Our implementation results serve as a proof for the practicability of network coding in relevant applications
Protograph LDPC Code Design For LS-MIMO 1-bit ADC Systems
Get PDFRecently, two emerging research topics are protograph low-density parity-check (P-LDPC) and large-scale multi-input multi-output (LS-MIMO) with low-resolution analog-to-digital (ADC) converters (LS-MIMO-LOW-ADC). In these directions, many research works have proposed 1-bit ADC as a good candidate for LS-MIMO systems in order to save both transmission power and circuit energy dissipation. However, we observed that previously reported P-LDPC codes might not have good performance for LS-MIMO systems with 1-bit ADC. Hence, we perform a re-design of the P-LDPC codes for the above systems in this paper. The new codes demonstrate a good coding gain from 0:3 dB at rate 1/2 to 0:5 dB at rate 2/3 in different LS-MIMO configurations with 1-bit ADC
Robust Subspace Tracking Algorithms in Signal Processing: A Brief Survey
Get PDFPrincipal component analysis (PCA) and subspace estimation (SE) are popular data analysis tools and used in a wide range of applications. The main interest in PCA/SE is for dimensionality reduction and low-rank approximation purposes. The emergence of big data streams have led to several essential issues for performing PCA/SE. Among them are (i) the size of such data streams increases over time, (ii) the underlying models may be time-dependent, and (iii) problem of dealing with the uncertainty and incompleteness in data. A robust variant of PCA/SE for such data streams, namely robust online PCA or robust subspace tracking (RST), has been introduced as a good alternative. The main goal of this paper is to provide a brief survey on recent RST algorithms in signal processing. Particularly, we begin this survey by introducing the basic ideas of the RST problem. Then, different aspects of RST are reviewed with respect to different kinds of non-Gaussian noises and sparse constraints. Our own contributions on this topic are also highlighted
Security for Two-Way Untrusted Relay against Constant and Reactive Jamming with Fixed Signals
Get PDFActive attacking in physical-layer security has not been significantly studied while potentially causing serious consequences for the legitimate networks. In this paper, we propose a novel method to estimate and remove the jamming signals from multiple multi-antenna jammers in a two-way relay network with multi-antenna legitimate and relay nodes. We carefully consider the signals in the time slots in order to exploit the repetition of the signals and design the transmitted signals which can work in different cases. The numerical results show that the secrecy maximum achievable sum-rate (MASR) at the legitimate nodes is higher than that of the conventional method when considering the affect of transmit SNR; the number antennas at the legitimate and relay nodes; normalized distance between one legitimate node and the relay; and the vertical coordinate of the relay