Overview
The ten participating students each year will have opportunities to engage in a wide variety of research projects related to smart environments. Here are some examples of the projects that student's can participate over the course of the REU.
Collect and Visualize Inhabitant Data (supervised by Diane Cook and Larry Holder)
REU students will work with faculty mentors to develop tools for automatic collection and visualization of inhabitant activity and health data. These students will use sensors and controllers that provide data through powerline or wireless communication to the software. The students will then assimilate the raw observations and provide features such as a report of observed activities organized by time frame and desired amount of detail, a list of frequent patterns of activity, and a visualization of inhabitant activities. Efficient techniques for collecting data while ensuring security and privacy will be investigated, and initial analysis of the data will be provided through visualization based on software similar to our ResiSim simulator. ResiSim is a residential simulator and graphical user interface that permits experiments to be performed without the requirement for access to actual hardware. The goal here is to provide a software tool with a graphical interface that can simulate the devices in the environment as well as inhabitants operating within the home or other setting.
Inhabitant Modeling and Prediction (supervised by Diane Cook)
A feature that separates smart environments from environments that are simply user-controllable is the ability to model and predict inhabitant behavior. If such a model can be built, the model can be used to customize the environment to achieve goals such as automation, security, or energy efficiency. If the model results in an accurate enough baseline, the baseline can provide a basis for detecting anomalies and changes in inhabitant patterns. The REU students working on this project will implement and compare a number of approaches to modeling and predicting inhabitant activities. We have designed modeling approaches based on text compression algorithms. The students can refine these and compare them with Markov model and neural network-based approaches. They can test the ideas using artificial data and data collected from the smart environment sites.
Pervasively Secured Infrastructures (supervised by Behrooz Shirazi)
This framework makes use of smart sensors, wireless networks, mobile agents, data mining, and profile-based learning in an integrated, collaborative and distributed manner. Through this project we are developing the concept of community computing as a framework for the development of pervasive computing applications. In community computing, goal-oriented software agents working on behalf of lower-end sensors and devices, collaborate with each other to carry out application-specific services. The problems addressed in this project include: (i) data collection and aggregation from heterogeneous, seemingly unrelated sensors; (ii) real-time, secured, authenticated information processing and exchange; and (iii) discovery of smart
environment security threats using data mining, learning and intelligent decision making techniques. This project was initially designed as a homeland security application. The REU-SE participants will design a community computing approach with PSI to provide on-demand security services for smart homes, workplaces, shopping malls, airports, hospitals, and other critical environments.
Automated Decision Making for Smart Environments (supervised by Larry Holder)
While supporting technologies for smart environments have matured, fully automating an environment using these technologies is still a rarity. We propose to investigate technologies for learning automation strategies based on work in AI planning systems, control theory, and machine learning. These algorithms have already been used together in a preliminary study to reduce manual interactions with an automated apartment. Working together with REU students on the other projects, REU students in this project will employ decision making techniques to optimize the environment for energy efficiency, minimal manual interactions, maximization of comfort, and other potential automation goals.
Middleware (supervised by Dave Bakken)
As with any large endeavor, smart environments integrate many components including sensor hardware, modeling and automation software, and controllers. We propose to research operating system and middleware software that facilitates the goal of making environment “smart”. This project will be a good choice for REU students wanting to gain expertise in software engineering and embedded systems. REU students will be designing middleware that reduces the complexities of the network, the operating system, and any other resources required by the smart environment. This middleware should ease the integration of new devices into the environment. Ensuring the privacy and confidentiality of collected data is particularly important in smart environments, and will be researched by REU students as well.
Natural Interfaces for Smart Environments (supervised by Christopher Hundhausen)
Although smart environments designers are encouraged by the progress that has been made in the field, much of this progress will go unused if the technologies are difficult or unnatural for inhabitants. Explicit input must now be replaced with more human-like communication capabilities and with implicit actions. An earlier undergraduate project created a PDA interface for our smart environment, called the Mavigator . We will incorporate interface technologies into smart environments such as motion tracking, gesture recognition, interactive surfaces, and speech processing. REU students will experiment with one or more interactive mechanisms and will evaluate the impact of the interactive tools on the overall experience for inhabitants in smart environments.
Mixed-signal Integrated Circuit Design and Low Power Sensor Circuits (supervised by George La Rue)
Previous research in the MavHome smart environment project led to the development of the “Argus” sensor network module, a low-cost module that permits the networking of a large number of sensors using a combination of I2C, RS232, and Ethernet connections. A number of copies of the final version of these modules were masked and assembled for integration into the physical smart environments. However, a great deal of new research can be performed to improve design of a general sensor network for smart environments.
At WSU, our research into clock and data recovery circuits, track and hold amplifiers (THAs) and
analog-to-digital converters (ADCs) in SiGe technology is supported by CDADIC, the Center for the Design of Analog-Digital Integrated Circuits. In addition, our research on direct digital frequency synthesizers is funded by the Air Force through CDADIC. An innovative 800 MHz radiation-tolerant non-linear digital-to-analog converter for direct digital frequency synthesizer (DDFS) in Honeywell’s MOI CMOS process was successfully designed, fabricated and tested. We are currently designing a 2 GHz DDFS using radiation-hardening-by-design techniques on a commercial CMOS process. We also have a project designing a 12-channel preamplifier IC and a low-power 16-bit ADC for neurosensor applications. Based on simulations, the preamplifier IC is expected to provide state-of-the-art performance in noise, bandwidth and power dissipation. The REU students will gain knowledge of highspeed or low power integrated circuit design, layout and testing. They will learn to use a variety of software tools for laying out ICs, printed circuit board test fixtures, circuit simulation and interfacing with test equipment. The REU students will test the sensor networks for use in smart environments and test selected devices in the context of the physical automated environments.
Low Power, High Data Rate Wireless Link for Sensor Networks (supervised by Deukhyou Heo)
A key component of practical automated environments is a fast, robust sensor network. There have been several recent demonstrations of telemetry systems incorporating a wireless link. To implement a wireless link in those telemetry systems, different methods have been applied. Investigators at the University of Michigan [32, 33] used a passive wireless link using inductive coupling between the primary and secondary coils for both power and data transfer. Their recent wireless link [34-36] is targeted to use a 4~20 MHz carrier frequency to obtain the baseband amplitude modulated carrier envelope and designed to extract a data stream up to 2 Mbit/sec from the frequency shift keying (FSK) modulated RF carrier.
system for humans that could be beneficial in smart environment settings. In this research project, the REU participants will aim to integrate a high-data-rate wireless link telemetry system on a single piece of silicon maintaining low power consumption. As an initial target system, we will use a multi-chip solution to implement the whole telemetry system. As a final goal, we will focus on the development of a single chip solution to achieve higher data rate, lower power, lighter weight, and smaller size. The resulting technology will be tested for integration and use with smart environment technologies.
Energy Efficiency (supervised by Anjan Bose)
End user energy consumption can be a complicated function of user needs. For example, devices such as smart blinds that open to reduce the need for artificial lighting may increase the need for air conditioning and inadvertently the need for overall energy consumption. Cooling and heating, in particular, can be difficult to manage due to long thermal time constants in a well-insulated building and the variation in desired temperature with the level of physical activity. Moreover, the time-of-day use of energy can be more important than the overall efficiency since energy (specifically, electric power) may be more readily available at far lower cost during off-peak hours. Historically, many attempts to directly control demand have failed because of the difficulty of reducing consumption without significant consumer inconvenience. Based upon our expertise in power systems and demand-side load response, REU students will build a model of energy consumption in a smart environment and learn automation policies that optimize energy usage while maintaining inhabitant comfort.
Smart Environments to Support Aging in Place (supervised by Diane Cook)
Perhaps the greatest impact of smart environment research will be the assistance that can be provided to individuals with disabilities and elderly living at home. Not only with the number of individuals aged 60 and over triple by 2050, but more of these elderly are living alone and want to continue living at home. We are investigating methods by which smart environment technologies can assist these individuals through assurance (making sure inhabitants are safe), support (helping individuals compensate for impairment), and assessment (determining physical or cognitive status) technologies. REU students will technologies to support this goal. REU students involved in this project will also volunteer at the assisted care facility in Pullman during as part of their summer program in order to become more aware of the issues and needs surrounding this problem.