Projects
ULISSES – Air sensors for everyone, everwhere
I am currently involved in the Horizon 2020 European Union Project Ulisses, where I work for the work package (WP) 5 'Sensor intelligence at the IoT edge'.
For this WP, we first develop mathematical models of the physical processes causing sensor aging and drift. In more detail, we build a stochastic model which characterize the statistical relationship between sensor parameters, sensor measurements, and the environmental factors. Based on the model, we do time series analysis and prediction for the sensor drift. Calibration is performed based on the standard statistical inference tools under hidden Makrov model.
[1] Y. You and T. J. Oechtering, "Hidden Markov Model Based Data-driven Calibration of Non-dispersive Infrared Gas Sensor," in 28TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO 2020), 2020, pp. 1717-1721.
Along this line, we further study the networked calibration mechanisms using belief function fusion approach and the reinforcement learning approaches. One of our featured recent works is the Deep Q-network based calibration for NDIR CO2 sensor networks.
[2] Y. You, A. Xu and T. J. Oechtering, "Belief Function Fusion based Self-calibration for Non-dispersive Infrared Gas Sensor," in 2020 IEEE SENSORS, 2020, pp. 1-4.
[3] Y. You, K. You, H. Chen, and T. J. Oechtering, “On Data-Driven Self-Calibration for IoT-Based Gas Concentration Monitoring Systems”, IEEE Internet of Things journal， pp. 1-1, 2022.
Two filed patents:
Y. You and T. J. Oechtering “Method for determining a gas concentration from a group of sensors,” Swedish patent (Patent #: 2051034-3).
C. Yang, T. J. Oechtering, and Y. You “Method for updating baseline calibration parameter,” filed June 2021.
Learn more at: https://www.ulisses-project.eu/
COPES - COnsumer-centric Privacy in smart Energy gridS
From Oct 2017 to Jan 2019, I was involved in the European CHIST-ERA project COPES, where we focus on the design of privacy-preserving and cost-efficient energy management strategies for smart grid consumers.
We first model the adversary's attacking behavior as hypothesis testing. And we further propose to use KL-divergence as the privacy leakage measure, and the expected cost-saving as the consumers' utility measure. The problem is formulated into a partially observed Markov decision process, and the reinforcement learning algorithms are utilized to solve the proposed MDP problem.
[1]. Y. You, Z. Li and T. J. Oechtering, "Optimal Privacy-Enhancing and Cost-Efficient Energy Management Strategies for Smart Grid Consumers," in 2018 IEEE Statistical Signal Processing Workshop, SSP 2018, 2018, pp. 144-148.
[2]. Y. You, Z. Li and T. J. Oechtering, "Energy Management Strategy for Smart Meter Privacy and Cost Saving," IEEE Transactions on Information Forensics and Security, vol. 16, pp. 1522-1537, 2021.
As an extension, we further study the attacking behavior which is known as non-intrusive load monitoring. Specifically, we assume the adversary infer on the consumers' energy consumption behavior based on the statistical inference under the factorial hidden Markov model. As the first attempt, we propose an online learning based solution for the privacy-preserving for single consumer. And we further study the privacy and cost trade-off problem under a non-cooperative game framework for multiple consumers.
[3] Y. You and T. J. Oechtering, "Online Energy Management Strategy Design for Smart Meter Privacy Against FHMM-based NILM," in 2020 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm), 2020.
[4] Y. You, Z Li, and T. J. Oechtering. ''Non-cooperative Games for Privacy-preserving and Cost-efficient Smart Grid Energy Management'', submitted to IEEE Transactions on Information Forensics and Security.