Open Access
Add to BibSonomy
Abstract
An electromagnetic interference (EMI)-free wide-range indoor dust monitoring system that employs the optical orthogonal frequency-division multiplexing (OFDM)-based visible-light communication (VLC) is proposed. For the long-term transmission of dust information, VLC can be utilized even in EMI-restricted areas, such as medical centers, emergency rooms, and nursing homes. Discrete cosine transform-based optical OFDM is adopted to transmit a large amount of dust information. For robust light detection from eliminate ambient light and low-frequency noise, an average voltage-tracking technique is utilized and as a result LED illumination is detected over 18 m distance with reliable error rate. Wide-range dust information from multiple dust sensors are clearly displayed through the designed user interface. Users can then monitor the air quality in real-time, improving the environmental awareness of individuals.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
In the past few years, air pollution has drawn considerable research attention. Public concern regarding air pollution has increased significantly owing to serious health hazards [1,2]. Among air pollutants, fine particulate matter (PM2.5), which describes pollutants that are less than 2.5 μm in diameter, draws the greatest concern [3]. PM2.5 particles penetrate the deepest parts of the lungs, such as the bronchioles and alveoli (gas-exchange regions), and even affect other organs by passing through the lungs. Studies have revealed that PM2.5 may cause serious health problems, such as respiratory diseases, cardiovascular disease, asthma, lung cancer, birth defects, and premature death [4–6]. Thus, monitoring the air quality, particularly the PM2.5 concentration, is very important. Recently, various studies have been performed for monitoring the outdoor air quality, e.g., monitoring using a low-cost urban PM2.5 monitoring system [7], dust information monitoring after blasting in open pit mines [8], and large-scale PM monitoring [9].
It is reported that humans spend more than 90% of their lives in indoor environments; thus, the indoor air quality must be managed [10,11]. Indoor air quality, including the dust concentration, differs from outdoor air quality [12]. For patients with respiratory illnesses, accurate PM2.5 information in living spaces is very important, and the PM2.5 concentration in indoor environments should be monitored. In particular, air-quality information is usefully in public places where people with poor immune systems are concentrated, such as medical centers and emergency rooms.
Visible-light communication (VLC), which has emerged as an alternative solution to short-range radio frequency communication, can be utilized for the long-term transmission of indoor air-quality information without electromagnetic interference (EMI). VLC, which covers the visible-light wavelength (380–740 nm), has various advantages. For example, it can be utilized for both illumination and wireless links, is EMI free, is harmless to the human body, has a large unregulated bandwidth, and provides security against unwanted wireless connections because visible light does not penetrate through building walls [13]. Therefore, it can be employed for real-life applications, such as vehicle-to-everything communication [14], indoor navigation [15], environmental monitoring [16], and healthcare monitoring [17,18]. Important environmental data, including PM2.5 information, can be transmitted via VLC even in places where EMI is restricted, e.g., medical centers, nursing home and hospitals, which are associated with respiratory diseases. Thus, medical experts and patients can monitor the indoor air quality in real time, improving the environmental awareness of individuals.
For transmitting and monitoring a large amount of dust information, optical orthogonal frequency-division multiplexing (OFDM) can be used [19–21]. As major optical OFDM schemes, asymmetrically clipped optical (ACO)-OFDM and direct current (DC)-biased optical (DCO)-OFDM have been studied [20]. ACO-OFDM has a merit regarding the optical power consumption, as only the symbols in the positive real part are transmitted, while half of the subcarriers are used to carry data. In DCO-OFDM, all the subcarriers carry data symbols, and this scheme is less efficient than ACO-OFDM with regard to the average optical power. Recently, discrete cosine transform (DCT)-OFDM, which can provide both an effective bandwidth and low-power transmission, was introduced in fiber optics field [21]. A study on DCT-OFDM was performed in laboratory environments with field-programmable gate array (FPGA)-based offline processing [22]. Considering its merits, such as its bandwidth efficiency and power consumption, DCT-based optical OFDM is promising for the transmission of a large amount of dust information in real time.
In this study, a DCT-OFDM-based VLC system is demonstrated for EMI-free wide range indoor PM2.5 monitoring. The focus of this study is to implement a cost-effective and low-complexity fine particulate matter monitoring system for providing wide-range indoor air quality trends in real time. To transmit a large amount of dust information through VLC, a DCT-based optical OFDM system is designed. For robust optical detection over a wide range, an average voltage tracking (AVT)-based optical detection scheme is adopted. The contributions of this study are three fold: 1) this is the first work that proposes the use of DCT-based VLC for monitoring wide-range multiple PM concentration, 2) the fixed-point DCT-OFDM has low implementation costs and a simple design, 3) the feasibility of long-range dust-data optical transmission with a low error rate in real time is demonstrated. To avoid misunderstanding due to various abbreviations in this work, useful abbreviations are listed in Table 1.
Table 1. Abbreviations and acronyms
2. DCT-based optical OFDM
A novel technique for optical OFDM, which employs the DCT instead of the inverse fast Fourier transform (IFFT) to ensure real-value signal transmission, was introduced in the field of fiber optics [21]. DCT-based OFDM can be combined with conventional optical OFDM (i.e., ACO-OFDM or DCO-OFDM) to improve the spectral efficiency or power efficiency. In this study, to increase the bandwidth efficiency, we focus on DCO-DCT-OFDM, because DCO-OFDM fully utilizes the subcarriers. Assuming the N-DCT transition (where N is the number of subcarriers for the DCT), DCT-DCO-OFDM has N bandwidth efficiency; i.e., subcarriers are fully utilized among the N subcarriers. In Table 2, the characteristics of DCT-based OFDM are compared with those of ACO-OFDM and DCO-OFDM.
Table 2. Comparison of conventional OFDM and DCT-based OFDM
2.1 DCT-DCO-OFDM
In a general OFDM scheme, the IFFT and fast Fourier transform (FFT) are used, and the function consists of a cosine wave (real part) and a sine wave (imaginary part). With DCT-OFDM, only a cosine wave is utilized for signal conversion, and the N-order inverse DCT (IDCT) and DCT are defined as follows [21]:
where is when ; otherwise, . An input symbol to the IDCT block is defined as , where is the M-level pulse amplitude modulation (PAM) symbol. The number of bits per symbol is defined as .In DCT-DCO transmission, the output of the IDCT block in the time domain, , is a real number, but to transmit signals below zero via light-emitting diode (LED) light, a suitable DC-bias is added, and remaining negative peaks are clipped. The DCT-DCO-OFDM symbol is then defined as [20]
where , which is related to the power of , is the DC-bias parameter and is derived as [20]Here, is the proportionality constant and the DC-bias level can be defined as [20]. In Fig. 1, an example of a DCT-based DCO-OFDM symbol is shown. Under the condition of 16-PAM, N-IDCT (), and a 13-dB bias, the IDCT block output is as shown in Fig. 1(a), and the DC-bias signal is as shown in Fig. 1(b), where the solid black lines and red lines with circle symbols represent the analog DCT-OFDM signal and the discrete signal, respectively.At the receiver, after the removal of DC-bias component (), the DCT block is processed. The cosine output in Eq. (2) has even and odd symmetry when index is even and odd, respectively. Thus, the iteration loop in Eq. (2) can be reduced, and the computational complexity can be reduced via the following operation [21]:
Finally, the recovered sequence, , can be derived from the DC-removed signal () using Eq. (5), as follows:
Table 3 compares the computational complexity between FFT-based OFDM and DCT-based OFDM [21]. DCT-based OFDM requires less multiplications and less additions than FFT-based OFDM. The reduction of the computational complexity increases with N. Moreover, DCT-based OFDM does not need a Hermitian symmetry operation; therefore, it is useful for digital signal processing, e.g., simple one-dimensional modulation and real-valued operations.
Table 3. Computational complexity of N-DCT and N-FFT
2.2 Implementation of fixed-point optical OFDM
To ensure real-number OFDM transmission, DCT and IDCT schemes are adopted for wide-range PM2.5 monitoring. However, there are problems regarding the implementation of all levels of a real number. In this study, we consider truncated OFDM symbol transmission, which is implemented using a fixed-point 8-bit decimal. Figure 2 shows the signal conversion from the DCT-OFDM symbol to the fixed-point data. First, the generated OFDM symbol is quantized by the defined signal level, where 8-bit data are considered. Next, the quantized values are converted into a value in the range of 0–255. Finally, the converted value (0–255) is transmitted via a universal asynchronous receiver/transmitter (UART) interface. Thus, LED transmission of the DCT-OFDM symbols is simply implemented; however, there is a potential error component due to signal truncation. In Fig. 3, curves obtained from a floating-point simulation and a fixed-point simulation (8-bit decimal) executed in MATLAB under the conditions of a 13-dB DC bias and the modulation levels of binary phase-shift keying (BPSK), 4-PAM, 8-PAM, and 16-PAM are compared. The bit error rate (BER) gap between the floating point and fixed point increases with the modulation level and is particularly large for the curve with 16-PAM. This degradation is caused by two main error components: fixed-point expressions of small decimals and the effect of zero clipping. In actual optical transmission, fixed-point implementation with DCT-based OFDM symbols is used and the details are presented in Section 5.1.
3. Selection of dust sensors
In this study, a PMS (Plantower) dust sensor was employed for wide-range PM2.5 monitoring. A recent study using PMS series showed that it is reliable and stable compared with professional instruments, such as TSI AM510 and GRIMM Model 1.109 as the ground level, particularly for PM2.5 measurements [23]. The PMS series employs a “virtual impactor” to separate different-sized particles and measures the light scattering (laser light) to estimate the concentration of particles. To test the measurement of the PMS5003 sensor, PMS dust sensors were connected to the processor via a UART and programmed to transfer dust information at specified time intervals; here, an Atmega128 processor was employed. Each set of data was obtained using the moving-average method, with 20 samples per minute. Figure 4 compares the moving-averaged data and current data for actual PM2.5 measurements performed in a test on Feb. 21, 2019. The current data fluctuate around the averaged data owing to various factors, such as cleaning and smoke from outside.
For a reliability test of PMS5003, outdoor dust measurements were compared with government-certified measurements (GCMs) provided by the Korea Environment Corporation [24]. An air-quality report was provided every hour, and our experiment location was 1.5 km from a government measuring station. The beta-attenuation method, which is a widely used air-monitoring technique that employs the absorption of beta radiation by solid particles extracted from air flows, was used for the GCMs. For reliable data analysis, we collected PM2.5 information in three different months: June 17–19, 2018; September 22–24, 2018; and December 22–25, 2018. In Fig. 5, the coefficient of determination () is used to evaluate the correlation between the government-certified data and the measured data [25]. ranges from 0 (non-correlated) to 1 (perfect fit). It is generally considered that coefficient values above 0.7 ( > 0.7) indicate a high correlation between test data and reference data [25]. According to the graph, the coefficients of determination for the two sets of data (measured data and government-certified data) were . Considering the experimental conditions, such as the distance from the government measuring station (1.5 km) and the office conditions (e.g., room cleaning, dust from printers, and people moving), the results indicate a high correlation.
4. System model
Wide-range PM2.5 monitoring can be applied in various locations where dust monitoring is required, such as medical centers, basement parking lots, indoor gymnasiums, and cement factories. Figure 6 illustrates the concept of implementing VLC in real life. In Fig. 7, a block diagram of the DCT-based optical OFDM system is presented. A large amount of dust information from multiple dust sensors is transferred using a single LED via DCT-based optical OFDM, and the light illumination is detected by a single photodiode (PD). After the demodulation process including the DCT and PAM, the collected data are stored on a database server and can be monitored using a smartphone or personal computer (PC) at the lighting facilities. Thus, the indoor air quality can be checked via a user interface in real time or at a later time. Additional details are presented in the following subsections.
4.1 Sensor modules and interface
To transmit data from multiple sensors without data loss, a reliable interface module between the sensors and the transmitter processor is required. In this study, ATmega 2560 is employed, because data transmission of PMS5003 is based on a UART, and ATmega2560 includes four UART channels. Thus, one ATmega2560 can control four dust sensors. Figure 8 shows the interface between multiple sensors and the microcontroller for the transmitter module. The transmitter module consists of a master microcontroller, a slave microcontroller and multiple dust sensors. The master microcontroller is connected to the slave microcontroller via a two-wire interface (TWI), which has high performance for multiple connections with a synchronization clock. Each dust sensor is connected to the slave microcontroller via the UART. For multiple tasks, the slave microcontroller employs FreeRTOS operating systems. FreeRTOS is a powerful multiprocessing library builder (multiple tasks are simultaneously executed via time-division processing) for microcontrollers such as ATmega. Thus, the slave microcontrollers can save the data-collection time, and multiple dust sensors simultaneously transmit PM2.5 information to ATmega2560. In Fig. 9(a), the multiple FreeRTOS-based tasks (task1–task6) for each slave microcontroller are shown. The task loops are shown in Figs. 9(b)-9(d). The collected data are transferred to the main processor via the TWI when there is a call from the main processor. In the master microcontroller, the collected data are ready to be processed for optical OFDM transmission.
Fig. 9 FreeRTOS-based task loops (slave microcontroller). (a) FreeRTOS-based slave MCU tasks. (b) Loop for task1–task4. (c) Loop for task5. (d) Loop for task6.
4.2 Transmitter module
To transmit multiple data collected from the slave microcontrollers, data are converted for optical transmission via PAM modulation followed by DCT-based optical OFDM in the master microcontroller, as shown in Fig. 7. First, multiple bits are modulated via M-PAM. Next, optical OFDM symbols are generated by the IDCT defined in Eq. (1) with an N-symbol input. After DC-bias and zero-clipping, the output of Eq. (3) is converted into a fixed-point decimal (8-bit) for UART transmission. Here, the data streams are prepared in a one-frame format (unit), which consists of (N + 3) bytes, i.e., a start byte (0xFF), a header byte (0x80), N OFDM symbol bytes, and an end byte (0x00). Figure 10 shows the designed transmitter module, where a plano-convex lens is used for long distance transmission. The transmitter module includes eight dust sensors (PMS5003), microcontrollers (one master and two slaves), a single LED (Hyper Flux 5 pie, red), an optical circuit, and the plano-convex lens.
4.3 Receiver module
Figure 11 shows a prototype of the DCT-based OFDM receiver module, which includes a PD, our custom-made optical-detector circuit, and a microcontroller (ATmega128). The optical-detector circuit consists of a voltage regulator (IC1), a buffer (IC2), an amplifier (IC3), an AVT block (IC4), a differential amplifier (IC5), and a comparator (IC6). Here, a recently introduced AVT block (IC4) was employed to eliminate ambient light and low-frequency noise below 100 Hz [26]. This allows strong light detection even when the light source is weaker than the ambient light. The cutoff frequency is set as 160 Hz, according to the equation for the RC filter cut off frequency: (R = 10 KΩ and C = 100 nF are used).
After the output signal passes through IC2–IC6 to obtain the required electric properties, it goes to the microcontroller (ATmega 128) via the UART1 channel and can be processed. First, a starting point is synchronized with the 0xFF and 0x80 bytes. Next, the N bytes are saved for DCT-OFDM demodulation. After the removal of the DC-bias, the N-symbol input is demodulated using Eq. (6), and multiple dust data are recovered. The recovered dust data are arranged in the one-frame format, i.e., a sync byte (0xFF), a header byte (0x80), dust data bytes (), and an end byte (0x00). Finally, the data are transferred via a UART connection to the user interface in order to provide the real-time wide-range PM2.5 information.
5. Experimental results
Figure 12 shows the experimental setup. A single LED (Hyper Flux 5-pie, red) and a single PD were employed to transmit wide-range dust data in a corridor at night. The proposed system was evaluated under various conditions, and the parameters considered for indoor PM2.5 monitoring with VLC are presented in Table 4. A plano-convex lens [17] can be used at the transmitter end to ensure light transmission across a room, e.g., an indoor gymnasium (transmission coverage over 10 m). The light source can be focused and the transmission distance extended without an additional LED or power supply.
Table 4. Experimental parameters
5.1 Transmission bandwidth of DCT-based OFDM
To precisely estimate transmission bandwidth in intensity modulation direct detection (IM/DD) optical transmission, various conditions, such as sampling frequency, current input, max/min voltage levels, signal to noise ratio, processor speed, and modulation scheme, have to be considered. Referring to recent study [27], the data rate of the IM/DD optical transmission system is calculated as:
where is the single-sided bandwidth of the system and denotes the spectral efficiency of the system. and are derived as, where is the BW efficiency listed in Table 2, is the sampling frequency, is the sign function, and is the cyclic prefix length. Without loss of generality, it is assumed that average signal level is constant in each frequency band (no missed subcarrier) and that is omitted in this study. Then, the spectral efficiency of Eq. (9) is revised as:Considering signal processing of low-cost microcontroller (ATmega 128), is stable for real-time transmission. Therefore, the data rate of the proposed system (DCT-OFDM) is derived as with conditions of, , and . With same conditions except , the data rate of the conventional optical OFDM (DCO-OFDM and ACO-OFDM) are estimated as (DCO-OFDM, ) and (ACO-OFDM, ), respectively. From the analysis, the proposed OFDM system achieves higher transmission bandwidth and a maximum data rate could be improved adopting high-performance microcontroller (high sampling frequency).
5.2 DCT-based optical transmission
The light intensity measured within the corridor (ambient light: 50–100 lux) with a plano-convex lens is shown in Fig. 13. A digital lux meter (MS6612) was used to measure the light intensity. As the main noise light source, an LED lamp was used, which was powered by a 220-V alternating-current supply with a 60-Hz frequency. In the graph, the curve with blue triangles (“Source Light Only”) represents the desired source light without ambient light, and the curve with red circles (“Total Rx Light”) represents the total received light intensity with ambient light. The intensity gap between “Source Light Only” and “Total Rx Light” increases with transmission distance. This is because the light illumination decreased relative to the fixed ambient light with the increase of the transmission distance; i.e., the light from the light source was greater than the interfering light at a short transmission distance, whereas the interfering light was more dominant at a long transmission distance.
The BER was analyzed with interfering ambient light, as shown in Fig. 14 For evaluation, sample signals collected from dust sensors were used for each modulation (BPSK, 4-PAM, 8-PAM, and 16-PAM) and 106 packets of data were transmitted using the proposed DCT-based optical OFDM method through the UART connection to the PC. To check the BER, the received signal was compared to reference data at the receiver module. As the transmission distance increases over 18.5 m, a rapid performance degradation occurred in the order of 16-PAM to BPSK. The single LED (Hyper Flux 5pie) only needed below 0.5 W (low power consumption). However, the BER result with the plano-convex lens is remarkable: the graph indicates reliable transmission (error-free with 106 packets) at a transmission distance of 18.5 m for all the modulation cases (BPSK, 4-PAM, 8-PAM, and 16-PAM). Thus, the transmitted multiple dust data were clearly observed on the designed monitoring interface at 18 m, as shown in Fig. 15. The experimental results indicate that the designed transmitter module can achieve cross-room transmission, e.g., in an indoor gymnasium, in real time.
Fig. 15 Designed user interface for wide-range PM monitoring (PC version, monitoring of data from 8 dust sensors at a transmission distance of 18 m).
5.3 Wide-range dust-sensor monitoring
The user interface for monitoring the indoor PM concentration via the PC was designed using Microsoft Visual Studio. Figure 15 shows the designed indoor PM monitoring screen that was used to observe the wide-range PM2.5 and PM10 data in real time. Selections for the location and particle type (PM2.5 or PM10), a room map, and sensor position information are provided. By pressing the Start button after selecting the port number, wide-range multiple dust concentrations are displayed in real time and are recoded for further analysis. For clear dust-level indication for each sensor position, the dust index is provided with four colors: blue (Good), green (Moderate), yellow (Bad), and red (Very Bad). The indices Good, Moderate, Bad, and Very Bad correspond to PM2.5 concentrations of below 15 μg/m3, below 35 μg/m3, below 75 μg/m3, and over 76 μg/m3, respectively and PM10 concentrations of below 30 μg/m3, below 80 μg/m3, below 150 μg/m3, and over 151 μg/m3, respectively. Similar standards have been defined by other countries and organizations [28–30].
To collect the wide-range PM concentration data, eight PMS5003 sensors (S1 and S2 in room 2418; S5 and S6 in room 2430; S3, S4, and S8 in the corridor; and S7 in room 2417) were installed at a height between 1 and 2 m, as shown in Fig. 15. Figure 16 shows the indoor dust history measured by Fig. 16(a) S3, S4, and S8 for February 16–18, 2019; Fig. 16(b) S2 and S8 for February 18–21, 2019; and Fig. 16(c) S5, S7, and S8 for February 19–22, 2019. Figure 17 shows the coefficients of determination between the sensors.
Fig. 17 Analysis of correlation (a) between S3 and S8, (b) between S8 and S4, (c) between S2 and S8, (d) between S5 and S7, and (e) between S7 and S8.
As shown in Fig. 16(a), PM2.5 information was collected from three sensors in the corridor (refer to the sensor positions shown in Fig. 15). The dust levels indicated by the sensors are similar; however, the dust level of S4 is sometimes higher than the others for the daytime. This is because S4 was located near the gate to the outside, increasing the particle count owing to factors such as smoke, movement of people, and dust from the outside. For the correlation between S3 and S8, , and for the correlation between S4 and S8 , as shown in Figs. 17(a) and 17(b), respectively. In Fig. 16(b), the dust data obtained by S2 in room 2418 and S8 in the corridor are compared. The main door in room 2418 was frequently opened by several members of the laboratory. Therefore, the PM2.5 concentration was affected by dust in the corridor, particularly when the main door was open in the daytime. The coefficient of determination was calculated as , as shown in Fig. 17(c). The dust levels between rooms 2417 and 2430, including the dust data for the corridor, are compared in Fig. 16(c). The dust levels in the two rooms (S5 in room 2430 and S7 in room 2417) were similar, except when a member was conducting an experiment in room 2430. Therefore, the dust indices between room 2417 and room 2430 had a very high correlation (). The dust levels in the two rooms were lower than those measured by S8 in the corridor, but we observe a similar trend between two sensors (S5 and S7) and S8; the doors in these two rooms mostly remained closed during the dust-data collection. Therefore, the coefficient of determination between the dust levels of S5 and S8 was high (). The wide-range dust-monitoring results indicate that the dust level differed significantly according to the local conditions (smoke input, room cleaning, printer use, movement of people), even indoors. The indoor dust level is usually affected by outdoor dust level; however, the results indicate that the dust levels in the rooms were lower than that in the corridor when the main door was closed in the rooms.
5.4 Further discussion
For practical deployment of the proposed wide-range PM monitoring system, there still remain several issues, such as, bi-directional VLC, backbone network, and overall monitoring and transmission system. For multiple sensors communication, bi-directional VLC is very useful. The practical environment monitoring system, which is eco-friendly, cost-effective and energy-efficient, was introduced [31]. To prevent the information isolated island, a backbone network is significant. For example, power line communication (PLC) could be an excellent backbone network for VLC system. It not only supplies power for VLC system but also transmits data via power line without additional data cable [32,33]. With implementation of bi-directional VLC and backbone network, LED base station located on the ceiling could collect multiple sensor data at each position and a large amount of data could be stored in the database sever through backbone network in real time. Then, the user could monitor wide-range PM information under indoor lighting facilities by user interface at any time.
6. Conclusion
A DCT-based VLC system for wide-range EMI-free PM2.5 monitoring is proposed. Fixed-point implementation-based optical OFDM transmission was implemented with low cost and low complexity. The proposed system has the following advantages: VLC-embedded wide-range indoor PM monitoring, a simple optical OFDM implementation for long distance transmission, and real-time remote PM2.5 monitoring via our self-designed user interface. According to experimental results, the proposed system can transfer eight different sets of PM2.5 data simultaneously over a distance of 18 m with a low error rate, allowing the dust level to be easily monitored through the designed user interface. In the experiment, the indoor dust information differed in different areas, depending on the local conditions. Accurate dust information is very important for patients with poor immune systems. Moreover, owing to the continuous deterioration of air quality worldwide, studies on smart dust forecasting are required. The proposed monitoring system is a useful reference for future smart air-quality forecasting systems with EMI-free characteristics.
Funding
Mid-Career Researcher Program through an NRF Grant funded by MSIT (2016R1A2B4015).
References
1. S. Fuzzi, U. Baltensperger, K. Carslaw, S. Decesari, H. Denier van der Gon, M. C. Facchini, D. Fowler, I. Koren, B. Langford, U. Lohmann, E. Nemitz, S. Pandis, I. Riipinen, Y. Rudich, M. Schaap, J. G. Slowik, D. V. Spracklen, E. Vignati, M. Wild, M. Williams, and S. Gilardoni, “Particulate matter, air quality and climate: lessons learned and future needs,” Atmos. Chem. Phys. 15(14), 8217–8299 (2015). [CrossRef]
2. World health organization, 7 million premature deaths annually linked to air pollution. [Online] Available: http://www.who.int/mediacentre/news/releases/2014/air-pollution/en/
3. C. A. Pope 3rd, R. T. Burnett, M. J. Thun, E. E. Calle, D. Krewski, K. Ito, and G. D. Thurston, “Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution,” JAMA 287(9), 1132–1141 (2002). [CrossRef] [PubMed]
4. R. D. Brook, S. Rajagopalan, C. A. Pope 3rd, J. R. Brook, A. Bhatnagar, A. V. Diez-Roux, F. Holguin, Y. Hong, R. V. Luepker, M. A. Mittleman, A. Peters, D. Siscovick, S. C. Smith Jr., L. Whitsel, J. D. Kaufman, and American Heart Association Council on Epidemiology and Prevention, Council on the Kidney in Cardiovascular Disease, and Council on Nutrition, Physical Activity and Metabolism, “Particulate matter air pollution and cardiovascular disease: An update to the scientific statement from the American Heart Association,” Circulation 121(21), 2331–2378 (2010). [CrossRef] [PubMed]
5. J. Lelieveld, J. S. Evans, M. Fnais, D. Giannadaki, and A. Pozzer, “The contribution of outdoor air pollution sources to premature mortality on a global scale,” Nature 525(7569), 367–371 (2015). [CrossRef] [PubMed]
6. Y. F. Xing, Y. H. Xu, M. H. Shi, and Y. X. Lian, “The impact of PM2.5 on the human respiratory system,” J. Thorac. Dis. 8(1), E69–E74 (2016). [PubMed]
7. Y. Hu, J. Fan, H. Zhang, X. Chen, and G. Dai, “An estimated method of urban PM2.5 concentration distribution for a mobile sensing system,” Pervasive Mobile Comput. 25, 88–103 (2016). [CrossRef]
8. M. Alvarado, F. Gonzalez, A. Fletcher, and A. Doshi, “Towards the development of a low cost airborne sensing system to monitor dust particles after blasting at open-pit mine sites,” Sensors (Basel) 15(8), 19667–19687 (2015). [CrossRef] [PubMed]
9. L. J. Chen, Y. H. Ho, H. H. Hsieh, S. T. Huang, H. C. Lee, and S. Mahajan, “ADF: An anomaly detection framework for large-scale PM2.5 sensing systems,” IEEE Internet Things J. 5(2), 559–570 (2018). [CrossRef]
10. Columbia Tribune, We spend too much of our lives indoors. [Online] Available: http://www.coumbiatribune.com/arts_life/community/we-spend-too-much-of-out-lives-indoors/article_8aee8294%E2%80%9308fa-11e3%E2%80%939c55%E2%80%9310604b9f6eda.html.
11. Forbes. How our buildings are making us sick. [Online] Available: https://www.forbes.com/sites/amywestervelt/2012/08/08/how-our-buildings-are-making-us-sick/.
12. Y. Han, M. Qi, Y. Chen, H. Shen, J. Liu, Y. Huang, H. Chen, W. Liu, X. Wang, J. Liu, B. Xing, and S. Tao, “Influences of ambient air PM2.5 and meteorological condition on the indoor PM2.5 concentrations,” Environ. Pollut. 205, 307–314 (2015). [CrossRef] [PubMed]
13. A. Jovicic, J. Li, and T. Richardson, “Visible light communication: Opportunities, challenges and the path to market,” IEEE Commun. Mag. 51(12), 26–32 (2013). [CrossRef]
14. T. Yamazato, I. Takai, H. Okada, T. Fujii, T. Yendo, S. Arai, M. Andoh, T. Harada, K. Yasutomi, K. Kagawa, and S. Kawahito, “Image-sensor-based visible light communication for automotive applications,” IEEE Commun. Mag. 52(7), 88–97 (2014). [CrossRef]
15. H. S. Kim, D. R. Kim, S. H. Yang, Y. H. Son, and S. K. Han, “An indoor visible light communication positioning system using a RF carrier allocation technique,” J. Lightwave Technol. 31(1), 134–144 (2013). [CrossRef]
16. J. An and W. Y. Chung, “Wavelength-division multiplexing optical transmission for EMI-free indoor fine particulate matter monitoring,” IEEE Access 6, 74885–74894 (2018). [CrossRef]
17. J. An and W. Y. Chung, “Single-LED multichannel optical transmission with SCMA for indoor healthcare,” J. Lightwave Technol. 36(23), 107–113 (2017).
18. J. An, Q. N. Pham, and W. Y. Chung, “Single cell three-channel wavelength division multiplexing in visible light communication,” Opt. Express 25(21), 25477–25485 (2017). [CrossRef] [PubMed]
19. J. Armstrong and B. J. C. Schmidt, “Comparison of asymmetrically clipped optical OFDM and DC-biased optical OFDM in AWGN,” IEEE Commun. Lett. 12(5), 343–345 (2008). [CrossRef]
20. S. Devasmitha and J. Armstrong, “Comparison of ACO-OFDM, DCO-OFDM and ADO-OFDM in IM/DD systems,” J. Lightwave Technol. 31(7), 1063–1072 (2013). [CrossRef]
21. J. Zhou, Y. Qiao, Z. Cai, and Y. Ji, “Asymmetrically clipped optical fast OFDM based on discrete cosine transform for IM/DD systems,” J. Lightwave Technol. 33(9), 1920–1927 (2015). [CrossRef]
22. D. Astharini, K. Azrina, D. Gunawan, D. I. Savitri, and O. N. Samijayani, “Design of visible light communication with DCT and M-ary PAM in Xilinx system generator,” in International Conference on Engineering Technologies and Applied Sciences (ICETAS),20185th International Conference On (IEEE 2018), pp. 1–6. [CrossRef]
23. L. J. Chen, Y. H. Ho, H. C. Lee, H. C. Wu, H. M. Liu, H. H. Hsieh, Y. T. Huang, and S. C. C. Lung, “An open framework for participatory PM2.5 monitoring in smart cities,” IEEE Access 5, 14441–14454 (2017). [CrossRef]
24. Air Korea, Air quality monitoring. [Online] Available: http://www.airkorea.or.kr/web
25. A. Ritter and R. M. Carpena, “Performance evaluation of hydrological models: Statistical significance for reducing subjectivity in goodness-of-fit assessments,” J. Hydrol. (Amst.) 480, 33–45 (2013). [CrossRef]
26. Q. N. Pham, V. P. Rachim, J. Y. An, and W. Y. Chung, “Ambient light rejection using a novel average voltage tracking in visible light communication system,” Appl. Sci. (Basel) 7(7), 670 (2017). [CrossRef]
27. D. Tsonev, H. Chun, S. Rajbhandari, J. McKendry, S. Videv, E. Gu, M. Haji, S. Watson, A. E. Kelly, G. Faulkner, M. D. Dawson, H. Haas, and D. O’Brien, “A 3-Gb/s Single-LED OFDM-based wireless link using a gallium nitride μLED,” IEEE Photonics Technol. Lett. 26(7), 637–640 (2014). [CrossRef]
28. D. Mintz, “Technical assistance document for the reporting of daily air quality – the air quality index (AQI),” U.S. Environmental Protection Agency, Dec. 2016.
29. S. Elshout, K. Leger, F. Nussio, M. Cagnoil, D. Gombert, S. Anselmi, E. Donato, H. Heich, C. Lad, M. Luts, and A. Snijder, “Air quality index - comparing urban air quality across borders,” Eur. Union (2012).
30. U. K. Air, Daily Air Quality Index. [Online] Available: https://uk-air.defra.gov.uk/air-pollution/daqi.
31. N. Q. Pham, V. P. Rachim, and W. Y. Chung, “EMI-free bidirectional real-time indoor environment monitoring system,” IEEE Access 7, 5714–5722 (2019). [CrossRef]
32. J. Song, W. Ding, F. Yang, H. Yang, B. Yu, and H. Zhang, “An indoor broadband broadcasting system based on PLC and VLC,” IEEE Trans. Broadcast 61(2), 299–308 (2015). [CrossRef]
33. W. Ding, F. Yang, H. Yang, J. Wang, X. Wang, X. Zhang, and J. Song, “A hybrid power line and visible light communication system for indoor hospital applications,” Comput. Ind. 68, 170–178 (2015). [CrossRef]
