Passive Infra-Red (PIR) sensors are an integral part of modern living. They have diverse applications ranging from automatic lighting and heating control in smart buildings, towel dispensers in washrooms, security alarms (for intrusion detection) to even human detection robots (for search and rescue). Unfortunately, PIR sensors are prone to failures during deployment due to reasons such as environmental damage, incorrect installation and component degradation among others that can lead to incorrect or faulty data. Currently, such failures are typically detected using either : (a) heavily engineered data-driven, statistical approaches that can have high false positive rates due to unseen data patterns or (b) expensive methods that use additional hardware such as video cameras or a golden reference sensor. The second approach inhibits scalability. In this work, we first create a taxonomy for the most common PIR sensor failures. We then present PIRMedic---a physics-driven approach, implemented at the edge, to detect the various classes of failures. We show that we can both detect and diagnose the failures in a PIR sensor using an intrinsic hardware signal viz., the analog output from the pyroelectric element in the sensor. Using this hardware signal in conjunction with frequency analysis and supervised machine learning methods, we obtain a high accuracy of 98 -- 99% in failure detection and diagnosis. We evaluate our methods using real-world deployments in three distinct locations, in different environment and usage conditions.
Smart water meters will soon provide real-time access to instantaneous water usage in many homes, and disaggregation is the problem of deciding how much of that usage is due to individual fixtures in the home. Existing disaggregation techniques require additional water sensing infrastructure and/or a manual characterization of each water fixture, which can be expensive and time consuming. In this paper, we describe a novel technique called WaterSense to perform fixture-level disaggregation using only a handful of inexpensive motion sensors. WaterSense automatically infers how many fixtures are in each room, and how much water each fixture uses. We evaluate the system using a 7-day in-situ evaluation in 2 diverse multi-resident homes with a total of 10 different water fixtures and 467 fixture events and observe that our approach achieves 86% classification accuracy in identifying individual water fixture events and 80--90% accuracy in determining the water consumption of individual water fixtures.
The commercial building sector is one of the largest energy consumers in the U.S., and lighting, heating, ventilating and air conditioning contribute to more than half of the energy consumption and carbon emissions in buildings. Controls are the most effective way of increasing energy efficiency in building systems; however, the interdependencies among building subsystems must be taken into account to achieve deep energy savings. A networked sensing and actuation infrastructure shared among building systems is the key to optimal integrated control of the interdependent building elements in low energy and zero net energy buildings. This paper presents a rapid-prototyping controls implementation platform based on the Building Controls Virtual Test Bed (BCVTB) framework that is capable of linking to building sensor and actuator networks for efficient controller design and testing. The platform creates a separation between the controls and the physical systems so that the controller can easily be implemented, tested and tuned with real performance feedback from a physical implementation. We realized an integrated lighting control algorithm using such a rapid-prototyping platform in a testing facility with networked sensors and actuators. This implementation has demonstrated an up to 57% savings in lighting electricity and 28% reduction in cooling demand.
We report on an experimental case study of personalized lighting controls built on top of an infrastructure designed to enable rapid development of applications in commercial buildings. Our personalized lighting controls (PLC) use an existing standard commercial building lighting automation system and require no new hardware to deploy. PLC presents occupants with a "shared virtual light switch" accessible online and easily viewable on smart phones by scanning a QR code. It embodies three important design principles: individual empowerment with localized human-centered resolution, token effort for energy consumption and return to a low-power state when inactive. After deploying our lighting controls on two new floors of a large research building on campus, we show a sustainable reduction in lighting energy of 50% to 70% on both floors over 12 weeks, continuing to this day. These savings are found to come from a combination of reducing brightness and keeping lights on less often, especially during evenings and weekends.
Existing electricity market segmentation analysis techniques only make use of limited consumption statistics (usually averages and variances). In this paper we use power demand distributions (PDDs) obtained from fine-grain smart meter data to perform market segmentation based on distributional clustering. We apply this approach to mining 8 months of readings from about 1000 US Google employees.
