1. School of Architecture
Occupancy Estimation in Smart Building using Hybrid CO2/Light Wireless Sensor Network
Chen Mao1, Qian Huang (Jenny)2
1Senior Student, Electrical and Computer Engineering
2Assistent Professor, School of Architecture
Southern Illinois University Carbondale
Presenter: Qian Huang

2. Outline Introduction Related Works Proposed System Prototype
Experimental Results
Conclusion

3. Smart Building
Intelligently delivers useful services to residents at lowest cost and maximum comfort
Through a variety of emerging technologies:
wireless sensor network
Internet of things (IoT)
Big data analytics
Smart Building applications
Intelligent parking, health monitoring, shopping assistance
Smart building, which delivers useful services to residents at lowest cost and maximum comfort, has gained increasing attention in recent years. A variety of emerging information technologies have been adopted in modern buildings, such as wireless sensor networks, internet of things, and big data analytics.
There are many interesting smart building applications. For example, in a smart shopping mall, internet of things technology connects every item for sale through wireless internet. Various retailers have implemented smartphone based indoor localization technology. The benefits of location based shopping assistance include knowing shopper’s location and trajectory, conducting shopping history data analysis, learning user’s shopping interest and preference, and then offering product recommendation and advertisement.
Source:

4. Building Energy Reduction
According to US Department of Energy
39% CO2 and 70% electricity from Building operation
Inefficient operation of HVAC (heating, ventilation, and air conditioning) equipment results in remarkable energy loss
It is common that HVAC systems keep running in active ON mode, while certain thermal zone is empty
Smart buildings should be energy efficient and low energy bill to building owners
Demand-driven HVAC control
Adaptive HVAC control based on room occupancy
According to U.S. Department of Energy (DOE), annual energy cost of buildings in United States reaches 200 billion. Much of building energy is consumed in heating, ventilation and air conditioning (HVAC) processes. It is common that HVAC systems keep running in active ON mode, while certain thermal zones or even the entire building is empty.
Most people agree that a smart building should be energy efficient, and consequently, much more affordable to building owners.
Occupancy information has a big impact on dynamic optimization of HVAC operation parameters and set points

5. Existing Occupancy Detection
Effectiveness of demand-driven HVAC control heavily depends on accurate occupancy detection
Passive Infrared (PIR) motion sensor
RFID sensor
Acoustic recognition/processing sensor
Video/image sensor
CO2 sensor
Acoustic
Video
RFID
CO2
PIR

6. Related Works PIR motion sensor (2001, 2009, 2010) RFID (2008, 2012)
Only detect if a person has moved in or out of an area, cannot detect actual occupancy
RFID (2008, 2012)
The location and trajectory of an occupant wearing a RFID sensor is easily observed and tracked – privacy and security concerns
Acoustic processing (2014)
Detection performance largely depends on environment where this technique is applied (quiet office vs. noisy supermarket)
Video/image sensor (2009, 2011, 2013)
Constraint of light of sight, high cost, privacy concern
CO2 sensor (2011, 2012, 2015)
CO2 level proportional to occupants, but varies case by case

7. Literature Review Summary
None of existing design meets the requirement of low-cost, high-accuracy, and better privacy

8. Challenges of Occupancy Detection
Low cost
High detection accuracy
Non-intrusive (due to increased concerns on personal privacy)
Source:
Source:

9. Our Contribution
We proposed a hybrid occupancy detection method using CO2 and light sensors.
Unlike video/image sensors, light sensors only report the illuminance level of light situations, hence privacy is protected.
With the assistance of light sensor, the hybrid detection method achieves better accuracy than using CO2 sensor alone.
We integrated this hybrid sensor with a wireless sensor node, and visualize the measurement data.

10. Wireless Sensor Network
Wireless sensor network is composed of numerous distributed autonomous sensor nodes
Each node senses ambient environment (Humidity, temperature or air quality)
Cost-effective, ease of use, flexibility, small size
Source:

11. Proposed System Architecture
Entire system consists of proposed hybrid sensors and a central control computer. The measurement results of CO2 and light levels are transmitted via wireless communication.

12. Prototype Implementation
Wireless sensor node from Texas Instruments (TI), miniature CO2 sensor from COZIR and light sensor from Adafruit are selected for the proposed system.

13. Experimental Results
Output voltage of CO2 sensor vs. Time when room ocupancy varies 3->5->4->3
The output voltage of this CO2 sensor precisely indicates room occupancy.

14. Experimental Results
A light sensor is taped on a door frame for experimental study
Once a person walks through the door and blocks lighting, the light sensor outputs a deep pulse response to this entrance or exit event

15. Experimental Results
Experimental setup to check response of light sensor under different illuminance conditions
The right figure shows how the output votlage of light sensor varies with illuminance. It looks like logarithmic relationship

16. Experimental Results
To verify the proposed hybrid detection method, we carried out experiments in an office building
Measured response of hybrid sensors when two occupants walk in and out of a room

17. Comparison References Mechanism Cost Intrusive
Occupancy Detection Performance
(Emmerich, 2001)
(Lam, 2009)
(Agarwal, 2010)
Passive infrared
High
Yes
Failure
(Lee, 2008)
(Li, 2012)
RFID
Low
Coarse-grained
(Uziel, 2013)
(Kelly, 2014)
(Huang, 2016)
Acoustic recognition No
Varying with environment
Failure when people keep silence
(Erickson, 2009)
(Benezeth, 2011)
(Ahmed, 2013)
Image camera
Failure when line of sight is not satisfied
(Sun, 2011)
(Nassif, 2012)
(Labeodan, 2015)
CO2 sensor
Accuracy depends on case by case, false detection may exist due to CO2 level fluctuation
This work
CO2 + Light
Improved accuracy with the assistance of light sensor

18. Conclusion
Smart building has great potential to increase quality of life, while significantly reducing energy usage and cost
Room occupancy is important information, which helps to realize energy-efficient demand-driven HVAC operation
Existing building occupancy detection or estimation methods can not meet all the requirements of low cost, high accuracy and privacy. A hybrid CO2/light sensor is proposed to meet design challenges.
The proposed system has been assembled and tested experimentally in an office building. The measurement results validate the functionality and benefits.

19. THANK YOU
&
QUESTIONS?