1,720,998 research outputs found
Autoencoder based Physical Layer Authentication for UAV Communications
No full textThe use of flying Unmanned Aerial Vehicles (UAVs) for communications is becoming more and more widespread, especially in 5G and beyond networks. In such a context, detection and authentication of UAVs is assuming an increasingly important role. In this paper we show that it is possible to distinguish different drones which communicate with a fixed ground base station (BS) on the basis of their channel characteristics and of the micro-Doppler signature associated to the specific features of each UAV. An urban scenario is simulated where UAVs fly at a constant height and channels are affected by Additive White Gaussian Noise (AWGN) and fading. With the aim of helping the BS in its authentication task, we take advantage of a sparse autoencoder trained on the channel of the legitimate transmitter, while data coming from possible attackers are classified as anomalies. We prove that, with proper network training, low levels of false alarm and missed detection can be achieved, especially if the attacker has no line-of-sight link, and that the presence of micro-Doppler actually contribute to enhance the authentication performance
Towards a Smart Extractor Hood to Improve Indoor Air Quality in Home Living Environments
No full textIndoor air quality is a key determinant of a healthy life and personal well-being. Indeed, unlike outdoor atmospheric air, quality and pollution of indoor air are not subject to specific regulations in various countries, while it is of paramount importance to get consciousness about sources and type of pollutants that have direct effects on human health. As reported in the literature, the contribution of emitted pollutants from cooking is estimated to be between 12% and 20% of the total pollutants found in indoor air composition. At the same time, kitchens or living rooms are among the spaces where people spend most of their time at home. In this paper, the possibility to turn an extractor hood into a smart autonomous system to improve indoor air quality in home living environments is explored. The study focuses on the evaluation of two different multi-sensor units, equipped with the capability to monitor temperature, relative humidity, volatile organic compounds, and particulate matter, in sensing pollutants generated by food overcooking. The experimental tests carried out in a realistic condition, by repeated overcooking of five different types of food, show a good correlation among the signals collected from the two sensing units, expressed by the coefficient of determination which is found to be typically greater than 0.8 (compared to a maximum value of 1), and repeatability of the sensor measurements provided by the two systems. This confirms the possibility to use the acquired values to control the activation of the extractor hood, when the level of detected pollutants is too high
Autoencoder-based physical layer authentication in a real indoor environment
Full text linkAuthentication of wireless nodes, as in fifth-generation (5G) and Internet of Things (IoT) networks, is an increasingly pressing issue, in order to limit the required computational effort and the necessary overhead. A simplification of the authentication process may therefore be of interest to achieve the satisfaction of stringent performance requirements, such as those envisaged for sixth-generation (6G) networks. This paper provides a study on the feasibility of physical layer authentication (PLA) in a real indoor environment, as an alternative solution to the traditional authentication schemes. To ensure the reliability of the proposed approach a simulated scenario is firstly tested. Subsequently, real-world data are collected through a laboratory setup using a Vectorial Signal Transceiver (VST) and two Universal Software Radio Peripherals (USRPs) to emulate the behavior of the receiver, the legitimate transmitter, and the potential adversary. A machine learning (ML) algorithm is then exploited to act as authenticator. This means that channel fingerprint is extracted from signals to create a dataset used to train a sparse autoencoder. To emulate a real authentication scenario, the autoencoder is trained only on the class of the legitimate user. Once a new message arrives, the autoencoder task is to discern authentic signals from those forged by the adversary. It is shown that a geometric mean of accuracy of more than 90%, with corresponding low levels of false alarm and missed detection, is achievable irrespective of the nodes location, underlining the robustness and versatility of the proposed ML-based PLA approach
Automotive radar application for structural health monitoring
No full textMonitoring of communication and residential infrastructures, such as buildings, bridges and tunnels, has always been important to ensure the safety of citizens. In particular, it is of interest to verify the level of mechanical stress to which bridges and buildings are subjected during earthquakes, measuring the oscillations that the structure undergoes during the seismic event, in order to verify the overcoming of safety thresholds, in which stability may be impaired. In this paper the mmWave radars designed for the automotive world is taken into account, since it allows high precision measurements at a relatively low cost. mmWave radars are able to detect vibrations of the order of tens of microns and, therefore, are very useful for monitoring buildings. Furthermore, using MIMO (Multiple Input Multiple Output) mmWave radars, it is possible to implement the Beamforming technology for the spatial shaping of the radio beam, in order to detect more targets simultaneously. In this way it is possible to monitor multiple buildings or structures simultaneously, or different parts of the same structure with a single sensor. The objective of the present work is to show the applicability of automotive radar to the measurement of oscillations suffered by buildings and bridges, in order to be used as a monitoring tool in the event of earthquakes. Results of software simulations, laboratory test and real measurements on infrastructures are provided
Identification of Smartphone Zombies and Normal Pedestrians Using FMCW Radar and Machine Learning
No full textMobile phone usage represents a source of distraction for pedestrians, who are losing awareness of external hazards given by vehicles and environment. Radars could be a solution to monitor continuously and privately the behaviours of pedestrians in the main public spaces in order to find solutions based on the way pedestrians walk, their habits and their walking speed. Being able to identify a pedestrian with the head down on the phone, usually called 'smartphone zombie'', is crucial to intervene to make the road safer and discourage the behaviour. We study the feasibility of identifying the walking pattern of 'smartphone zombie'' against a control pedestrian walking normally exploiting an automotive frequency modulated continuous wave radar working at 77 GHz. By applying principal component analysis and machine learning we obtain a classification accuracy of 92.4% of smartphone zombies against normal walk and 87.6% when adding a third class of fast walkers
MmWave radar features extraction of drones for machine learning classification
No full textWith the progressive reduction of cost, in the market it is possible to find a very large assortment of Unmanned Aerial Vehicles (UAV) that are used in general for non-warlike activities. Unfortunately, it may happen that malicious subjects use these objects to cause damage or inconvenience, then the availability of solutions to predict these situations can be crucial for alerting the population and saving lives. In this work, we present a technique to identify drones from their micro-Doppler features, by analyzing their variations during the flight. The characterization of the features and how they evolve in time is useful to predict dangerous situations and classify the drone type, with the help of Machine Learning techniques
Measuring UAV Propeller RPM with FMCW Radar: Validation with Calibrated Accelerometers
No full textThe use of Unmanned Aerial Vehicles (UAVs) in non-military applications has become more widespread in recent years. Safety concerns during their operation have also increased. For this reason, developing detection techniques targeting such devices is an area of scientific interest. The detection of drones can be achieved through the use of various sensors, including optical (video, or Light Detection And Ranging, or electromagnetic sensors. By resorting to the latter family of sensing technologies, different types of information about the UAV can be collected, which motivates the focus of this work on Radar sensors. Machine Learning and Deep Learning approaches may allow to classify the type of UAV, but more quantitative figures can be obtained as well, from the Radar signals. Among them, the rotational speed of a UAV propeller was already quantified by a Frequency Modulated Continuous Wave Radar, but the technique used can be investigated more in-depth to better understand the interaction with chassis vibrations, to evaluate the accuracy of the obtained values. To this aim, a series of tests are carried out on a mockup quadcopter. The results output by the Radar, are compared to the values provided by a calibrated accelerometer, showing that the mean vibration frequency is exactly measured, while a difference in the order of tens of micrometers is found on the mean vibration displacement. These outcomes prove that the vibration detected by the Radar is actually relatable to the rotational speed of the UAV propeller
ML-Based Edge Node for Monitoring Peoples’ Frailty Status
Full text linkThe development of contactless methods to assess the degree of personal hygiene in elderly people is crucial for detecting frailty and providing early intervention to prevent complete loss of autonomy, cognitive impairment, and hospitalisation. The unobtrusive nature of the technology is essential in the context of maintaining good quality of life. The use of cameras and edge computing with sensors provides a way of monitoring subjects without interrupting their normal routines, and has the advantages of local data processing and improved privacy. This work describes the development an intelligent system that takes the RGB frames of a video as input to classify the occurrence of brushing teeth, washing hands, and fixing hair. No action activity is considered. The RGB frames are first processed by two Mediapipe algorithms to extract body keypoints related to the pose and hands, which represent the features to be classified. The optimal feature extractor results from the most complex Mediapipe pose estimator combined with the most complex hand keypoint regressor, which achieves the best performance even when operating at one frame per second. The final classifier is a Light Gradient Boosting Machine classifier that achieves more than 94% weighted F1-score under conditions of one frame per second and observation times of seven seconds or more. When the observation window is enlarged to ten seconds, the F1-scores for each class oscillate between 94.66% and 96.35%
Physiological Parameters Extraction by Accelerometric Signal Analysis During Sleep
No full textSleep quality is an index of well-being, since sleep disorders, such as sleep apnea, may constitute a health risk. A constant monitoring of subjects, especially when there are heart or respiratory diseases, is essential. The present paper aims to offer a non-invasive and comfortable sleep monitoring, by employing a BallistoCardioGraphic (BCG) signal processing. In particular, with a BCG device located below the mattress, we are able to extract the heart rate, respiratory rate and, therefore, to exploit this information to develop an automatic sleep apnea recognition algorithm. The automatic approach presented has proven to achieve accuracy and reliability and could represent a valid resource to prevent serious damages during sleep
- …
