Wearable sensors are increasingly used for continuous health monitoring, but their small size limits battery capacity, affecting user experience and monitoring capabilities. To overcome this, we introduce an ultra-low power analog Folded Neural Network (FNN) for physiological signal processing in a batteryless fashion. Our proposed FNN, by serializing computation, provides several benefits over traditional analog implementations, such as lower space, lower power consumption, and lower peak-to-average power ratio. We evaluate our method extensively using a dataset designed for ECG-based screening and diagnosis. Our analysis considers factors such as thermal noise, spatial requirements, and power consumption. Additionally, we evaluate detection performance, investigating various parameters of the proposed FNN. This evaluation provides insights into the optimal configuration for accurate anomaly detection. We observe a good trade-off for accuracy around 6 layers and a hidden size of 30 and further demonstrate that such architecture could be implemented in a wearable device and executed in a batteryless fashion.

This paper introduces a novel, lightweight, on-board approach to crowd pattern identification, ingeniously using the processes of existing video compression standards, particularly H.264 or MPEG-4. Piggy-backing on the H.264 video-encoding algorithm, we propose real-time crowd pattern recognition and identification methodologies that can identify macroscopic patterns in as low as 2 milliseconds on NVIDIA TX2, resulting in around 45 x execution time reduction compared to existing approaches. Furthermore, we introduce a temporally aware approach to pinpoint and adapt to crowd movement patterns, continuously recalibrating as a drone's Point Of View (POV) varies or observed motions diverge. Evaluating our method against publicly available datasets, we emphasize our system's performance and computational advantages, especially when faced with real-time observational shifts. In conclusion, our approach elegantly bridges the gap between crowd safety imperatives and the challenges of UAV monitoring, heralding a new era of real-time drone-centric crowd management intelligence.

In this article, we introduce a groundbreaking approach for ultra-low-power hybrid analog-digital processing of multimodal physiological data at multiple locations, emphasizing EEG signals. We propose an innovative analog Convolutional Processing Unit (CvPU) that uniquely harnesses the properties of anisotropic diffusion in electrical circuits for convolution. This novel use of anisotropic diffusion-driven convolution sets our work apart. Additionally, we present a controller architecture that allows for the sequential execution of multiple consecutive convolutional layers using the same CvPU array. The proposed neural network architecture to detect seizures using EEG signals is evaluated on a publicly available clinical dataset. Our CvPU array-based convolution’s performance and feasibility metrics have been assessed using SPICE simulation software. Furthermore, we have delved deep into studying the scalability of our approach in terms of power and space and its feasibility for battery-less and implantable applications and have compared it with both digital and hybrid analog-digital methods.

With the increase in the use of unmanned aerial vehicles (UAVs)/drones, it is important to detect and identify causes of failure in real time for proper recovery from a potential crash-like scenario or postincident forensics analysis. The cause of crash could be either a fault in the sensor/actuator system, a physical damage/attack, or a cyber attack on the drone's software. In this work, we propose novel architectures based on deep convolutional and long short-term memory neural networks to detect (via autoencoder) and classify drone misoperations based on real-time sensor data. The proposed architectures are able to learn high-level features automatically from the raw sensor data. Empirical results show that our solution is able to detect (with over 90% accuracy) and classify various types of drone misoperations [with about 99% accuracy (simulation data) and up to 85% accuracy (experimental data)]. Furthermore, with the help of field programmable gate array-based hardware acceleration, we achieved a speedup of 40x ($\sim\!\text{2.6 ms}$) for detection, while consuming half the amount of power compared with onboard GPU devices, such as NVIDIA Jetson TX2.

With the rapid growth of the use of Machine Learning (ML) techniques in Unmanned Aerial Vehicles (UAVs), there is an opportunity to use ML techniques to detect and prevent anomalous behavior in drones. However, limited drone power thwarts successful implementation of contemporary power-hungry ML techniques. Therefore, we propose a hybrid analog-digital system to solve the problem of continuous anomaly detection. In this paper, a series of pure analog ML methods including SVMs (linear, polynomial, Radial Basis Function (RBF) and Sigmoid kernels) as well as pure analog fully-connected Neural Network (NN) are presented. We validate our method with sensor data from a series of drone experiments to detect and identify causes of failure in real-time. The results show that RBF kernel provides at least 88.17 % and at most 99.99 % accuracy under different time window and crash-like scenarios with an extremely low False Negative (FN) ratio with sensor data especially in the z-axis.

Recently, much progress has been made in using Neural Networks (NNs) for important yet narrowly focused tasks such as image classification (e.g., VGG-Net, ResNet), playing complex games like GO or other Computer Vision (CV) tasks. While these achievements are impressive, they are either achieved on computers with virtually unlimited resources or with little regard to real-time actionability. In this paper, we propose to combine the ubiquity of low-resource mobile devices, e.g., drones, with approximate- and distributed-computing techniques in order to make these NN techniques deployable on resource-constrained devices as well as to provide realtime information about the environment. We target situational awareness, which involves sensing the crucial factors in a new environment on a real-time basis. Specifically, we introduce intelligence to a team of drones in the form of real-time detection of a suspect/weapon using local resources and suspect identification in an emergency situation. We validate our proposed methods using Microsoft AirSim simulator via both simulations and hardware-in-the-loop emulations.