Search

People-counting data can be used in building-control systems to improve comfort and in space management applications to optimize building space. In this work, we consider a combi-sensor with a single-pixel thermopile and passive infrared sensor for people counting. We first develop a thermopile signal model for object temperature measurements under multiple people occupancy. We then propose a people counting method based on: cumulative sum (CUSUM) change detection in the object temperature signal, forming a people count estimate using likelihoods of differential mean temperature in detected changes, and decision fusion with an infrared vacancy sensor. The proposed method is evaluated with data generated using the developed signal model as well as experimental data from a cell office/meeting room environment. We obtain an average counting error of 0.11 and 0.19 for 90% of the instants respectively when considering 15 minute windows for simulated and experimental datasets.

To alleviate the energy constraint and coverage limitations in Internet of Things applications, in this paper we investigate a dual-hop hybrid visible light communication (VLC)/radio frequency (RF) cooperative communication system based on simultaneous lightwave information and power transfer. The proposed system consists of a light-emitting diode source (S), an off-the-grid relay (R) that moves randomly within the coverage of S, and a destination (D) that is located outside the coverage of S. To facilitate communications between S and D, the received optical signal at R in the first phase can be separated into alternating current and direct current components for information decoding and energy harvesting, respectively. Then the energy harvested by R is used to forward the decoded information to D using RF in the second phase. In view of the distinct channel conditions across the two hops, the decoding status and the energy available at R, the end-to-end system outage probability is analytically derived. Simulation results are used to evaluate the impact of different system parameters and demonstrate the efficiency of the proposed system.

People counting and occupancy sensor systems are used in diverse smart building applications like lighting controls, HVAC controls and workspace management. Anomalies in people counting and occupancy sensor data can result in undesirable application behavior. We address the problem of anomaly classification of sensor data using 3-class random forest classifiers in this work. We specifically consider sensor deployment scenarios where sensor fields-of-view may overlap. Depending on the sensor type, we devise signal features in time, frequency and spatio-temporal domains. The proposed random forest classifiers are evaluated using people counting and binary occupancy sensor data in an office environment, and are shown to have higher true positive rate and a lower false positive rate in comparison to an unsupervised {k} -means method and a random forest classifier with a single signal energy feature.

Business models around energy performance of lighting systems and managed services make increasing use of the energy metering sensor data. However, accurate submetering solutions remain expensive and complex to commission. With the goal of realizing a low-cost metering sensor solution that provides reasonable measurement accuracy, we consider nonintrusive load monitoring for lighting systems using virtual energy submetering. We propose analytics methods using metered lighting energy sensor data to address three problems: 1) provide initial validation of assets of a lighting control system; 2) disaggregate energy consumption across distinct control groups of a validated system enabling usage and performance monitoring; and 3) monitor and manage lighting system asset over lifecycle. The proposed methods are evaluated using an experimental testbed in a commercial building.

We consider the problem of detecting sensor commissioning in the form of determining the sensor layout. We address this problem for single-pixel thermopile sensors, located at the ceiling, that provide remote temperature measurements for people counting applications and HVAC controls. We employ a random forest classifier to determine the deployed layout in an area. For this classifier, we propose spatio-temporal distance features using two-sided cumulative sum recursive least squares (CUSUM RLS) filtering of the thermopile temperature sensor signals. Using sensor data generated with simulated occupancy patterns and a thermopile signal model, we show that the proposed method achieves a true positive rate (determining the correct layout) of 90.2% and false positive rate of 1.3%.

We consider the problem of classifying anomalous occupancy sensor behavior in connected indoor lighting systems. Anomalous occupancy sensor behavior may occur in the form of either a high number of false alarms (type-1 anomalies) or missed detection (type-2 anomalies). We consider a supervised machine learning approach to determine whether the detection signal of an occupancy sensor is normal, or exhibits type-1 or type-2 anomalies. We devise occupancy signal features in the time and frequency domains and employ a random forest classifier to perform 3-class classification. The proposed method is evaluated using motion sensor data from an office building, and is shown to have higher true positive rate and a lower false positive rate in comparison to an unsupervised k-means method and a random forest classifier with a single signal energy feature.

We consider the problem of controlling a smart lighting system of multiple luminaires with collocated occupancy and light sensors. The objective is to attain illumination levels higher than specified values (possibly changing over time) at the workplace by adapting dimming levels using sensor information, while minimizing energy consumption. We propose to estimate the daylight illuminance levels at the workplace based on the daylight illuminance measurements at the ceiling. More specifically, this daylight estimator is based on a model built from data collected by light sensors placed at workplace reference points and at the luminaires in a training phase. Three estimation methods are considered: regularized least squares, locally weighted regularized least squares, and cluster-based regularized least squares. This model is then used in the operational phase by the lighting controller to compute dimming levels by solving a linear programming problem, in which power consumption is minimized under the constraint that the estimated illuminance is higher than a specified target value. The performance of the proposed approach with the three estimation methods is evaluated using an open-office lighting model with different daylight conditions. We show that the proposed approach offers reduced under-illumination and energy consumption in comparison to existing alternative approaches.

Indoor lighting systems need to be designed to balance energy consumption and the visual comfort of occupants. Achieving this goal with multiple luminaires and sensors is, however, not a simple problem. Lighting control systems need to adjust the dimming levels of luminaires in real time based on occupancy conditions and external daylight changes. We propose a system based on visible light communication to monitor and control artificial lighting in a robust manner. In our system, luminaires modulate the emitted light to broadcast basic control information while fulfilling their main purpose of illumination. Based on the broadcasted information, luminaires use collocated light sensors to estimate the optical channel gains and daylight contributions. This information is then used in a control algorithm to determine the dimming levels of luminaires under specified illumination constraints. We provide an analytical framework, simulations, and an empirical evaluation of our approach in an office space. We compare our method with a state-of-art lighting control system that uses radio transceivers to communicate information, and thus, cannot monitor optical channels in real time. Our results show that while the radio-based system may underilluminate and even oscillate around the desired illumination due to large reflectance changes in the environment, our method provides stable illumination close to desired levels.

We consider a smart lighting control system for daylight and occupancy adaptation with illumination and dimming constraints. The lighting system has multiple luminaires, with co-located light and occupancy sensors, and a central controller. The sensors communicate local occupancy state and illuminance measurements to the central controller where dimming levels for the luminaires are optimized. We consider an optimization framework wherein the objective is to minimize the power consumption subject to illumination and dimming constraints. The illumination constraint is to achieve an illumination value at the light sensor that is above a specified set-point, depending on the occupancy state. The dimming constraints are to achieve a level of spatial uniformity in dimming levels of the luminaires within a neighbourhood, with the dimming levels being within physical limits. We propose an iterative method for optimization to determine the dimming levels that (i) does not require explicit daylight knowledge for every iteration, and (ii) induces smoother dimming level changes in luminaires. The proposed method and optimization framework is compared with existing approaches, in terms of achieved illumination and dimming values, in an open-plan office lighting model.

We consider the problem of quantifying energy savings when a lighting control system is upgraded. A conventional approach to determine energy savings resulting from the control upgrade is using the energy consumption values over the pre-upgrade and post-upgrade time periods. However, energy consumption depends on operational factors, such as occupancy and daylight conditions, which may be different over the two periods. This may result in incorrect estimation of the energy savings. To address this problem, we consider energy performance prediction using sensor data from the lighting system. Specifically, occupancy and light sensor data in the lighting control system over the pre-upgrade period are used to construct a prediction model of the energy consumption using support vector regression. Sensor data from the post-upgrade period is then used in this model to estimate the energy consumption of the pre-upgrade control system under operating conditions in the post-upgrade period. Using data from an indoor office lighting model, we show that the proposed method provides better estimates of energy savings than the conventional approach.

Adaptation of artificial light based on the available amount of daylight is known to be effective for energy savings. To achieve such daylight control, the state-of-the-art lighting systems use external photodetectors. The photodetector measures the combined contribution of artificial light and daylight, and closed-loop control schemes are used to determine the dimming levels of luminaires to produce the right amount of artificial light. In this paper, we propose LED luminaires that can perform the dual function of illumination and daylight sensing, obviating the need of additional photodetectors. We present a daylight sensing luminaire prototype and consider two driving protocols for sensing and illumination. An open-loop control scheme is then considered for daylight control. The proposed system is shown to be more robust to reflectance changes in comparison with a photodetector-based closed-loop lighting control system.

Location data analytics refers to the statistical analysis of data pertaining to user location. Such analytics offers useful insights into the interactions among occupants and with their environment. Based on the insights, building facility management may be improved. We consider location data analytics using wireless sensors, which are part of a wireless lighting system, to collect radio received signal strength indicator data resulting from device transmissions from user mobile devices. Zone positioning of users is first performed. Zone position data are then used for analytics to obtain: 1) occupancy density; 2) environment characterization; 3) crowd detection; and 4) occupant connectivity. The approach is evaluated in a simulated open-plan office lighting environment considering simple occupant motion profiles and wireless device transmission models.

We consider the problem of classifying anomalous occupancy sensor behavior in connected indoor lighting systems. Anomalous occupancy sensor behavior may occur in the form of either a high number of false alarms (type-1 anomalies) or missed detections (type-2 anomalies). Two anomaly discovery scenarios are considered: one, in which no anomalies exist post-deployment, and two, in which both anomaly types are found together with normally functional sensors. We address the problem of classifying anomalies that may occur subsequently using a machine learning approach. Under scenario 1, we consider a one class random forest classifier to determine whether an occupancy signal is normal or not. In scenario 2, we consider a supervised random forest classifier to determine whether the detection signal of an occupancy sensor is normal, or exhibits type-1 or type-2 anomalies. We devise occupancy signal features in time and frequency domains to perform 2-class classification in scenario 1, and 3-class classification in scenario 2. The proposed method is evaluated using motion sensor data from an office building, and is shown to have higher true positive rate and a lower false positive rate in comparison to an unsupervised $k$ -means method and a random forest classifier with a single signal energy feature.