Design and Implementation of a Leader-Follower Smart Office Lighting Control System Based on IoT Technology

Most installed lighting systems are outdated and have poor energy efficiency. Therefore, there is an urgent need for smart LED lighting systems that are energy efficient, easy to install, and inexpensive. In this article, a leader-follower smart office lighting control system based on Internet of Things (IoT) technology is proposed to satisfy the goal of saving energy through the coordinated operation of a system counter and an infrared (IR) human movement sensor (passive infrared (PIR) sensor). When no one is present in the space, all indoor LED lights are turned to low-light mode. Otherwise, all indoor LED lights are turned to high-light mode together. In addition to change the brightness of LED lights in the same area at the same time to save energy, the parameters of the LED lights can be set directly through microcontrollers via the IoT and the internet. If a general 15 W T8 LED tube (noninduction light) is replaced with the proposed leader-follower office lighting system (assuming that the office is occupied for 10 hours a day and that the hourly low-light mode is 20 minutes), then the power-saving rate is as high as 28.13%.


I. INTRODUCTION
The rapid industrial and economic development of various countries has irreparably damaged the Earth's environment. As the Earth becomes unable to bear this load, people are becoming aware of the serious adverse effects on the climate and the environment. In turn, the importance of global warming and environmental protection issues such as energy conservation and the sustainable development of green energy are gradually being recognized by countries worldwide, and relevant protective measures have been put forward to maintain resources.
Residential electricity consumption is increasing rapidly worldwide, and energy demands are increasing daily. There is increasing interest in high-efficiency devices, particularly after the doubling of oil prices has led to rising electricity rates [1]. Population growth has resulted in a lack of natural resources. For example, manufactured resources such as electricity can no longer meet the needs of rapid population growth. In the past few decades, various measures have been taken to improve equipment and system design to improve production and transmission efficiency and to reduce power consumption [2].
In the United States, the energy consumed by buildings accounts for approximately 40% of the energy consumption, of which lighting for residential buildings accounts for approximately 10% and that for commercial buildings accounts for 20% [3]. Therefore, lighting is the most energyconsuming aspect of buildings worldwide. Reducing the energy consumption of lighting and the development of LEDs have had a significant impact on the lighting industry [4]- [5]. The use of LED technology in lighting systems can save energy [6], and LED technology development and applications continue to evolve. LED technology can be used to reduce the power required for lighting and to enhance its durability and environmental protection [6], [7]- [10].
The governments of various countries are actively promoting energy efficiency campaigns. Research has found that simple lighting control using motion sensors can effectively reduce energy consumption in buildings. However, smart lighting control strategies can achieve greater energy savings and better services and provide many advantages over simple on/off control [11].
According to TrendForce's latest report, the outlook for the global LED lighting market is more optimistic than previously estimated. The global LED lighting market is estimated to grow by 5.1% in 2021. COVID-19 is spreading rapidly around the world. LED lighting products are necessary for human production and life. People stay at home, leading to higher household electricity consumption. Therefore, rigid market demand still exists. During the COVID-19 pandemic, the demand for home lighting has not decreased but rather increased, highlighting the essential nature of lighting. In addition, lighting products will further develop toward digital intelligent dimming and control lights. The lighting industry will also likely pay more attention to the intelligent systemization of products, the relationship between human health and lighting, and market demand for segmented applications in the future. TrendForce estimates that the lighting industry will reach 44.3 billion US dollars in 2025 [12].
Until now, due to the high cost of installation and maintenance and the difficulty of retrofitting, smart lighting control systems have not been very successful [13], and there is still room for improvement. Moreover, the previously used systems were expensive and required the setting up of servers, rewiring, and extensive renovation. This article proposes a low-cost, easy-to-renovate intelligent lighting control system suitable for office environments that can retain the original lamp holders without the need for wiring, server setup, and leading to energy savings. Immediately after the office space is unoccupied, it is changed to run in the low-light mode, ensuring that the office space will not turn into a dim environment during working hours. This paper is an extended application work of [50]. Although [50] has an excellent energy-saving effect, it is only suitable for installation in public spaces such as parking lots and corridors and is not ideal for office space. The system proposed in this paper is suitable for office space through the coordinated operation of the PIR Sensor on the lamp and the system counter above the office door. When it is confirmed that no one exists in the office, all the LED lights in the room will be turned into low-light mode together, so there will be no misjudgment situation.
The rest of this article is organized as follows. Section II discusses related work. Section III introduces the proposed system. Section IV conducts and discusses energy-savings calculations related to the proposed system. Finally, we conclude the article in Section V.

II. RELATED WORK
LED lighting is environmentally friendly but is still not popular, mainly because its installation cost remains higher than that of existing lighting devices. With advances in LED manufacturing technology, this price is falling, but the overall lighting cost is still prohibitive.
Sensors and network technology are being explored to improve the economic benefits of LED lighting, and such LED lighting systems have been commercialized. When the user is not in the room or the room is too bright, these systems reduce energy consumption by reducing the output power of the lighting instead of maintaining the same brightness [14].
Byun and Shin [14] proposed an energy-efficient lighting control system, namely ESLiCoS, which considered user satisfaction. The ESLiCoS improves energy efficiency and occupant satisfaction by controlling lighting control parameters considering spatial characteristics and occupant behavior patterns.
In recent years, the lighting field's energy efficiency and control research results have been fruitful, and the various technologies and solutions that have been proposed [14]- [29] can mainly be divided into two methods: wired and wireless systems. However, the maintenance and installation costs of wired systems are relatively high, and thus, these systems are slowly being replaced by wireless technology.
Wireless sensor networks (WSNs) are an effective technology for building energy control because they are easier to implement than wired networks. Combining advanced WSN control with DC-powered LED lighting systems should achieve greater energy savings in green smart buildings [16].
In [16], artificial and daylight lighting can be measured and recorded through sensors in the local area network or a set of data logger devices to modify the light intensity and thus its energy consumption. Tang et al. [19] investigated a new method to perform closed-loop color control of lighting systems in smart homes using cameras available on modern smartphones.
Elejoste et al. [20] proposed an intelligent street light management system based on LED lights deployed in existing facilities. The proposed wireless communication technology significantly reduced the investment cost of traditional wired systems.
Uhm et al. [26] designed and implemented a power-aware LED light enabler with a light sensor, motion sensor, and network interface. The LED light enabler also communicates with context-aware middleware using an intelligent power gateway that uses sensing data to analyze user life patterns to determine optimal power control adaptively.
With the advancement of WSN technology, it is now easier than ever before to monitor industrial buildings, commercial buildings, and offices. Due to the flexible distribution of WSN equipment, WSNs are used in a variety of virtual applications, including industry, medical care, security, and environmental monitoring.
WSNs are the backbone of cyber-physical system (CPS) applications [30]- [32]. Each device contains a network node that integrates computing, wireless communication, power management, and sensing functions to collect and process data from sensors, usually with cooperative coordination [33]. Combining WSNs with LED lights and advanced drivers can greatly reduce lighting power consumption in multiple applications [21].
Recently, WSNs have been applied to energy savings, such as light control [34], [35]. In [34], light control was used to study the trade-off between energy consumption and user Lighting in an office space or home is usually directly controlled by an on/off switch. Of course, users can connect lighting equipment to a personal computer (PC) to remotely turn on or off lights, but this PC must control the operation 24 hours a day, leading to high costs and the consumption of a considerable amount of electricity [36]- [39]. In some designs, specific hardware and software must be installed to control the lights, resulting in unacceptable costs. In addition, this type of system cannot detect human body temperature or indoor light intensity [40], [41].
Kaleem [42] studied an outdoor lighting system based on ZigBee. Sensors controlled the system parameters. Experiments showed that the system reduced energy use by 70.8% in outdoor street environments. Wang [43] studied a WSN-based smart lighting control system. Based on the illuminance and uniformity of the light, he designed a smart lighting control algorithm that can be stepped down.
References [44] and [45] presented a system composed of an infrared human motion sensor (passive infrared (PIR) sensor), Arduino Uno, and a 2-channel relay module that used the PIR sensor to detect motion and control light through computer processing. The function of this system is to automatically turn on the lights when someone enters the classroom and to automatically turn off the lights when no one is there. Wahyuni & Irawan [46] studied a similar intelligent lighting system in which the PIR sensor detects the presence of people through body temperature and the movement of people in the room.
Mohamed et al. [47] studied the use of two PIR sensors to detect the entry and exit of people at the entrance and then determined the brightness of the lighting according to the number of people in the space. The more people there are, the higher the brightness, thereby avoiding energy waste. However, there is a problem with this system. When the number of people is small, the space's lighting is obviously insufficient, and the number of people should not determine the brightness of the lighting.
The most significant problem with current smart lighting systems is that although they can reduce energy, they are usually costly and difficult to install and are mostly used in simple spaces such as toilets and corridors. They are not widely used in complex spaces such as offices that include complex tasks. Therefore, this article proposes a leaderfollower smart office lighting control system based on Internet of Things (IoT) technology that is very suitable for office spaces.
The proposed system is a low-cost, wireless, easy-toinstall, and adaptable leader-follower smart LED lighting system. When the system's counter is zero and the infrared human body movement sensor (PIR sensor) cannot detect people, the system turns all the indoor LED lights to "lowlight mode." In contrast, when the counter is not 0, the system turns all the indoor LED lights to a "high-light mode." In addition to change the brightness of the LED lights in the same area at the same time to save energy, the LED lights can be directly parameterized through the IoT. Table 1 compares the proposed system and related works.
Previously, we developed a new type of LED sensor light with a low-light mode [48]. We changed the full-dark mode of the general sensor light to a low-power low-light mode for standby, which solved many problems with the general sensor light. When a person approaches the sensor light, their eyes will not feel uncomfortable due to strong light, and the sensor light is in a weak light state when no one is passing by. People who return at night no longer need to face dark corridors, achieving multiple purposes.
In addition, the power consumption of the new sensor light in low-light mode is only 1/10 of that in high-light mode, but its brightness is only half of that in high-light mode. It is very suitable for sensor lighting in building parking lots, corridors, stairwells, or toilets.
The original light tube can be directly replaced without changing the light fixture, and the basic brightness can be maintained, as shown in Fig. 1 [48]. This work is based on the system described in [48] and transforms it into a low-cost, easy-to-renovate smart lighting control system suitable for office environments.
Furthermore, the main idea and goal of [50] are regionalized synchronization, while the proposed system is designed for smart office lighting applications. Through the PIR sensor on the lamp, the energy-saving lighting in the space works in conjunction with the system counter above the office door. The LED lights in the room switch to the low-light mode together only when it is confirmed that there is no one in the office, so there is no misjudgment in the proposed system unlike for the system in [50].

A. THE DESIGN CONCEPT
The design concept of this work is based on [48], modified its algorithm for calculating the number of people in space regarding [47], and [ 50] to design and implement a leaderfollower smart lighting control system suitable for the office environment. The design concept is introduced in detail sequentially.

FIGURE 1. A new type of LED sensor light (T8 LED light tube) [48].
In an office, busy schedules and hard work sometimes result in people forgetting to turn off lights, which ultimately results in a waste of electricity. The office lighting system we propose can reduce power consumption and optimize power. Fig. 2 shows an application scenario of the proposed system installed in an office. Fig. 2 (a) is a side view, and Fig.  2 There are two PIR sensors above the office door. The sequence of the two PIR sensors is used to determine whether a person is entering or exiting the office. If the PIR sensor outside the door senses the person's presence first, then the person is judged to be "entering", and the system counter is increased by 1. Otherwise, if the PIR sensor inside the door senses the person's presence first, then the person is judged to be "exiting", and the system counter is decreased by 1.
In the beginning, the system counter=0; when someone enters the office from outside, the counter increases by 1, and LED 1 closest to the door is also triggered as long as the LEDs in the same office (area) are on the same channel (only the LED lights of the same channel can send signals to or receive signals from each other, as explained in detail in the next section). At this time, trigger signals are transmitted to the other LED lights in the same area through the Wi-Fi wireless transmission module. The LED lights in the same area all simultaneously change to the high-light mode.
In Fig. 2, when someone approaches LED 1, LED 1 immediately transmits the trigger signal to LEDs 2 to 6. There are two ways to receive the signal: one is directly receiving the signal sent by LED 1, and the other is receiving a signal sent by the other LEDs that received the signal  Take LED 6 as an example. It can receive the signal sent by LED 1 or the trigger signal sent by LED 5 (whichever signal arrives first). After the trigger signal of LED 1 (as a leader) is received, LED 2 to LED 6 (as followers) immediately operate in high-light mode, so LED 1 to LED 6 change to the strong light mode nearly simultaneously. Different areas are defined by different channel settings. LEDs in different areas are not triggered. The user can set the areas to which the LEDs belong according to the environmental requirements. When LED lights are set to the same area, they act synchronously.
The LED light used in the proposed smart office lighting system in this paper is similar to the induction light proposed in [1] (it is equipped with a PIR sensor), so the strong light lasts for T seconds (the parameter T can be set through IoT; the default value is 30 seconds). When the PIR sensor of the previous LED light cannot detect a person, the lights automatically change to low-light mode, but the system is modified to the requirements that the PIR sensor cannot detect people and the system counter = 0.
These two conditions must exist at the same time for the LEDs to be turned to low-light mode. This method prevents system misjudgment and clearly prevents turning off of lights when someone is there. Because people in a space may stay still for a long time, the PIR sensor alone cannot detect them, and misjudgment can occur.
Therefore, when someone is present in the office, a person entering only increases the system counter by 1 and does not cause the LED lights to change (because the high-light mode is maintained); if someone exits the office, this only decreases the system counter by 1 and does not change the  No system to low-light mode (because the counter is not equal to zero). Only after the last person in the office leaves for T seconds (the default value is 30 seconds) do all LED lights change to low-light mode. Because the proposed system does not require the replacement of the light fixtures of the original LED lights, we can directly replace old LED tubes or fluorescent tubes in the office with new sensor lights and then install PIR sensors inside and outside the door. Hence, the installation is simple; there is no need for wiring and no need to set up a server, and the original light fixtures can be used, allowing the smart LED lighting control system to reduce energy consumption.
Furthermore, we take [26] as an example. In this article, the energy-savings concept is the same, but a server and complicated wiring must be set up to control the brightness of each LED light and the route that people pass by. It takes considerable time and money to achieve a 58% energysavings effect. In our design, as long as the general T8 LED tube is replaced with a new induction light, no server or wiring is required, and good energy-savings effects can be achieved. We use the proposed system in Fig. 2 as an example. Each lamp tube costs NT$800, and the door control circuit and PIR sensor cost NT$1,200. The total cost is 800×6+1,200 = NT$6,000 (approximately US$241.29).

B. THE PROPOSED SYSTEM FLOWS AND ARCHITECTURE
A flow chart of the proposed system is shown in Fig. 3. The proposed system first establishes a connection with the cloud server. When there is no connection, the proposed system repeatedly tries to establish a connection. After the connection is established, the proposed system waits for a command from the application and starts to receive external triggers. The command of the application program is to set the LED parameters (e.g., high-light duration T, high-light mode power %, and low-light mode power %). Then, the proposed system checks whether the connection is continuous and repeats the whole system flow.
The proposed system first receives the trigger of the door PIR sensor. If the entry of a person is detected, the system counter increases by 1; if the exit of a person is detected, the system counter decreases by 1; if no one enters or exits, the counter remains unchanged, and then the proposed system judges the value of the counter. If the counter is not 0, all LEDs are set to high-light mode.
The time continues for T seconds. If the counter = 0, the system immediately checks whether the PIR sensor of each LED has been triggered (indicating that there are still people in the office); if so, all the LEDs are set to high-light mode. If the PIR sensor of the LED is not triggered, then the counter = 0, and no one is in the office at this time. Therefore, all LEDs are set to low-light mode (we wait until the T-second countdown of the high-light mode expires before turning to low-light mode).    Therefore, when the counter>1, a person entering only increases the counter by 1. If someone exits the office, it only decreases the counter by 1, and the LED lights remain in high-light mode. All the LED lights change to low-light mode only when the last person in the office leaves, T seconds have elapsed, and the PIR sensor of the LED has not been triggered.
When the PIR sensor of the LED light cannot detect anyone and the system counter=0 simultaneously, the LEDs are changed to low-light mode. This method can prevent many situations of misjudgment. For example, if there is no one in the office at the beginning, the counter will erroneously be set to 1 if two people simultaneously enter side by side. When one person walks out of the office, the counter will change to 0, but the PIR sensor of the LED light will detect that there are people in the office at this time.
Therefore, the LED lights are not changed to low-light mode, and when the last person leaves the office, the counter remains at 0 (we set the counter to never become negative), and the LED lights change to low-light mode. Previous systems reported in the literature have installed only a system counter or used only a PIR sensor for detection. Under these conditions, it is very easy to misjudge the number of people, resulting in malfunctions (the lights are turned off when someone is present or are still on when no one is present).
The architecture of the proposed system is shown in Fig.  4. The proposed system can upload the LED high-light duration, high-light mode power %, low-light mode power %, and other parameters, and the leader-follower mode control data are sent to the cloud through Wi-Fi wireless transmission.
The terminal device performs parameter settings. The terminal device can be a mobile phone, tablet, or computer. When the leader LED light receives commands from the cloud, it immediately transfers the command to other follower LED lights via Wi-Fi wireless transmission. The parameters of the LED lights in the same area can be easily set together instead of individually setting them. The settable range and preset value of the LED light parameters of the proposed system are shown in Table 2. The parameters that can be set are the duration of high light T, the power of the high-light mode WH, and the power of the lowlight mode WL. These parameters are all set. The default values are suitable for office lighting. The user can connect to the LED lights to modify them through the internet. The   setting range is shown in the table. The modified parameter values are also uploaded to the Wi-Fi wireless transmission module. The cloud platform is used for big data analysis and suggestions for setting parameter values in the future. A prototype of the proposed system is demonstrated in Fig. 6.

C. HARDWARE IMPLEMENTATION OF THE PROPOSED SYSTEM
The LED lights in the office operate in low-light mode at the same time when there is no one present. When any of the LED lights in the office detect people passing by (as shown in Fig. 7), it transmits the trigger signals through the Wi-Fi wireless transmission mode to the other LED lights in the same area and simultaneously adjusts the brightness to the high-light mode instead of sequentially lighting up the LED lights one by one.
The communication module in Fig. 6 is a Wi-Fi wireless transmission module responsible for the data transmission to the cloud and transmission between the LED lights. In the transmission method between the LED lights, the LED lights set to the same channel send signals to or receive signals from each other. For example, the channel of the leader LED light in Fig. 8 is set to '000' (the dual inline package (DIP) switch is set to '000').
We use Wi-Fi wireless transmission to transfer the set parameters of 000 to other channels and to the follower LED lights. This information cannot be received by the other channels. This is a function designed for a large field because, in a large field, there may be hundreds of LED lights that need to be set. Setting so many LED lights is a very large project, so the channel value of the light can simply be set the same.
Take the office in Fig. 2 (b) as an example. We can set the channel of all LED lights in the same office to '001' and those in different offices to different channels so that a large number of LED lights can be set easily. The user can set the area of the LEDs according to the environmental requirements, and when the LED lights are set to the same area, they act synchronously. In addition to the various parameters set through the DIP switch of the LED control circuit, parameters can be set via the mobile device app, as shown in Fig. 9.
In this work, a Wi-Fi wireless transmission module is adopted as the communication module in Fig. 5. Moreover, this Wi-Fi wireless transmission module can be replaced with ZigBee, Bluetooth, LTE, etc., according to different field conditions. We note that the communication module should support universal asynchronous receiver/transmitter (UART) interfaces for communication with the MCU module used in the system proposed in this work.
The data transmission method of the Wi-Fi communication module can generally be divided into two types: message queuing telemetry transport (MQTT) and HyperText Transfer Protocol (HTTP). In this article, we adopt the MQTT architecture [51]. Compared to the HTTP architecture, the MQTT architecture can reduce power consumption during transmission, and the proposed system will synchronize the time with the primary server every minute, avoiding the problems caused by time out of synchronization and data delay [2], as follows. 1) The finite element machine architecture is adopted to write programs, and the device will automatically ignore it to avoid errors when assisting false data injection [52]. 2) A timestamp is adopted for the data recognition judgment, so that the same timestamp will only recognize one piece of data, and the data with errors due to delay will be ignored. 3) Due to the solution of the first two problems, system resources can be prevented from being exhausted due to multiple occupation.

IV. ENERGY SAVINGS CALCULATION AND DISCUSSION
Each LED induction lamp calculates the energy consumption per hour and year. Then, it is compared with the noninduction LED light to calculate its energy-saving rate (ESR), the number of electricity consumption savings, cost savings, reduced carbon emissions, etc. The proposed office lighting system's LED lights have high-light and weak light modes, so it is more complicated to calculate the power consumption, unlike ordinary non-induction LED lights. Factors that affect the power consumption include the frequency of people entering and leaving the office, the duration of high-light T, the power of high-light mode WH, the power of low-light mode WL, the presence or absence of continuous triggering. To facilitate the deriving of the formula, we refer to calculation methods [2], [48], [49] of power consumption. Finally, consider a company as an example to calculate the total savings for all of its offices. WL: The power of the LED light in low-light mode. The default power in this work is 1.5 W in low-light mode (default value is 10%) WE: The average power consumption of the LED light circuit (including the power consumption of the control circuit and PIR sensor); the WE value in this work is approximately 1 W. Fig. 10 shows that the use of the new type of LED light sensor maintains the activated state within 1 hour (the life cycle state is the low-light mode state, not the 0-second lowlight mode). There are 3,600 L seconds to maintain the highlight mode status, so we use the method for calculating the area to simulate and calculate the energy Eph consumed by the new LED induction light: In addition, we assume that the new type of LED sensor light in a company's office is turned on for 10 hours a day, for a total of 10×365=3,650 hours in a year. Therefore, the annual electric energy Epy consumed by each new type of LED sensor light is as follows:

A. CALCULATION OF THE POWER CONSUMPTION AND ENERGY-SAVINGS RATE (ESR)
Because noninductive LED lights are always in the starting state, the electric energy Fpy consumed by each LED light per year compared with noninductive LED lights is (where Fph is the electric energy consumed by noninductive LED lights per hour): First, (4) shows that ESR is inversely proportional to the electrical energy Eph consumed by the new LED induction light per hour. From (1), Eph is inversely proportional to L (because (WL-WH) must be negative), so ESR is proportional to L. The more seconds accumulated in low-light mode, the better ESR is.
Next, to calculate the ESR, the actual power values of WH, WL, and WE must be determined. The power of the LED light in high-light mode is 15 W, the power WL in low-light mode is 1.5 W, and the average power consumption of the circuit WE is approximately 1 W. Below, we bring different L values into the formula for Eph to calculate the Epy, ESR, Dpy, Mpy, and Cpy values of each new type of LED sensor light with different L values (compared with those of nonsensing LED lights). In Table 3, due to space, only the important parameters are listed (the T value is listed at intervals of 120 seconds). Table 3 shows that the maximum ESR of 84.38% occurs when L = 3,600 (indicating that the new LED sensor lights are in low-light mode within 1 hour, that is, there is no light in high-light mode in the entire office within 1 hour); the same is true for Dpy, Mpy, and Cpy. The ESR is smallest when L is zero (ESR=0%) and linearly increases with increasing L. When L = 3,600 seconds, the ESR rises to 84.38%. Fig. 11 shows the relationship between ESR and L. In Fig. 11, ESR is the energy-saving rate, L is the number of seconds accumulated per hour in the low-light mode for each LED lamp, and the relationship between the two is linear. A greater number of accumulated seconds in low-light mode corresponds to a higher the energy-saving rate (no one in the office) so that the system will run in low-light mode only when at the appropriate time to save energy.

C. ENERGY-SAVINGS CASE CALCULATION
We use a company as an example to calculate the total cost savings and energy savings of all its offices. Assume that the company has 16 offices and that each office has 12 LED lights. There are a total of 192 LED lights. For the new type of LED induction light, we have demonstrated its energysavings effects and expected results from reducing electricity consumption over the same period of the year. Because our new LED sensor light can uniformly set the parameters of each LED light in the office through the Wi-Fi wireless transmission module, there is no need to waste effort to individually set them. We use 192 LED lights to estimate the reduced power consumption over the same period of the year. The following assumes that the lights in the office are turned on for 10 hours a day and that the hourly time of the new LED sensor light mode is 1,200 seconds. We assume that the office is occupied for 1,200 seconds and that otherwise no one is there.
We can directly refer to we can obtain the following expected results (assuming that the office is occupied for 10 hours a day, that the hourly time of the new type of LED sensor light in low-light mode is 1,200 seconds, that NT$4 is saved per degree calculation, that the power WH of the high-light mode is 15 W, that the power WL of the low-light mode is 1.5 W, and the average power consumption of the circuit WE is approximately 1 W):

D. USER SATISFACTION ANALYSIS AND DISCUSSION
A total of 180 respondents completed questionnaires in this study, of which 51.7% were males and 48.3% were females. The analysis results are shown in Table 4. The overall average ranking of the satisfaction found that the top three satisfaction levels for sensor light settings are "The respondent thinks that this office lighting system can operate in low-light mode when there is no one in the office space and does not allow the office space to become totally dark during working hours (m=4.9)," with the highest score; "The respondent thinks that the low-light mode function of this office lighting system is very energy-efficient (m=4.85);" and finally "The respondent thinks that this office lighting system can confirm that no one is present in the space and turn all the LED lights in the room to low-light mode together and that there is no misjudgment (m=4.74)." This satisfaction analysis was conducted at Cheng-Shiu University, Kaohsiung, Taiwan. Currently, the proposed system is installed in the life creative building of the school and has been used for nearly half a year. Currently, we are evaluating whether the office lighting system of the whole school has been fully renovated.
Furthermore, the reliability of the proposed system is as high as 95% after the actual test. Errors occur only in rare cases. Generally, errors occur when two people enter side by side at the same time. The proposed system counter and PIR sensor must confirm that no one is in the state before changing to the low-light mode. However, if the person in the office is entirely still, the PIR sensor still cannot sense it. Currently, this is the major problem that must be resolved.
Finally, comparing to the related works, the differences and advantages of the proposed method are summarized as follows. 1) When the office space is unoccupied, the system operates in low-light mode and does not allow the office space to become totally dark during working hours. 2) Through the coordinated operation of the system counter and the infrared human body movement sensor (PIR sensor), when it is confirmed that no one is present in the space, all the indoor LED lights are turned to lowlight mode, and there is no misjudgment.
3) The original light fixtures can be retained, with no wiring and no need to set up a server, and the energysavings purposes of smart lighting in the office can be achieved. 4) The parameters of LED lights can be set directly through replaced with the proposed office lighting system (assuming that the office is occupied for 10 hours a day and that the hourly low-light mode is 20 minutes), the power-saving rate is as high as 28.13%.
Finally, there are some existing smart lighting control products on the market. They are also similar to the proposed system. Hence, we also compare our system to them. Table  5 compares the system developed in this work with two popular smart lighting bulb products: Philips Hue full-color ambient light bulbs [53] and VOCOline smart lighting bulbs [54].

V. CONCLUSIONS
This article proposes an energy-saving, easy-to-install, wireless, low-cost IoT-based leader-follower smart office lighting control system that is very suitable for installation in the office of a large company or factory. It uses a system counter and an infrared human motion sensor (PIR Sensor) working in coordination; only when no one is in the space do all the LED lights in the room enter low-light mode together, and there is no misjudgment.
The system can easily achieve energy savings and convenient control of a large-scale smart LED lighting system. Moreover, the cost of this system is only 1/10 that of installing a smart system, and there is no need to replace the original LED lights. The original light fixtures can be used in the system without additional wiring or setting up a server.
In addition to directly setting the parameters of LED lights through the IoT, unlike general induction lights, the recommended office lighting system operates in low-light mode instead of total darkness when there is no one in the office space. Taking a general 15 W T8 LED tube (induction light) as an example, if users switch to the recommended office lighting system (assuming that the office is occupied for 10 hours a day and that the hourly low-light mode time is 20 minutes), the power-saving rate is as high as 28.13%.
In addition, when the system counter is 0, but people are still present in the office, the PIR sensor is still unable to sense if the people in the office are completely still. Therefore, in the future, we will fine-tune the strategy and add a judgment condition that the count is set to be 0, but when the PIR sensor senses that people are still present in the office many times, the counter will be changed to a policy of 1, solving this problem. Furthermore, in future work, we plan to focus on considering the lighting effects of natural light so as to obtain increased energy savings and make the system more efficient.