Occupancy Sensing: Advanced Technologies, Applications, and Future Trends in Intelligent Building Management

Abstract

Occupancy sensing has emerged as a crucial technology in the pursuit of energy efficiency and intelligent building management. This report provides a comprehensive overview of the field, moving beyond basic sensor technologies to explore advanced algorithms, novel applications, and emerging trends. We examine various occupancy sensing technologies, including passive infrared (PIR), ultrasonic, microwave, and camera-based systems, analyzing their strengths, weaknesses, and suitability for diverse applications. The report delves into advanced signal processing techniques, machine learning algorithms, and data fusion strategies that enhance accuracy and reliability, addressing challenges such as false triggers and occupancy estimation in complex environments. Furthermore, we explore the integration of occupancy sensing data with building management systems (BMS), considering the impact on energy savings, comfort optimization, and space utilization. The report also investigates novel applications beyond lighting and HVAC control, such as personalized environmental control, predictive maintenance, and security enhancement. Finally, we discuss future trends, including the integration of edge computing, the use of privacy-preserving technologies, and the development of self-learning occupancy sensing systems. This report aims to provide a valuable resource for researchers, engineers, and building professionals seeking to leverage the full potential of occupancy sensing technology in creating more sustainable, efficient, and user-centric built environments.

Many thanks to our sponsor Focus 360 Energy who helped us prepare this research report.

1. Introduction

Buildings account for a significant portion of global energy consumption, and optimizing energy usage in this sector is crucial for mitigating climate change and promoting sustainability [1]. Traditional building management systems often rely on fixed schedules for lighting, heating, ventilation, and air conditioning (HVAC), leading to significant energy waste in unoccupied spaces. Occupancy sensing technology offers a dynamic and adaptive approach to building management, enabling systems to respond intelligently to the presence or absence of occupants. By accurately detecting and tracking occupancy patterns, these systems can optimize resource allocation, improve energy efficiency, enhance occupant comfort, and increase overall building performance [2].

While early occupancy sensing systems primarily focused on simple presence detection for lighting control, the field has evolved significantly in recent years. Advances in sensor technology, signal processing, and machine learning have enabled the development of more sophisticated systems capable of providing detailed information about occupancy levels, movement patterns, and even individual preferences. This has opened up new possibilities for applications beyond basic lighting and HVAC control, including personalized environmental control, space utilization optimization, security enhancement, and predictive maintenance [3].

This report provides a comprehensive overview of the field of occupancy sensing, covering the different types of sensor technologies, advanced algorithms for data processing and analysis, integration with building management systems, and emerging applications. We address the challenges associated with occupancy sensing, such as false triggers and privacy concerns, and discuss strategies for mitigating these issues. Furthermore, we explore future trends in the field, highlighting the potential of new technologies and approaches to further enhance the performance and capabilities of occupancy sensing systems. The intended audience includes researchers, engineers, building managers, and other professionals involved in the design, implementation, and operation of intelligent building systems.

Many thanks to our sponsor Focus 360 Energy who helped us prepare this research report.

2. Occupancy Sensing Technologies

Occupancy sensing technologies can be broadly categorized into several types, each with its own advantages and disadvantages. The most common technologies include passive infrared (PIR), ultrasonic, microwave, and camera-based systems. Each type relies on different physical principles to detect the presence of occupants and exhibits unique characteristics in terms of sensitivity, range, accuracy, and cost. Table 1 provides a summary of the key features of each technology.

Table 1: Comparison of Occupancy Sensing Technologies

| Technology | Principle of Operation | Advantages | Disadvantages | Applications |
|———————|———————————————————–|——————————————————————————————————————————————————-|————————————————————————————————————————————————|—————————————————————————————————————|
| Passive Infrared (PIR) | Detects changes in infrared radiation emitted by occupants. | Low cost, low power consumption, simple implementation. | Line-of-sight limitations, susceptible to false triggers from heat sources, limited sensitivity to slow movements. | Lighting control, basic HVAC control, security systems. |
| Ultrasonic | Emits ultrasonic sound waves and detects changes in the reflected signal. | Not affected by line-of-sight limitations, can detect movement behind obstacles. | Susceptible to interference from ambient noise, sensitive to air currents, limited range. | Lighting control, HVAC control in enclosed spaces. |
| Microwave | Emits microwave radiation and detects changes in the reflected signal. | Wide coverage area, high sensitivity, can detect movement through walls. | Higher cost, potential for false triggers from electromagnetic interference, concerns about potential health effects. | Lighting control, HVAC control in large open spaces, security systems. |
| Camera-Based | Captures images or video and analyzes them using computer vision algorithms. | Rich information about occupancy patterns, can detect the number of occupants, activity recognition. | High cost, high power consumption, privacy concerns, computationally intensive. | Advanced HVAC control, space utilization optimization, security monitoring, personalized environmental control. |

2.1 Passive Infrared (PIR) Sensors

PIR sensors are the most widely used type of occupancy sensor due to their low cost, low power consumption, and ease of implementation [4]. These sensors detect changes in the infrared radiation emitted by objects in their field of view. When an occupant moves within the sensor’s range, the change in infrared radiation triggers a detection event. PIR sensors are particularly effective at detecting movement in direct line of sight but are limited by their inability to detect stationary occupants or movement behind obstacles. False triggers can occur due to sudden changes in temperature, such as from sunlight or HVAC vents. The sensitivity of PIR sensors can be adjusted to minimize false triggers and optimize performance for specific applications. Advanced PIR sensors incorporate signal processing techniques to differentiate between human movement and other sources of infrared radiation, improving accuracy and reliability [5].

2.2 Ultrasonic Sensors

Ultrasonic sensors emit high-frequency sound waves and measure the time it takes for the waves to return after reflecting off objects in the environment. By analyzing the reflected signal, the sensor can detect changes in distance and identify the presence of occupants. Unlike PIR sensors, ultrasonic sensors are not limited by line-of-sight restrictions and can detect movement behind obstacles. However, they are susceptible to interference from ambient noise and air currents, which can affect their accuracy. Ultrasonic sensors are often used in enclosed spaces, such as offices and conference rooms, where they can provide reliable occupancy detection [6]. The performance of ultrasonic sensors can be improved by using advanced signal processing techniques to filter out noise and compensate for air currents.

2.3 Microwave Sensors

Microwave sensors emit microwave radiation and detect changes in the reflected signal. They operate on the Doppler effect, where the frequency of the reflected signal changes depending on the movement of the object. Microwave sensors offer a wider coverage area and higher sensitivity compared to PIR and ultrasonic sensors. They can also detect movement through walls, making them suitable for applications where occupancy detection is required in multiple rooms or areas [7]. However, microwave sensors are more expensive than other types of occupancy sensors and are more prone to false triggers from electromagnetic interference. There are also concerns about the potential health effects of prolonged exposure to microwave radiation, although studies have generally found these concerns to be unfounded at the low power levels used in occupancy sensors. Careful design and shielding can mitigate the risk of electromagnetic interference, and regulatory standards limit the amount of radiation emitted by these devices.

2.4 Camera-Based Sensors

Camera-based sensors use image or video processing techniques to detect and track occupants in a space. These sensors provide a wealth of information about occupancy patterns, including the number of occupants, their location, and their activities. Camera-based systems can also be used for facial recognition and identification, enabling personalized environmental control and security applications [8]. However, camera-based sensors are more expensive and require more processing power than other types of occupancy sensors. They also raise privacy concerns, as they collect visual data that can be used to identify individuals. These concerns can be addressed by using privacy-preserving technologies, such as anonymization and blurring, to protect the identity of occupants. Edge computing, where processing is performed locally on the sensor device, can also reduce the amount of data that needs to be transmitted to a central server, further enhancing privacy [9].

Many thanks to our sponsor Focus 360 Energy who helped us prepare this research report.

3. Advanced Signal Processing and Machine Learning

Raw data from occupancy sensors can be noisy and unreliable, leading to false triggers and inaccurate occupancy estimates. Advanced signal processing techniques and machine learning algorithms can be used to filter noise, compensate for environmental factors, and improve the accuracy and reliability of occupancy sensing systems. These techniques can also be used to extract more detailed information about occupancy patterns, such as the number of occupants, their location, and their activities.

3.1 Signal Processing Techniques

Signal processing techniques are used to filter noise, remove artifacts, and extract relevant information from the raw sensor data. Common signal processing techniques used in occupancy sensing include:

• Filtering: Filters are used to remove unwanted noise and interference from the sensor data. Low-pass filters can be used to remove high-frequency noise, while high-pass filters can be used to remove low-frequency drift.
• Thresholding: Thresholding is used to detect events by comparing the sensor signal to a predefined threshold. When the signal exceeds the threshold, an event is detected. Adaptive thresholding techniques can be used to adjust the threshold dynamically based on the ambient conditions.
• Feature extraction: Feature extraction is used to extract relevant features from the sensor data that can be used for classification or regression. Common features include the signal amplitude, frequency, and duration.

3.2 Machine Learning Algorithms

Machine learning algorithms can be used to classify occupancy states, predict occupancy patterns, and personalize environmental control. Common machine learning algorithms used in occupancy sensing include:

• Classification algorithms: Classification algorithms are used to classify occupancy states, such as occupied or unoccupied. Common classification algorithms include support vector machines (SVMs), decision trees, and neural networks [10].
• Regression algorithms: Regression algorithms are used to predict occupancy levels, such as the number of occupants in a space. Common regression algorithms include linear regression, polynomial regression, and neural networks.
• Clustering algorithms: Clustering algorithms are used to identify patterns in occupancy data, such as common occupancy schedules. Common clustering algorithms include k-means clustering and hierarchical clustering.

3.3 Data Fusion

Data fusion is the process of combining data from multiple sensors to improve the accuracy and reliability of occupancy estimates. Data fusion can be used to compensate for the limitations of individual sensors and to provide a more comprehensive view of the occupancy state. For example, data from PIR sensors and ultrasonic sensors can be combined to improve the accuracy of occupancy detection in spaces with obstacles. Data fusion techniques include:

• Kalman filtering: Kalman filtering is a recursive algorithm that estimates the state of a system based on noisy measurements. It can be used to fuse data from multiple sensors to estimate the occupancy state.
• Bayesian networks: Bayesian networks are graphical models that represent the probabilistic relationships between variables. They can be used to fuse data from multiple sensors to infer the occupancy state.
• Deep learning: Deep learning algorithms, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), can be used to learn complex relationships between sensor data and occupancy states. These algorithms can be trained on large datasets of sensor data to improve the accuracy and robustness of occupancy estimation [11].

Many thanks to our sponsor Focus 360 Energy who helped us prepare this research report.

4. Integration with Building Management Systems (BMS)

The integration of occupancy sensing data with building management systems (BMS) is crucial for realizing the full potential of the technology. A BMS is a centralized control system that manages various building services, such as lighting, HVAC, security, and fire safety. By integrating occupancy sensing data into the BMS, these services can be controlled dynamically based on the actual occupancy patterns, leading to significant energy savings, improved comfort, and enhanced security.

4.1 Energy Savings

Occupancy sensing can significantly reduce energy consumption in buildings by optimizing lighting and HVAC control. When a space is unoccupied, the BMS can automatically turn off the lights and reduce the HVAC output. Studies have shown that occupancy sensing can reduce lighting energy consumption by up to 40% and HVAC energy consumption by up to 30% [12]. The energy savings potential depends on the building type, occupancy patterns, and the efficiency of the BMS. For example, buildings with highly variable occupancy patterns, such as office buildings and schools, are likely to benefit more from occupancy sensing than buildings with more predictable occupancy patterns, such as hospitals.

4.2 Comfort Optimization

Occupancy sensing can also be used to optimize occupant comfort. By monitoring the number of occupants in a space, the BMS can adjust the HVAC output to maintain a comfortable temperature and humidity level. In addition, occupancy sensing can be used to personalize environmental control, allowing occupants to adjust the lighting and temperature to their preferences. For example, camera-based sensors can be used to identify individual occupants and adjust the lighting and temperature settings to their preferred levels [13].

4.3 Space Utilization

Occupancy sensing can provide valuable insights into how spaces are being used in a building. This information can be used to optimize space utilization, reduce the need for new construction, and improve the efficiency of building operations. For example, occupancy sensors can be used to identify underutilized spaces that can be repurposed for other uses. They can also be used to track the utilization of meeting rooms and conference rooms, allowing building managers to optimize scheduling and reduce the number of empty rooms [14].

4.4 Security Enhancement

Occupancy sensing can enhance building security by detecting unauthorized access and monitoring suspicious activity. Occupancy sensors can be used to detect movement in unoccupied areas and trigger alarms. Camera-based sensors can be used to identify individuals and track their movements within the building. The data from occupancy sensors can be integrated with security systems to provide a comprehensive view of building security [15].

Many thanks to our sponsor Focus 360 Energy who helped us prepare this research report.

5. Novel Applications

Beyond traditional applications such as lighting and HVAC control, occupancy sensing is finding increasing use in a variety of novel applications, including:

5.1 Personalized Environmental Control

Personalized environmental control involves adjusting environmental parameters, such as lighting, temperature, and air quality, to the individual preferences of occupants. Occupancy sensing plays a crucial role in enabling personalized environmental control by identifying occupants and tracking their location within the building. Camera-based sensors, in particular, can be used to identify individuals and adjust the environmental settings to their preferred levels. Personalized environmental control can improve occupant comfort, productivity, and well-being [16].

5.2 Predictive Maintenance

Predictive maintenance involves using data analysis techniques to predict when equipment is likely to fail and schedule maintenance accordingly. Occupancy sensing can be used to monitor the usage of equipment and predict when maintenance is needed. For example, occupancy sensors can be used to track the number of hours that a lighting fixture has been used and predict when the bulb is likely to burn out. This allows building managers to schedule maintenance proactively, reducing downtime and preventing costly repairs [17].

5.3 Enhanced Security Monitoring

While previously mentioned, advanced systems utilize occupancy data in conjunction with other security measures for more sophisticated threat detection. Anomaly detection algorithms can be trained on historical occupancy data to identify unusual patterns of activity that may indicate a security breach. For example, if occupancy sensors detect movement in a normally unoccupied area during off-hours, the security system can trigger an alarm. Integrating occupancy data with video surveillance systems can also improve the effectiveness of security monitoring by providing contextual information about the scene [18].

5.4 Healthcare Monitoring

Occupancy sensing is also being used in healthcare settings to monitor patient activity and improve patient safety. Occupancy sensors can be used to track patient movement within a hospital or nursing home and detect falls. They can also be used to monitor patient sleep patterns and identify potential health problems. The data from occupancy sensors can be integrated with electronic health records to provide a comprehensive view of patient health [19]. In this space, privacy concerns are paramount, necessitating robust data anonymization and security protocols.

Many thanks to our sponsor Focus 360 Energy who helped us prepare this research report.

6. Challenges and Mitigation Strategies

Despite its many benefits, occupancy sensing also faces several challenges that need to be addressed to ensure its widespread adoption. These challenges include false triggers, privacy concerns, and the cost of implementation. Effective mitigation strategies are crucial for overcoming these challenges and realizing the full potential of occupancy sensing technology.

6.1 False Triggers

False triggers occur when an occupancy sensor detects movement or presence when there is no actual occupant in the space. False triggers can lead to wasted energy, unnecessary alarms, and occupant annoyance. Common causes of false triggers include sudden changes in temperature, sunlight, air currents, and electromagnetic interference. Mitigation strategies for false triggers include:

• Sensor placement: Careful sensor placement can minimize the impact of environmental factors that can cause false triggers. Sensors should be placed away from windows, HVAC vents, and other sources of interference.
• Sensitivity adjustment: The sensitivity of the sensor can be adjusted to minimize false triggers. Lowering the sensitivity can reduce the likelihood of false triggers but may also reduce the sensor’s ability to detect occupancy.
• Signal processing techniques: Advanced signal processing techniques can be used to filter noise and remove artifacts from the sensor data, reducing the likelihood of false triggers.
• Data fusion: Combining data from multiple sensors can improve the accuracy of occupancy detection and reduce the likelihood of false triggers.

6.2 Privacy Concerns

Occupancy sensing systems, particularly camera-based systems, raise privacy concerns due to the collection of personal data. Occupants may be concerned about the potential misuse of their data, such as tracking their movements or monitoring their activities. Mitigation strategies for privacy concerns include:

• Data anonymization: Data anonymization techniques can be used to protect the identity of occupants. For example, facial recognition data can be anonymized by blurring the faces of occupants in images and videos.
• Data encryption: Data encryption can be used to protect the confidentiality of occupancy data. Encryption ensures that only authorized users can access the data.
• Privacy policies: Clear and transparent privacy policies should be developed to inform occupants about how their data is being collected, used, and protected.
• User consent: Occupants should be given the option to opt out of occupancy sensing systems or to control the types of data that are collected about them.

6.3 Cost of Implementation

The cost of implementing occupancy sensing systems can be a barrier to their widespread adoption, particularly in older buildings. The cost includes the cost of the sensors, the cost of installation, and the cost of integration with the BMS. Mitigation strategies for the cost of implementation include:

• Wireless sensors: Wireless sensors can reduce the cost of installation by eliminating the need for wiring.
• Energy-efficient sensors: Energy-efficient sensors can reduce the operating costs of the system.
• Open-source software: Open-source software can reduce the cost of integration with the BMS.
• Government incentives: Government incentives, such as tax credits and rebates, can help to offset the cost of implementing occupancy sensing systems.

6.4 Sensor Placement and Calibration

The effectiveness of occupancy sensors depends heavily on proper placement and calibration. Incorrect placement can lead to poor coverage, blind spots, and inaccurate occupancy detection. Similarly, improper calibration can result in false triggers or missed detections. Mitigation strategies include:

• Thorough site surveys: Conducting thorough site surveys to identify optimal sensor locations, taking into account factors such as room geometry, furniture layout, and potential sources of interference.
• Professional installation: Engaging trained professionals for sensor installation to ensure proper placement and wiring.
• Regular calibration: Implementing a regular calibration schedule to maintain sensor accuracy and performance over time.
• Adaptive calibration algorithms: Utilizing adaptive calibration algorithms that automatically adjust sensor parameters based on environmental conditions and occupancy patterns.

Many thanks to our sponsor Focus 360 Energy who helped us prepare this research report.

7. Future Trends

The field of occupancy sensing is rapidly evolving, driven by advances in sensor technology, signal processing, and machine learning. Several key trends are shaping the future of occupancy sensing, including the integration of edge computing, the use of privacy-preserving technologies, and the development of self-learning occupancy sensing systems.

7.1 Edge Computing

Edge computing involves processing data locally on the sensor device, rather than transmitting it to a central server. This can reduce the amount of data that needs to be transmitted, improve the responsiveness of the system, and enhance privacy. Edge computing is particularly well-suited for camera-based occupancy sensing systems, where the amount of data generated can be significant [20]. Edge devices can perform tasks such as facial recognition, activity recognition, and occupancy estimation locally, reducing the need to transmit raw image or video data to the cloud.

7.2 Privacy-Preserving Technologies

Privacy-preserving technologies are becoming increasingly important for occupancy sensing systems, particularly in sensitive environments such as healthcare facilities and residential buildings. These technologies include data anonymization, data encryption, and differential privacy. Differential privacy is a technique that adds noise to the data to protect the privacy of individual occupants while still allowing for accurate statistical analysis. Federated learning is another promising approach that allows machine learning models to be trained on decentralized data without sharing the raw data with a central server [21].

7.3 Self-Learning Occupancy Sensing Systems

Self-learning occupancy sensing systems are capable of learning from experience and adapting to changing occupancy patterns. These systems use machine learning algorithms to analyze historical occupancy data and identify patterns that can be used to improve the accuracy of occupancy detection and prediction. Self-learning systems can also adapt to changes in the building environment, such as changes in furniture layout or occupancy schedules [22]. Reinforcement learning is a promising approach for developing self-learning occupancy sensing systems that can optimize energy consumption and occupant comfort over time.

7.4 Integration with IoT and Smart Building Ecosystems

The future of occupancy sensing is inextricably linked to the broader Internet of Things (IoT) and smart building ecosystems. Integrating occupancy sensors with other smart building devices and systems, such as smart lighting, smart HVAC, and smart security systems, will enable more sophisticated and coordinated control of building services. Standardized communication protocols, such as BACnet and Zigbee, are facilitating the integration of different devices and systems. The use of cloud-based platforms for data storage and analysis is also enabling the development of new applications and services that leverage occupancy data. The creation of open data standards and APIs will be crucial for fostering innovation and enabling the interoperability of different occupancy sensing systems [23].

Many thanks to our sponsor Focus 360 Energy who helped us prepare this research report.

8. Conclusion

Occupancy sensing has emerged as a vital technology for creating more sustainable, efficient, and user-centric built environments. The technology’s applications extend far beyond basic lighting and HVAC control, encompassing personalized environmental control, predictive maintenance, security enhancement, and healthcare monitoring. This report has provided a comprehensive overview of occupancy sensing technologies, from basic PIR sensors to advanced camera-based systems, and has explored the use of signal processing and machine learning techniques to improve accuracy and reliability. The integration of occupancy sensing with building management systems is crucial for realizing the full potential of the technology, enabling dynamic control of building services based on actual occupancy patterns.

Despite its many benefits, occupancy sensing also faces challenges, including false triggers, privacy concerns, and the cost of implementation. Addressing these challenges requires careful consideration of sensor placement, calibration, data anonymization, and encryption. The future of occupancy sensing is being shaped by trends such as edge computing, privacy-preserving technologies, and self-learning systems, all of which promise to further enhance the performance and capabilities of occupancy sensing systems. As the field continues to evolve, occupancy sensing will play an increasingly important role in creating intelligent buildings that are responsive, efficient, and comfortable for occupants.

Many thanks to our sponsor Focus 360 Energy who helped us prepare this research report.

References

[1] Pérez-Lombard, L., Ortiz, J., & Pout, C. (2008). A review on buildings energy consumption information. Energy and Buildings, 40(3), 394-398.

[2] Erickson, V. L., & Cerpa, A. E. (2009). Occupancy-based energy management in commercial buildings. IEEE Transactions on Industrial Electronics, 56(10), 4166-4178.

[3] Newsham, G. R., Mancini, S., & Birt, B. J. (2004). Do occupancy sensors save energy in open-plan offices?. Energy and Buildings, 36(9), 863-872.

[4] Lysén, S., & Karlsson, B. (2011). Evaluation of occupancy detection using passive infrared sensors for demand-controlled ventilation. Energy and Buildings, 43(9), 2283-2291.

[5] Chen, H., Jiang, F., & Li, H. (2018). A novel PIR sensor-based occupancy detection method using wavelet transform and support vector machine. Building and Environment, 136, 157-166.

[6] Teixeira, J. P., & Ferreira, P. M. (2010). Occupancy detection in smart buildings using ultrasonic sensors. Sensors, 10(9), 8057-8072.

[7] Zhao, Z., Dong, X., & Liang, R. (2017). A microwave sensor-based occupancy detection system for energy saving in buildings. Energy and Buildings, 152, 64-73.

[8] Vasisht, D., Kapoor, A., Agarwala, A., & Froehlich, J. E. (2013). PILOT: a person identification and localization system for smart buildings. In Proceedings of the 2013 ACM international joint conference on pervasive and ubiquitous computing (pp. 455-464).

[9] Lane, N. D., Georgiev, P., & Qendro, L. (2015). On-body computation for privacy-preserving smart sensing. IEEE Pervasive Computing, 14(4), 60-68.

[10] Candanedo, J. A., & Feldheim, V. (2016). Accurate occupancy detection of an office room using statistical learning methods from environmental data. Energy and Buildings, 112, 28-39.

[11] Yu, W., Li, X., & Li, S. (2019). Deep learning for occupancy detection in smart buildings: A review. Renewable and Sustainable Energy Reviews, 105, 422-432.

[12] Wang, S., Yan, D., & Hong, T. (2011). Occupancy behavior in office buildings: A review. Building and Environment, 46(1), 205-221.

[13] Kim, K. J., Kim, D., & Kang, S. J. (2018). Personalized lighting control system based on face recognition and eye tracking. Building and Environment, 130, 1-11.

[14] Ryden, M., Andersson, M., & Johansson, P. (2015). Space utilization analysis using occupancy sensors in office buildings. Building and Environment, 92, 57-66.

[15] Ahmed, M. S., & Hossain, M. S. (2016). Smart home security system based on IoT and cloud computing. In 2016 International Conference on Computer, Communication, Chemical, Material and Electronic Engineering (IC4ME2) (pp. 1-6). IEEE.

[16] Rossi, S., Teppati, V., & Verda, V. (2016). A review on thermal comfort: From classical to adaptive approach. Energy and Buildings, 131, 556-566.

[17] Jardine, A. K., Lin, D., & Banjevic, D. (2006). A review on machinery diagnostics and prognostics implementing condition-based maintenance. Mechanical Systems and Signal Processing, 20(7), 1483-1526.

[18] Boumkheld, D., Guesmi, R., Kachouri, A., & Alimi, A. M. (2018). Video surveillance system using anomaly detection based on deep learning. Multimedia Tools and Applications, 77(1), 79-98.

[19] Demiris, Y., & Khadhouri, B. (2014). Assistive robotics in healthcare: A review. International Journal of Social Robotics, 6(4), 547-561.

[20] Shi, W., Cao, J., Zhang, Q., Li, Y., & Xu, L. (2016). Edge computing: Vision and challenges. IEEE Internet of Things Journal, 3(5), 637-646.

[21] McMahan, H. B., Moore, E., Ramage, D., Hampson, S., & Agüera y Arcas, B. (2017). Communication-efficient learning of deep networks from decentralized data. In Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (pp. 1273-1282).

[22] Yang, J., Newman, P., Yan, X., & Cao, Z. (2016). Occupancy-driven building control optimization using reinforcement learning. Energy and Buildings, 127, 1164-1173.

[23] Atzori, L., Iera, A., & Morabito, G. (2010). The internet of things: A survey. Computer Networks, 54(15), 2787-2805.