22 January 2024 in High-Confidence Computing
5 December 2022 in Scientific Reports
Recurrent incidents of economically motivated adulteration have long-lasting and devastating effects on public health, economy, and society. With the current food authentication methods being target-oriented, the lack of an effective methodology to detect unencountered adulterants can lead to the next melamine-like outbreak. In this study, an ensemble machine-learning model that can help detect unprecedented adulteration without looking for specific substances, that is, in a non-targeted approach, is proposed. Using raw milk as an example, the proposed model achieved an accuracy and F1 score of 0.9924 and 0. 0.9913, respectively, when the same type of adulterants was presented in the training data. Cross-validation with spiked contaminants not routinely tested in the food industry and blinded from the training data provided an F1 score of 0.8657. This is the first study that demonstrates the feasibility of non-targeted detection with no a priori knowledge of the presence of certain adulterants using data from standard industrial testing as input. By uncovering discriminative profiling patterns, the ensemble machine-learning model can monitor and flag suspicious samples; this technique can potentially be extended to other food commodities and thus become an important contributor to public food safety.
15 November 2019 in IEEE Internet of Things Journal
Visitor counting can be attractive to various applications, like business management and marketing investigation. Recently, many studies have employed wireless signals to achieve visitor counting without people's active participation and privacy intrusion. However, existing systems mainly count the overall visitors inside a certain area, which fails to provide the fine-grained information of the coming and leaving visitor flow. Unlike previous studies, this article proposes to count the in-and-out visitors to monitor visiting frequency and population, which can be applied for many indoor places, such as shops and restaurants. Therefore, we present the first WiFi-based in-and-out visitor counting system, Door-Monitor, which obtains the direction (enter or exit) and the number of visitors passing by the door. The WiFi signals enable us to count the visitors in a low-cost and nonintrusive way, and it can tell the exact number of visitors even when multiple persons pass by the door simultaneously. To detect the visitors' passing direction, we show that the patterns in the phase difference series can indicate the entering and exiting passing directions by analyzing the effects of the passing behavior on the signal's phase information. To count the passing visitors, we perform a short time Fourier transformation on the phase difference series to generate the spectrogram, on which the convolutional neural network is applied for building a counting model. The experimental results show that the average accuracies of passing direction detection and visitor counting are 95.2% and 94.5%, respectively.
21 December 2022 in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies
Vibration measurement is vital for fault diagnosis of structures (e.g., machines and civil structures). Different structure components undergo distinct vibration patterns, which jointly determine the structure's health condition, thus demanding simultaneous multi-point vibration monitoring. Existing solutions deploy multiple accelerometers along with their power supplies or laser vibrometers on the monitored object to measure multi-point vibration, which is inconvenient and costly. Cameras provide a less expensive solution while heavily relying on good lighting conditions. To overcome these limitations, we propose a cost-effective and passive system, called Multi-Vib, for precise multi-point vibration monitoring. Multi-Vib is implemented using a single mmWave radar to remotely and separately sense the vibration displacement of multiple points via signal reflection. However, simultaneously detecting and monitoring multiple points on a single object is a daunting task. This is because most radar signals are scattered away from vibration points due to their tilted locations and shapes by nature, causing an extremely weak reflected signal to the radar. To solve this issue, we dedicatedly design a physical marker placed on the target point, which can force the direction of the reflected signal towards the radar and significantly increase the reflected signal strength. Another practical issue is that the reflected signal from each point endures interferences and noises from the surroundings. Thus, we develop a series of effective signal processing methods to denoise the signal for accurate vibration frequency and displacement estimation. Extensive experimental results show that the average errors in multi-point vibration frequency and displacement estimation are around 0.16Hz and 14μm, respectively.
22 November 2023 in ACM Transactions on Internet of Things
Early detection of fatigue driving is pivotal for the safety of drivers and pedestrians. Traditional approaches mainly employ cameras and wearable sensors to detect fatigue features, which are intrusive to drivers. Recent advances in radio frequency (RF) sensing enable non-intrusive fatigue feature detection from the signal reflected by driver’s body. However, existing RF-based solutions only detect partial or coarse-grained fatigue features, which reduces the detection accuracy. To tackle the above limitations, we propose a mmWave-based fatigue driving detection system, called mmDrive, which can detect multiple fine-grained fatigue features from different body parts. However, achieving accurate detection of various fatigue features during driving encounters practical challenges. Specifically, normal driving activities and driver’s involuntary facial movements inevitably cause interference to fatigue features. Thus, we exploit unique geometric and behavioral characteristics of fatigue features and design effective signal processing methods to remove noises from fatigue-irrelevant activities. Based on the detected fatigue features, we further develop a fatigue determination algorithm to decide the driver’s fatigue state. Extensive experiment results from both simulated and real driving environments show that the average accuracy for detecting nodding and yawning features is about 96%, and the average errors for estimating eye blink, respiration, and heartbeat rates are around 2.21 bpm , 0.54 bpm , and 2.52 bpm , respectively. And the accuracy of the fatigue detection algorithm we proposed reached 97.63%.