240

Int'l Conf. Wireless Networks | ICWN'15 |

Energy-efficient Communication Algorithm Using Luminance Control of Ceiling Lighting for Wireless Sensor Networks Hiroki MURAKAMI1 , Hiroto AIDA2 , Motoi OKADA1 , Kento MATSUI1 and Mitsunori MIKI2 1 Graduate School of Science and Engineering, Doshisha University, Kyoto, Japan 2 Department of Science and Engineering, Doshisha University, Kyoto, Japan Abstract— A wireless sensor network enables monitoring in a wide area by distributing many sensor nodes. The typical method of wireless communication in sensor networks is flooding. In flooding, each node broadcasts data. This method requires each node to transmit data many times and entails such problems as low energy efficiency and packet collisions. An alternative method of indoor communication is visible light communication; however, visible light communication has a low cost-effectiveness for a sensor network which handles small-size data, considering the requirement of devices such as a modulator. Hence, the authors propose a data communication method which does not use wireless communication but uses commonly available lightings and illuminance sensors. This study examines a communication scheme which transmits data to illuminance sensors by causing continuous variations in illuminance within a range not sensible by human eyes in an environment free of external light, and methods to increase the communication speed. Keywords: sensor networks, sensor nodes, lights

been many research endeavors to develop more efficient methods of communication between sensor nodes based on flooding [2][3][4]．However, as long as using wireless communication, the possibilities of change in the communication environment from obstacles or node arrangement as well as packet losses, cannot be neglected. Hence, we propose a communication method not relying on wireless communication, to alleviate communication loads on a sensor network and increase the energy efficiency of sensor nodes. This study proposes a data communication method which does not rely on wireless communication, but utilizes the luminance control features of commonly available dimmable ceiling lights and luminance sensor nodes. The proposed method transmits data to the sensor nodes within the radiation of a lighting fixture all at once, eliminating the need of broadcasting and realizing high energy efficiency. We will also examine a communication algorithm without regard to effects of external factors other than the luminance of lighting fixtures such as daylight, and approaches to considering a higher communication speed.

1. Introduction

2. Related work

In recent years, wireless sensor networks have drawn attention as a technology to control office or home appliances, control crop production facilities in agriculture, traffic monitoring or natural disaster monitoring [1]．A wireless sensor network enables wide area monitoring by placing many sensor nodes in the space which transmit sensing data. Each sensor node has such capabilities as sensing the temperature/humidity status of the target under monitoring, sending and receiving the sensing data and arithmetic processing. Meanwhile, being wireless means that wireless sensor nodes have limited power supply. Hence, to prolong the life and minimize the operating cost of a wireless sensor network, a high level of energy efficiency is needed in sensor nodes control. In sensor networks, flooding is the most typical method of wireless communication for transmitting the status information from sensors. While flooding is the simplest and strongest method of communication between nodes, all sensor nodes comprising the network are subject to communication loads because all sensor nodes broadcast status information. In a sensor network composed of sensor nodes with restrained power supply, flooding causes heavy communication loads. To solve these problems, there have

2.1 Visible light communication In recent years, the advancement of communication technologies has made wireless communication ubiquitously available. Meanwhile, for the users of wireless communication, there are no means to know from where data are sent or where they are going. To solve this problem in wireless communication using radio waves, research in visible light communication has been underway [5]．Visible light communication is a communication method which uses lighting for data communication. Visible light communication is considered to have the following three advantages: first, users can visually know that a transmitter is sending data. The sender can recognize by light that the data are being sent while the receiver can know the sender by recognizing the light source. Secondly, it enables secure communication. Since the communication is visible, just physically blocking light is enough to realize secure communication. Thirdly, visible light communication using the light from lighting fixtures can be used safely in areas where the use of radio waves is restricted, such as hospitals and airplanes. Conventional methods of visible light communication use

Int'l Conf. Wireless Networks | ICWN'15 |

visible light, which is an electromagnetic radiation of 0.4 -0.7 micrometers in wavelength to communicate information. By flashing an LED at a speed too high for human eyes to recognize, it communicates information to terminals equipped with a receiving device. The receiving device is typically a high-resolution photo diode or a high-speed image sensor. Hence, it is difficult to apply these methods to an existing network consisting of illuminance sensors. This study intends to realize a communication method applicable to existing common sensor networks by realizing data communication using illuminance sensors.

2.2 Flooding Flooding is a method commonly used for transmitting status information obtained by sensors. In flooding, since packet communication to the whole sensor network is enabled by all sensor nodes broadcasting status information, all sensor nodes comprising the network are subject to communication loads. Meanwhile, the power consumption by packet communication shares a large part of the whole power consumption by sensor nodes. Hence, for a sensor network with restraints on power supply, energy efficient control of packet communication is essential. Against this backdrop, researches have been undertaken on ways to increase energy efficiency by improving the methods of route search in flooding [6]．However, power is nonetheless consumed as long as the communication uses a wireless technology. Therefore, we propose a communication method which does not use wireless communication to reduce communication loads on the sensor network to increase the energy efficiency of sensor nodes. In this study, the authors propose a data communication method which uses no wireless communication but uses the luminance control of commonly available dimmable ceiling lighting fixtures and illuminance sensor nodes. In data communication using luminance control, data transmission from the sink node to sensor nodes will be possible: for example, this may be used for sending commands to change the sampling frequency to each sensor node or to operate the power supply. When the luminance of a lighting fixture is altered, the illuminance measured by sensor nodes will change; then each sensor node refers to the luminance trend, and interprets and receives it as a transmitted data. This method, sending data to all sensor nodes within the area of radiation of a lighting fixture at once, eliminates broadcasting to increase energy efficiency.

3. Scheme of data communication using ceiling lighting luminance control

241

nodes will change, and sensor nodes calculate luminance changes, which realizes data communication. A possible problem from illuminance changes is that user comfort may be sacrificed if illuminances changes are sensed by users. Hence, the illuminance changes need to be within the range unnoticeable to users. Preceding studies have verified that illuminance changes within 7% from the current value cannot be sensed by human eyes [7]．Hence, assuming an indoor environment, the data communication scheme using ceiling lighting luminance control causes illuminance changes within 7% of the current illuminance to realize data communication. In order to propose a data communication algorithm using ceiling lighting luminance control, the authors investigated the trends of illuminance measured by sensor nodes at lighting luminance changes.

3.2 Illuminance trend experiment using wireless sensor nodes An experiment was conducted to verify how the illuminance values measured by an illuminance sensor mounted on a wireless sensor node change when changes within about 7% of the current illuminance were given. For the experiment, Crossbow’s MOTE MICAz was used as wireless sensor nodes [8]．On a MOTE MICAz node, a generalpurpose external sensor board MDA088 was installed and a lead-type NaPiCa [9] illuminance sensor was mounted for obtaining illuminance values. The resistance between the MDA088 and the NaPiCa illuminance sensor here is 430 Ω. The experiment was conducted in the laboratory at Doshisha University, using 28 sets of a fullcolor LED lighting (DLA016E, SHARP Corporation), a wireless sensor node with NaPiCa and a sink node. The illuminance measurement interval for the illuminance sensor was set to 0.1 second. Fig. 1 shows the plan of the experimental environment. Fig. 2 shows a photo of a scene from the experiment. The distance between a lighting and a wireless node placed perpendicularly below the lighting was 1.9m. As shown in Fig. 1, the wireless sensor node was placed right below the lighting.

Sensor Node Full-color LED

3.1 Overview In the data communication scheme using ceiling lighting luminance control, the system alters the luminance of lighting fixtures so that the illuminance measured by sensor

Figure 1: Environment for the illuminance measurement experiment

Int'l Conf. Wireless Networks | ICWN'15 |



  

242







Figure 2: A scene from the illuminance measurement experiment It should be noted here that the values obtained by a NaPiCa illuminance sensor do not directly show the illuminance values themselves. Besides, this experiment does not require accurate illuminance values because it is intended to enable the reception of data from a lighting by having a sensor node comparing illuminance values on relative terms. Hence, for the purpose of this experiment, the values obtained by a NaPiCa were corrected against a generalpurpose class A, ANA-F11 illuminance meter; using an ANA-F11 illuminance meter capable of measuring illuminance with high precision. The illuminance values obtained from NaPiCa were corrected in this experiment. Equation 1 below is the correction formula. Illana ≈ 2.096 ∗ Illnapica + 17

(1)

Illana : illuminance obtained by sensor ANA-F11 [lx] Illnapica : value obtained by NaPiCa illuminace sensor In an environment where the original desktop illuminance is 500 lx, the illuminance was raised by 15 lx, 3% of 500 lx, and then brought down to the original value in 1 second, while the taking illuminance data. The experiment was conducted in an environment in which the illuminance sensor is free from effects of external light such as daylight. Fig. 3 shows the history of illuminance measured at 0.1 second intervals. As is indicated by Fig. 3, the illuminance changes sharply when a luminance change is detected. In addition, there constantly are small fluctuations in illuminance, which need to be recognized as no change in illuminance. Based on the experiment, a data communication algorithm using ceiling lighting luminance control is proposed.

3.3 Data communication algorithm using ceiling lighting luminance control Since the illuminance measurement trend experiment using wireless sensor nodes demonstrated the existence of constant illuminance fluctuations, it became clear that there is a need of an algorithm which neglects these fluctuations

Figure 3: History of illuminance measured at 0.1 second intervals

but regards only large illuminance changes as significant illuminance changes. An algorithm of a data communication scheme using ceiling lighting luminance control satisfying this condition is described below. The proposed algorithm assumes an environment in which illuminance sensors are free from the effects of light other than from the lightings such as daylight. From Fig. 3, when the sensor detects a luminance change, the illuminance value changes sharply, while some errors in illuminance measurements occur even while the lighting is kept on at a constant luminance, causing fluctuations in illuminance values. Hence, it needs to be ensured that sensors recognize only those illuminance changes occurring upon a luminance change as illuminance changes caused by data communication. For this purpose, in the proposed method, the measured illuminance value is differentiated from the previous illuminance value, and the algorithm judges based on the gradient whether it is an illuminance change caused by a luminance change or not. The following paragraphs describe the algorithm more specifically. The proposed algorithm is premised on the condition that the data transmission interval is already known to sensor nodes. The data transmission interval here is T[s]. The illuminance sensor obtains illuminance values n times within T[s], and retains the current illuminance value and the previous illuminance value. Then the gradient of illuminance change is calculated using the current illuminance value and the previous illuminance value: if it is not below the threshold α, then it is deemed an illuminance change from a luminance change. Meanwhile, the lighting converts the data to transmit into a binary bit sequence and sends it bit by bit. For the purpose of this method, the bit is "0" if the gradient is below the threshold α or "1" if it is not below the threshold α. To indicate the start of communication to the sensor node, "1" is sent as a start bit. When bits of the same values are sent serially, the lighting is kept on at the same luminance, resulting in no change in illuminance. Hence, it determines whether 1 or 0 at the point of T seconds from the

Int'l Conf. Wireless Networks | ICWN'15 |

4. Increasing the speed of data communication using ceiling lighting luminance control 4.1 Outline of the strategy for speedup The speed of communication using the data communication algorithm based on ceiling lighting luminance control depends on the bit transmission interval (T seconds) and the notation of the bits transmitted at T second intervals. Hence, this chapter examines approaches toward increasing the speed of data communication when a data communication algorithm based on ceiling lighting luminance control is used. For the purpose of this study, two approaches to speedup are verified: shortening the transmission interval and sending more information with one luminance control. In the method using a shorter transmission interval, the interval T was shortened from 1 second by 0.1 second in an attempt to realize a higher speed in binary data communication. In the method transmitting more data at a time, illuminance levels graded in multiple stages as shown in Fig. 4 were used to increase the amount of data transmitted by one step of luminance control to realize a higher speed.

4.2 Experiment of higher speed communication using shorter transmission intervals A data communication experiment was conducted using the data communication algorithm based on ceiling lighting luminance control, varying the transmission interval

  
 



















   



Figure 4: Speedup by graded illuminance changes

(T second) between 1.0 second and 0.1 second by 0.1 second increments. The experimental environment was the same as that used in the illuminance trend experiment. The sample data used in the data communication are 100 bits of randomly generated binary data. Fig. 5 shows the resulting average matching rates. From Fig. 5, with transmission intervals between 1.0 second and 0.2 second, the communication is successful with a matching rate of 99% or more. However, at an interval of 0.1 second, the matching rate is 52.5%, indicating that the communication was unsuccessful. Checking the illuminance measurement intervals in communication with a transmission interval of 0.1 second revealed that the interval between illuminance measurements exceeded 0.01 second, and judgment on one bit took 0.1 second or longer, which disabled successful communication. Because the average minimum interval between illuminance measurements for a NaPiCa sensor is 0.0084 seconds, it was considered that individual differences and application delays may have disabled illuminance measurement at a transmission interval of 0.1 second.  
 




reception of a start bit, using the current illuminance value. （1） The illuminance sensor obtains the current illuminance value （2） The lighting raises the luminance by x1 %（when x1 < 7%） from the current luminance （3） When the illuminance sensor detects a change in illuminance, the gradient between before and after the change is calculated. （4） If the gradient is within the threshold, data communication is started. （5） The lighting resets the luminance to the original luminance value. （6） The gradient is calculated at every illuminance measurement interval. Then the system waits for T/2 seconds. （7） Upon detecting a gradient, the system calculates the amount of change from the maximum illuminance and minimum illuminance in the last T seconds, and determines received bit against the preconfigured threshold. （8） If the gradient remains below the threshold for T seconds after bit reception and the illuminance value is in the original range, it is deemed 0. This algorithm enables data communication using changes in illuminance obtained by a sensor node as a communication medium, realized by changing the lighting luminance.

243


















  
 



Figure 5: Average matching rates in data communication experiment at different transmission intervals These results show that the communication speed can be increased by using shorter transmission intervals, of which limit depends on the minimum measurement interval of a NaPiCa illuminance sensor. Also where an illuminance

244

Int'l Conf. Wireless Networks | ICWN'15 |

4.3 Experiment of higher speed communication using graded illuminance changes In this experiment, by converting every 2 bits of the binary sequence into a quaternary equivalent to make a quaternary bit sequence, the luminance controlling side transmits more data with each luminance control, to realize communication at a higher speed. The luminance controlling side transmits bits using luminance levels graded in four stages as shown in Fig .4. The sensor node calculates the illuminance gradient after each measurement interval, and then identifies it with one of the four grades referring to three thresholds as shown in Fig.4. The gradient is judged to mean 0 if illuminance remains unchanged from the reference illuminance level L [lx] for T seconds; 1, 2 or 3 when it has changed from L [lx] to L+a, L+2a or L+3a respectively. Fig. 4 shows an example in which illuminance L is set at 500lx. An experiment was conducted with a configuration to discriminate received bits as shown in Fig. 6. In the experiment, every 2 bits of the 100 bits of a binary bit sequence were converted into a quaternary equivalent to make a quaternary bit sequence, which were transmitted at 1.0 second intervals. The experimental environment was the same as the one used for the illuminance trend experiment using wireless sensor nodes. The data used in communication were randomly generated 100 binary bits. Fig. 7 shows a chart of average matching rates in the cases of two-stage grading and fourstage grading at a transmission interval of 1.0 second. Fig. 7 reveals that communication in four-stage grading was as successful as communication in two-stage grading.

5. Conclusion This study examined a data communication sheme using a MOTE MICAz wireless sensor node, a NaPiCa illuminance sensor and the luminance control of dimmable ceiling lightings as well as approaches to increasing the communication speed. First, to realize a data communication scheme based on luminance control, the trend of luminance values obtained by a NaPiCa luminance sensor was investigated. The luminance values obtained by the NaPiCa sensor showed some fluctuation even while the lighting luminance was constant. Hence, sensor nodes need to disregard the fluctuation in illuminance value occurring under a constant luminance, but to recognize an illuminance change only when the illuminance trend indicates that it is from a luminance change. An algorithm was developed in which a sensor node calculates the

   

  

 









  

 













Figure 6: Illuminance values obtained at a sensor node and received bits

 
 
 


sensor other than NaPiCa is used, the communication speed can be optimized by appropriately setting the transmission interval T. In the case of NaPiCa, since the average minimum measurement interval is 0.0084 seconds, it is considered that the limit of speedup may lie somewhere around 0.0084 seconds; but because a higher speed means a higher error rate, the transmission interval needs to be selected appropriately for the purpose of communication.









 



  





 



  


Figure 7: Average matching rates with two-stage luminance grading change and four-stage luminance grading change

gradient by differentiating the illuminance value obtained, and determines whether there was an illuminance change by comparing the gradient against a preconfigured threshold. As a method to increase the speed of data communication based on ceiling light luminance control using this algorithm, two approaches were proposed – using shorter transmission intervals and transmitting more data at once – for which data communication experiments were conducted. The experiment using shorter transmission intervals demonstrated that the interval can be 0.2 seconds at the shortest. The transmission speed (T seconds) depends on the illuminance sensor’s performance rating on the minimum illuminance measurement interval. This means that even when using an illuminance sensor other than NaPiCa, the communication speed can be increased by appropriately configuring the transmission interval (T seconds) according to the sensor’s minimum illuminance measurement interval. In the experiment of improving the communication speed by graded illuminance changes, the authors tried to improve the communication speed by increasing the amount of information sent at once by grading the illuminance in multiple

Int'l Conf. Wireless Networks | ICWN'15 |

stages. The result demonstrated that communication using illuminance changes graded in multiple stages was as successful as the high-speed communication at a transmission interval of 1.0 second. The future themes for study will include developing communication methods which can cope with changing communication environments. Although the experiments for this study were conducted in an ideal environment free from the effects of daylight, illuminance sensors may be affected by a workers’ shadow or display lights: hence, methods with considerations for these influences need to be developed. A disturbance such as a worker’s shadow on an illuminance sensor is expected to cause a sharp change in illuminance measurements. A possible solution to his problem is to configure sensor nodes so that they will disregard such sharp changes, judging that no change in illuminance has occurred. Incorporating a process to cope with disturbances like this is expected to enable data communication based on ceiling lighting luminance control taking account of disturbances.

References [1] I. F. Akyildiz and W. Su, Y. Sankarasubramaniam, "Wireless sensor networks: a survey”, Computer Networks Journal, vol.38, no.4, pp.393422, 2002. [2] J. Nagashima, A. Utani and H. Ymamoto, "Advanced Particle Swarm Optimization Algorithm Computing Plural Acceptable Solutions and Its Application to Wireless Sensor Networks -Forwarding Power Adjustment of Each Sensor Node for Query Dissemination-", Journal of Japan Society for Fuzzy Theory and Intelligent Informatics, vol.23, no.1 pp.65-77, 2002. [3] Y. Ohta, K. Yanagihara and S. Hara, "A Clustering Method Taking into Consideration of Wireless Ling Characteristic in Wireless Sensor Network", The IEICE Transactions on Information and Systems Technical Report.USN, no.107 pp.85-90, 2007. [4] K. Suzuki, B. Mandai and T. Watanabe, "Dispersive Packets Transmission to Multiple Sinks for Prolonging Network Lifetime in Sensor Networks", The IEICE Transactions on Information and Systems.B, vol.J91-B, no.8, pp.831-843, 2008. [5] S. Haruyama, "Visible Light Communication", The IEICE Transactions on Information and Systems.A, vol.86, no.12, pp.1284-1291, 2003. [6] M. Noto, J. Arikawa and M.Matsuda, "Low-Power Flooding Method in Ad-Hoc Networks", The IEICE Transactions on Information and Systems.D, vol.J91, no.5, pp.1252-1260, 2008. [7] T. Shikakura, H. Morikawa and Y. Nakamura, "Research on the Perception of Lighting Fluctuation in a Luminous Offices Environment", Journal of the Illuminating Engineering Institute of Japan., vol.85, pp.346-351, 2001. [8] Crossbow MOTE - Wireless Sensor Networks MTS/MDA Sensor Board User’s Manual, http://www.xbow.jp/mtsmdaj.pdf. [9] Panasonic: NaPiCa, http://www3.panasonic.biz/ac/download/control/ sensor/illuminance/catalog/bltn_jpn_ams.pdf.

245