A cyber-physical systems often consists of a large number of interacting physical and information components, e.g., a patient-care system may link an increasingly popular and sophisticated patient monitoring system with a network of patient/medical information and an emergency handling system; a transportation system may link a transportation monitoring network, consisting of many sensors and video cameras, with a traffic information/control system; and a battlefield commander system may link a sensor/reconnaissance network with a battle information/analysis system. These systems share an important common feature: they are networked cyber-physical systems, i.e., individual agents or components interact with a set of physical and information components, forming large, interconnected, and heterogeneous cyber-physical networks. Without loss of generality, we call such interconnected cyber-physical networks as cyber-physical information networks. Clearly, cyber-physical information network is ubiquitous and form a critical component of modern information infrastructure. However, despite their importance and growing prevalence in our world, we have only recently recognized the critical importance of studying cyber-physical information networks as an integrated research theme.
In this project, we investigate the foundations, design principles, methodologies, and scalable and effective algorithms of cyber-physical information networks based on a medical application, with an emphasis on four major research challenges for design and construction of cyber-physical information networks:
Theme 1. Rare event detection and analysis in cyber-physical data streams
Theme 2. Reliable and trusted data analysis with cyber-physical information networks
Theme 3. Spatiotemporal analysis in cyber-physical information networks, and
Theme 4. Privacy and security in cyber-physical information networks.
Architecture of cyber-physical information network.
Current Research Findings:
Our research findings can be categorized into two areas: (i) CPS infrastructure and application development, and (ii) data mining. To date, at UVA most of our results are in category (i), while most of the results for (ii) are with our collaborator UIUC. However, we are jointly investigating the integration of the data mining results with the application area of home health care. That is, we are building the CPS infrastructure and application requirements (as originally proposed) to be able to understand new data mining requirements and to be able to test new data mining solutions in a home health care application. In general, this includes improving data mining to handle human behavior patterns, time durations of activities, real-time sensor streams, and rare events that might indicate medical problems.
In one part of our work, we evaluated the use of height for biometric identification of residents, by mounting ultrasonic distance sensors above the doorways in a home. Height sensors are cheap, are convenient for the residents, are simple to install in an existing home, and are perceived to be less invasive than cameras or microphones. Height is typically only a weak biometric, but we show that it is well suited for identifying among a few residents in the home, and can potentially be improved by using the history of height measurements at multiple doorways in a tracking approach. We evaluated this approach using 20 people in a controlled laboratory environment and by installing in 3 natural, home environments. We combined these results with public anthropometric data sets that contain the heights of residents in 2077 elderly multi-resident homes to conclude that height sensors could potentially achieve at least 95% identification accuracy in 95% of elderly homes in the US.
Sleep monitoring is very important for elderly people since inadequate and irregular sleep are often related to serious diseases such as depression and diabetes. In many cases, it is necessary to monitor the body positions and movements made while sleeping because of their relationships to particular diseases (i.e., sleep apnea and restless legs syndrome). Analyzing movements during sleeping also helps in determining sleep quality and irregular sleeping patterns. This work developed a sleep monitoring system based on the WISP platform - active RFID-based sensors equipped with accelerometers. We showed how our system accurately infers fine-grained body positions from accelerometer data collected from the WISPs attached to the bed mattress. Movements and their duration are also detected by the system. We performed an empirical study of 10 subjects on 3 different mattresses in controlled experiments to show the accuracy of our inference algorithms. We evaluated the accuracy of the movement detection and body position inference for 6 nights on 1 subject, and compared these results with two baseline systems: one that uses bed pressure sensors and the other is an iPhone based application.
Many applications of Body Sensor Networks (BSNs) have emerged during the last few years. In the medical field, BSNs are commonly used for detecting health-detrimental accidents and monitoring health status. For most of these applications, reliable communication is essential: an accident detection application needs to send the emergency alarm in real time to healthcare professionals for help; a health monitoring application needs to log a large amount of physiological data continuously. However, BSNs have two important characteristics that impede achieving reliable communication: radio attenuation caused by the impermeability of the human body, and the highly variable operation environment caused by the mobility of BSNs, especially the gather and scatter of BSNs. In this work we performed a comprehensive empirical study to investigate important factors that affect link qualities in terms of Packet Reception Ratio (PRR) in BSNs. Both the body shadowing effect and the interference within one BSN and between multiple BSNs are examined by changing various factors such as the transmission power, the sensor placement, and the distance between two BSNs, etc. Based on the study, we proposed several approaches to guarantee reliable communication for BSNs, including allocating different frequency channels for nearby BSNs, a cross-body TDMA based MAC protocol, and a quality of service (QoS) algorithm to guarantee the delivery of the most important data when bandwidth is inadequate. All approaches are fully implemented and tested for 5 BSNs, and the results show that using our solution BSNs can achieve reliable communication.
To date our work concentrates on the medical domain. In this domain we have previously built a testbed to emulate WSNs in home health care and assisted living facilities. This system is called AlarmNet. We are working with the UVA medical school. See also Wireless Networking for Assisted Living and Other Medical Applications. See also Hardware We Developed for Medical Applications.
We have also created a Center for Wireless Health. Information can be found at Center for Wireless Health.
This project is supported by the National Science Foundation under Grant No.
CISE CPS 0931972. Any opinions, findings and conclusions or recommendations expressed
in this material are those of the author(s) and do not necessarily
reflect the views of the National Science Foundation (NSF).
R. Rajkumar, I. Lee, L. Sha, and J. Stankovic, Cyber Physical Systems: The Next Computing Revolution, invited, Design Automation Conference, 2010.