Automated electrocardiogram (ECG) classification using machine learning (ML) is extensively utilized for arrhythmia detection. Contemporary ML algorithms are typically deployed on the cloud, which may not always meet the availability and privacy requirements of ECG monitoring. Edge inference is an emerging alternative that overcomes the concerns of cloud inference; however, it poses new challenges due to the demanding computational requirements of modern ML algorithms and the tight constraints of edge devices. In this work, we propose a tiny convolutional neural network (CNN) classifier for real-time monitoring of ECG at the edge with the aid of the matched filter (MF) theory. The MIT-BIH dataset with inter-patient division is used for model training and testing. The model generalization capability is validated on the INCART, QT, and PTB diagnostic databases, and the model performance in the presence of noise is experimentally analyzed. The proposed classifier can achieve average accuracy, sensitivity, and F1 scores of 98.18%, 91.90%, and 92.17%, respectively. The sensitivity of detecting supraventricular and ventricular ectopic beats (SVEB and VEB) is 85.3% and 96.34%, respectively. The model is 15 KB in size, with an average inference time of less than 1 ms. The proposed model achieves superior classification and real-time performance results compared to the state-of-the-art ECG classifiers while minimizing the model complexity. The proposed classifier can be readily deployed on a wide range of resource-constrained edge devices for arrhythmia monitoring, which can save millions of cardiovascular disease patients.

Automated classification of Electrocardiogram (ECG) for arrhythmia monitoring is the core of cardiovascular disease diagnosis. Machine Learning (ML) is widely used for arrhythmia detection. The cloud-based inference is the prevailing deployment model of modern ML algorithms which does not always meet the availability and privacy requirements of ECG monitoring. Edge inference is an emerging alternative that addresses the concerns of latency, privacy, connectivity, and availability. However, edge deployment of ML models is challenging due to the demanding requirements of modern ML algorithms and the computation constraints of edge devices. In this work, we propose a lightweight self-contained short-time Fourier Transform (STFT) Convolutional Neural Network (CNN) model for ECG classification and arrhythmia detection in real-time at the edge. We provide a clear interpretation of the convolutional layer as a Finite Impulse Response (FIR) filter and exploit this interpretation to develop an STFT-based 1D convolutional (Conv1D) layer to extract the spectrogram of the input ECG signal. The Conv1D output feature maps are reshaped into a 2D heatmap image and fed to a 2D convolutional (Conv2D) neural network (CNN) for classification. The MIT-BIH arrhythmia database is used for model training and testing. Four model variants are trained and tested on a cloud machine and then optimized for edge computing on a raspberry-pi device. Weight quantization and pruning techniques are applied to optimize the developed models for edge inference. The proposed classifier can achieve up to 99.1% classification accuracy and 95% F1-score at the edge with a maximum model size of 90 KB, an average inference time of 9 ms, and a maximum memory usage of 12 MB. The achieved results of the proposed classifier enable its deployment on a wide range of edge devices for arrhythmia monitoring.

Convolutional layers (CLs) are ubiquitous in contemporary deep neural network (DNN) models, commonly used for automatic feature extraction. A CL performs cross-correlation between the input to the layer and a set of learnable kernels to produce the layer output. Typically, kernel weights are randomly initialized and automatically learned during model training using the backpropagation and gradient descent algorithms to minimize a specific loss function. Modern DNN models comprise deep hierarchical stacks of CLs and pooling layers. Despite their prevalence, CLs are perceived as a magical tool for feature extraction without solid interpretations of their underlying working principles. In this work, we advance a method for designing and analyzing CLs by providing novel signal processing interpretations of the CL by exploiting the correlation and equivalent convolution functions of the layer. The proposed interpretations enable the employment of CLs to develop finite impulse response (FIR) filters, matched filters (MFs), short-time Fourier transform (STFT), discrete-time Fourier transform (DTFT), and continuous wavelet transform (CWT) algorithms. The main idea is to pre-assign the CL kernel weights to implement a specific convolution- or correlation-based DSP algorithm. Such an approach enables building self-contained DNN models in which CLs are utilized for various preprocessing and feature extractions tasks, enhancing the model portability, and cutting down the preprocessing computational cost. The proposed DSP interpretations provide an effective means to analyze and explain the operation of automatically trained CLs in the time and frequency domains by reversing the design procedures. The presented interpretations are mathematically established and experimentally validated with a comprehensive machinery fault diagnosis application example illustrating the potential of the proposed methodology.

Security is a major challenge preventing wide deployment of the smart grid technology. Typically, the classical power grid is protected with a set of isolated security tools applied to individual grid components and layers ignoring their cross-layer interaction. Such an approach does not address the smart grid security requirements because usually intricate attacks are cross-layer exploiting multiple vulnerabilities at various grid layers and domains. We advance a conceptual layering model of the smart grid and a high-level overview of a security framework, termed CyNetPhy, towards enabling cross-layer security of the smart grid. CyNetPhy tightly integrates and coordinates between three interrelated, and highly cooperative real-time security systems crossing section various layers of the grid cyber and physical domains to simultaneously address the grid's operational and security requirements. In this article, we present in detail the physical security layer (PSL) in CyNetPhy. We describe an attack scenario raising the emerging hardware Trojan threat in process control systems (PCSes) and its novel PSL resolution leveraging the model predictive control principles. Initial simulation results illustrate the feasibility and effectiveness of the PSL.

Time series classification is an active research topic due to its wide range of applications and the proliferation of sensory data. Convolutional neural networks (CNNs) are ubiquitous in modern machine learning (ML) models. In this work, we present a matched filter (MF) interpretation of CNN classifiers accompanied by an experimental proof of concept using a carefully developed synthetic dataset. We exploit this interpretation to develop an MF CNN model for time series classification comprising a stack of a Conv1D layer followed by a GlobalMaxPooling layer acting as a typical MF for automated feature extraction and a fully connected layer with softmax activation for computing class probabilities. The presented interpretation enables developing superlight highly accurate classifier models that meet the tight requirements of edge inference. Edge inference is emerging research that addresses the latency, availability, privacy, and connectivity concerns of the commonly deployed cloud inference. The MF-based CNN model has been applied to the sensor-based human activity recognition (HAR) problem due to its significant importance in a broad range of applications. The UCI-HAR, WISDM-AR, and MotionSense datasets are used for model training and testing. The proposed classifier is tested and benchmarked on an android smartphone with average accuracy and $F1$ scores of 98% and 97%, respectively, which outperforms state-of-the-art HAR methods in terms of classification accuracy and run-time performance. The proposed model size is less than 150 KB, and the average inference time is less than 1 ms. The presented interpretation helps develop a better understanding of CNN operation and decision mechanisms. The proposed model is distinguished from related work by jointly featuring interpretability, high accuracy, and low computational cost, enabling its ready deployment on a wide set of mobile devices for a broad range of applications.

On-chip interconnects are the performance bottleneck in modern system-on-chips. Code-division multiple access (CDMA) has been proposed to implement on-chip crossbars due to its fixed latency, reduced arbitration overhead, and higher bandwidth. In CDMA, medium sharing is enabled in the code space by assigning a limited number of ${N}$ -chip length orthogonal spreading codes to the processing elements sharing the interconnect. In this paper, we advance overloaded CDMA interconnect (OCI) to enhance the capacity of CDMA network-on-chip (NoC) crossbars by increasing the number of usable spreading codes. Serial and parallel OCI architecture variants are presented to adhere to different area, delay, and power requirements. Compared with the conventional CDMA crossbar, on a Xilinx Artix-7 AC701 FPGA kit, the serial OCI crossbar achieves 100% higher bandwidth, 31% less resource utilization, and 45% power saving, while the parallel OCI crossbar achieves ${N}$ times higher bandwidth compared with the serial OCI crossbar at the expense of increased area and power consumption. A 65-node OCI-based star NoC is implemented, evaluated, and compared with an equivalent space division multiple access based torus NoC for various synthetic traffic patterns. The evaluation results in terms of the resource utilization and throughput highlight the OCI as a promising technology to implement the physical layer of NoC routers.