27 March 2024 in IEEE Transactions on Instrumentation and Measurement
Data-driven method for inertial navigation and positioning has absorbed attention in recent years and it outperforms all its competitor methods in terms of accuracy and efficiency. This paper introduces a new neural architecture framework called IMUNet which is accurate and efficient for inertial position estimation on the edge device implementation receiving a sequence of raw Inertial Measurement Unit (IMU) measurements. The architecture has been compared with the one-dimension version of the state-of-the-art Convolutional Neural Network (CNN) networks that have been introduced recently for edge device implementation in terms of accuracy and efficiency. Moreover, a new method of collecting a dataset using IMU sensors on cell phones and Google ARCore Application Programming Interface (API) has been proposed and a publicly available dataset has been recorded. A comprehensive evaluation using four different datasets as well as the proposed dataset has been done to prove the performance of the proposed architecture. Our experiments show that the proposed framework outperforms state-of-the-art CNN networks in terms of efficiency on a variety of datasets while preserving the accuracy. All the code in both Pytorch and Tensorflow framework as well as the Android application code for data collection has been shared to improve further research.
26 April 2021 in Medical & Biological Engineering & Computing
1 February 2020 in Measurement
4 November 2021 in IEEE Transactions on Emerging Topics in Computational Intelligence
Applications of machine learning (ML) models and convolutional neural networks (CNNs) have been rapidly increased. Although state-of-the-art CNNs provide high accuracy in many applications, recent investigations show that such networks are highly vulnerable to adversarial attacks. The black-box adversarial attack is one type of attack that the attacker does not have any knowledge about the model or the training dataset, but it has some input data set and their labels. In this paper, we propose a novel approach to generate a black-box attack in sparse domain whereas the most important information of an image can be observed. Our investigation shows that large sparse (LaS) components play a critical role in the performance of image classifiers. Under this presumption, to generate adversarial example, we transfer an image into a sparse domain and put a threshold to choose only $k$ LaS components. In contrast to the very recent works that randomly perturb $k$ low frequency (LoF) components, we perturb $k$ LaS components either randomly (query-based) or in the direction of the most correlated sparse signal from a different class. We show that LaS components contain some middle or higher frequency components information which leads fooling image classifiers with a fewer number of queries. We demonstrate the effectiveness of this approach by fooling six state-of-the-art image classifiers, the TensorFlow Lite (TFLite) model of Google Cloud Vision platform, and YOLOv5 model as an object detection algorithm. Mean squared error (MSE) and peak signal to noise ratio (PSNR) are used as quality metrics. We also present a theoretical proof to connect these metrics to the level of perturbation in the sparse domain.
7 February 2022 in IEEE Access
In the areas of communications engineering and biomedical engineering, cloud computing for storing data and running complex algorithms have been steadily increasing due to the increase in internet of things and connected health. As connected IoT devices such as wearable ECG recorders generally have less storage and computational capacity, acquired signals get sent to a remote center for storage and possible analysis on demand. Recently, compressive sensing has been used as a secure, energy-efficient and fast method of signal sampling in such recorders. In this paper, we propose a secure procedure to shift away the total recovery of compressively sensed measurement to cloud and introduce a privacy-assured signal recovery technique in the cloud. We present a fast, and lightweight encryption for secure CS recovery outsourcing that can be used in wearable devices, such as ECG Holter monitors. In the proposed technique, instead of full recovery of CS-compressed ECG signal in the cloud, to preserve privacy, an encrypted version of ECG signal is recovered by using a randomly bipolar permuted measurement matrix. The user with a key, decrypts the encrypted ECG from the cloud to obtain the original ECG signal at their end. We demonstrate our proposed method using the ECG signals available in the MIT-BIH Arrhythmia Database. We also demonstrate the strength of the proposed method against partial exposure of the key. Experimental results on client and cloud sides show our proposed method has lower complexity and consuming time compared to the recent related works, while maintaining the quality of outsourcing task in cloud.