742 IEEE SENSORS JOURNAL, VOL. 15, NO. 2, FEBRUARY 2015

A Low-Power Wireless Sensor forOnline Ambient Monitoring

Silviu C. Folea, Member, IEEE, and George Mois, Member, IEEE

Abstract— This paper presents the development of a compactbattery-powered system that monitors the carbon dioxide level,temperature, relative humidity, absolute pressure, and intensity oflight in indoor spaces, and that sends the measurement data usingthe existent wireless infrastructure based on the IEEE 802.11 b/gstandards. The resulted device’s characteristics and performanceare comparable with the ones provided by recognized solutions,such as ZigBee-based sensor nodes. By combining Wi-Fi con-nectivity with ambient sensors, this solution can be used for theremote gathering and further processing of measurement data.Testing revealed that the system can operate continuously for upto three years on a single 3 V small battery.

Index Terms— Sensor systems, wireless sensor networks, recon-figurable architectures, Internet.

I. INTRODUCTION

INDOOR air quality (IAQ) represents an important factoraffecting the comfort, the health and also the safety of

building occupants. IAQ problems lead to a set of symptoms,including headaches, dizziness, difficulties in concentrationand others, referred to as “sick building syndrome” (SBS).Basic measurements, such as temperature, relative humidityand CO2, can provide information useful in solving suchproblems [1]. The present paper presents the development ofa compact battery-powered system, that monitors the temper-ature, relative humidity, the carbon dioxide level, the absolutepressure and the intensity of light in indoor spaces, andthat sends the measurement data using the existent wirelessinfrastructure based on the IEEE 802.11 b/g standards. Thisprovides the possibility of the remote gathering and furtherprocessing of data from a large number of such wireless sens-ing systems. Furthermore, by combining wireless connectivitywith ambient sensors, this solution can be used for reducingthe overall energy consumption of an entire building [2].

The characteristics of the developed device, namely reduceddimensions, low power consumption, high flexibility androbustness, make it suitable for its use as a node in a wirelesssensor network (WSN) or in an Internet of Things (IoT)scenario. The reduced energy profile is achieved by the useof a low power core microcontroller, and of an nondispersive

Manuscript received June 20, 2014; accepted August 12, 2014. Date ofpublication August 22, 2014; date of current version November 20, 2014.The associate editor coordinating the review of this paper and approving itfor publication was Dr. M. R. Yuce.

The authors are with the Department of Automation, Faculty ofAutomation and Computer Science, Technical University of Cluj-Napoca,Cluj-Napoca 400114, Romania (e-mail: silviu.folea@aut.utcluj.ro;george.mois@aut.utcluj.ro).

Color versions of one or more of the figures in this paper are availableonline at http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/JSEN.2014.2351420

infrared sensor (NDIR) for CO2 measurements, having thelowest power consumption on the market. The temperatureand relative humidity sensor has a power consumption that iscomparable to the one of the gas sensor (1 mA), while theother attached sensors, measuring pressure and light intensity,are less power hungry than these, consuming 5 µA and0.24 mA, respectively. Moreover, a Wi-Fi module with anadvanced API software, named WiFly, which allows efficientpower management, was chosen for data transmission. These,combined with suitable power saving strategies, and dependingon preset measurement rates, lead to the achievement of abattery life between one month and several years. Although theacquired ambient data can be displayed locally on the attachedLCD, for testing the most probable usage scenario, they werevisualised using a commercial solution, provided by Xively,a “Public Cloud for the Internet of Things” [3].

The use of gas sensors in general, and of CO2 sensors,in particular, in small battery powered devices was not possibleuntil recently because of their large power consumption anddimensions. The low power NDIR sensors field is at thebeginning, with the Cozir® Ambient CO2 sensor dominatingthe market. The solution presented in this paper employs thissensor, achieving satisfactory accuracy and battery lifetime.The other attached sensors, namely the ones measuring tem-perature and relative humidity, pressure and light intensity, donot pose as many problems as the CO2 sensor and can beefficiently included in a portable device. As far as the authorsknow, there are no other devices with the same performanceand features.

Several solutions are presented in the literature or arepresent on the market, but they provide a limited set offunctionalities and a reduced number of attached sensors.A similar device represents the subject of paper [4], butthe power consumption here is not calculated. Based on theanalysis of the used chips and on the general description, thedesign in this case cannot achieve low power consumptionand, therefore, the device cannot be a mobile one. Anothersolution, this time closer to the one presented here, comesfrom the company Point SixTM, and employs an NDIR sensorand Wi-Fi connectivity [5]. However, this system does notinclude humidity, atmospheric pressure and light measurementcapabilities. The third monitoring system is a self-powered oneand is developed by EnOcean Alliance [6]. It is also based onthe Cozir® sensor, but it allows only a reduced number ofmeasurement rates, between 9 and 2 per hour, depending onlight intensity. The device presented in this paper is a smallbattery powered ambient (temperature, relative humidity, CO2,absolute pressure and light intensity) wireless sensor allowing

1530-437X © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

FOLEA AND MOIS: LOW-POWER WIRELESS SENSOR FOR ONLINE AMBIENT MONITORING 743

Fig. 1. Ambient sensor front view.

measurement rates between one and 60 samples per hour.When taking a single measurement per hour, it can run forup to three years without requiring maintenance. The testsshowed a battery lifetime of up to three years, comparablewith the one of the device presented in paper [7], based on802.15.4/ZigBee communication, which consumes less energythan Wi-Fi.

The rest of the paper is structured as follows. Section IIpresents the hardware architecture of the developed wirelesssensor, highlighting the system components and their features.Section III details the software architecture of the applicationrunning on the device, along with several usage scenarios. Thespecification of message components and some examples ofsent values are also given here. Section IV presents the powerconsumption of the entire system, along with scenarios andmechanisms for reducing it, while the final section gives theconcluding remarks.

II. HARDWARE ARCHITECTURE

A. General Overview

The developed ambient wireless sensor, shown in Figure 1,is a stand-alone device, which measures the CO2 level inthe air, the temperature, humidity, absolute pressure and lightintensity and which sends the acquired information using theIEEE 802.11 b/g standards to a preset IP address. The acquireddata can be displayed locally on an LCD with backlight,showing text on 2 rows and 16 columns, by pressing a button,or at a remote location, by using a specialized application or aweb page. The Internet of Things scenario offers the possibilityof remotely visualizing numerical and graphical values overtime, setting triggers, sending short text messages using theShort Message Service (SMS) in case of alarms triggering andso on. All these alarms and triggers are implemented in thesoftware running on a remote computer used for data gatheringand visualization.

B. Internal Structure

The device’s core is represented by a PSoC 3, a program-mable system on chip microcontroller. This is the central partof the ambient sensor, initiating all the main actions that haveto be performed for its proper operation. The componentsthat make up the system can be divided, in the same way as

Fig. 2. Wireless sensor hardware architecture.

for a wireless sensor node in a wireless sensor network, intofour main groups: the sensing unit, the processing and storageunit, the transceiver and the power supply. The sensing unitconsists of a CozirTM CO2 Ambient Sensor, a DHT22 digitaltemperature and humidity sensor, an MPL115A2 barometersensor and a TSL2561 light sensor. The sensors were chosento respect satisfactory range and accuracy requirements, whileachieving the smallest power consumption. Another importantcriterion was represented by the cost of the sensors, a rea-sonable price being achieved for small quantities. This makesthe system very competitive and suitable for mass-production.The processing and storage unit is represented by the coremicrocontroller, while data transmission is implemented by theRN-131C/G wireless LAN module, from Roving Networks.A 3 V CR123A battery and a DC/DC converter form thepower supply unit. The architecture is presented in Figure 2,where the main components of the measurement system arehighlighted. As it can be seen in Figures 2 and 3, all thesensors are attached to the core microcontroller and operate atthe same time.

C. The PSoC 3 Core

The advances in semiconductor industry, the smaller processtechnologies and the maximized circuit densities lead to a con-tinuously increasing number of System-on-Chip solutions ina large number of applications. These circuits integrate signalacquisition and conversion functions, data storage and process-ing capabilities and I/O, providing significant advantages, themost important consisting in low power consumption, reduceddimensions and low costs. By including a wide range ofsystem components into the chip, the number of parts on theprinted circuit board (PCB) is reduced, directly affecting thepower consumption and production costs of the digital system.Such a device is the PSoC, the acronym for programmablesystem on chip, produced by Cypress Semiconductor [8].It integrates discrete analog and programmable logic alongwith memory and a microcontroller, being suitable for thedesign of embedded systems. These are the reasons whya PSoC 3 microcontroller, namely CY8C3246PVI-147, waschosen as the data processing unit of the wireless sensorpresented in this paper. It has an 8-bit single cycle pipelined8051 processor running at 24 MHz, as core, 64 kB of flashmemory, an 8 kB SRAM and an on-chip EEPROM forstoring nonvolatile data. This chip was chosen because it

744 IEEE SENSORS JOURNAL, VOL. 15, NO. 2, FEBRUARY 2015

offers enough memory for implementing the application. Theprogram occupies 62% of flash memory and 3.5% of SRAM.

D. Wireless Module

A standalone embedded wireless 802.11 b/g networkingmodule, the RN-131C/G Wireless LAN Module from RovingNetworks, was chosen for sending measurement data throughUDP to a specific IP address [9]. The transmission mechanismalong with the message format will be presented in Section III.The sensor’s Wi-Fi communication capability makes use ofthe existent wireless infrastructure and provides high transferrates even when encryption is employed (WPA2), but it alsolimits the battery lifetime because of the increased powerconsumption. This effect was countered by the use of severalmechanisms for reducing the power consumption of the entiresystem.

E. CO2, Temperature and Humidity Sensors

The ultra low power CozirTM CO2 Ambient Sensor, spe-cially developed for battery powered applications, was selectedfor measuring the carbon dioxide level in the air. It canmeasure CO2 concentrations between 0 and 2000 ppm. Itsaverage power consumption is less than 3.5 mW, but the powersupply of the measurement system must generate a peak of33 mA for a short period of time. The noise is higher than±50 ppm, but by activating the digital filter, so that an averagebetween 2 . . . 32 instant measurements is computed, its valuecan be attenuated. There is also a drawback to this action,the filter value affecting the warm-up period of the sensor,3 to 32 seconds being necessary for allowing the responseto reach a final value. This period has an impact over theoverall power consumption profile and the user must carefullychoose an appropriate filter value. The atmospheric pressureis one of the factors affecting CO2 monitoring and altitudecompensation is required. It can be set manually in the systemconfiguration step or it can be computed depending on theinformation given by the absolute pressure sensor.

Another important issue that has to be addressed is theauto-calibration routine because carbon-dioxide sensors arewidely used as part of demand-control ventilation (DCV)systems. Therefore, the performance of the CO2 sensors cansignificantly affect energy use as well as indoor air quality inthese cases. Overestimation of the CO2 concentration leads toincreased outdoor air usage and increased energy costs, whileunderestimation may lead to poor IAQ and SBS [10]. Fresh aircalibration is implemented by the core microcontroller, sincethe CO2 sensor is powered down after each reading. This isperformed at intervals specified by the user in the measurementsystem configuration phase, and can also be disabled if desired.The operating environment in which the system operates isof great importance also, because testing revealed that thecarbon dioxide sensing component is very sensitive to thedew point, where the digital output can decrease down tozero.

The temperature and relative humidity values are acquiredby a very low cost digital temperature and humidity sensor,the DHT22. It consists of several electronic components on

a small PCB, encased in a plastic box: a capacitive humiditysensor, a 10 kohms thermistor as the temperature transducerand a small package microcontroller, STM8S103F3, used forsignal processing. This sensor’s accuracy is acceptable in manyapplications with values of ±2 % (with a maximum of ±5 %)for humidity and of ±0.5 °C for temperature. The powerconsumption value is 1 mA in active mode and 40 µA insleep mode. This high value during sleep mode is one of thereasons for implementing a separate power supply for sensors,which can be switched off by the central processing unit. Thetemperature and humidity ranges are given by the CO2 sensor’sspecifications, its operation conditions allowing temperaturesbetween 0°C and 50°C and relative humidities between 0%and 95% (non-condensing).

F. Absolute Pressure and Light Sensors

An absolute pressure sensor, MPL115A2, with an I2Cinterface, was chosen for measuring the atmospheric pressureand for compensating the CO2 deviation, if required. Theinitial accuracy is of ±1 kPa, which translates to an errorof approximately 100 m in altitude. The absolute pressurerange of this sensor is between 50 kPa and 115 kPa. Thepower consumption is 5 µA in active mode and only 1 µA inshutdown mode.

The light sensor, TSL2561, which also communicatesthrough an I2C interface, was chosen for determining the lightintensity. In this case, the power consumption is 0.24 mA inactive mode and 3.2 µA in power down mode. The need forpolling the sensor can be removed by programming it withan interrupt function. The sensor outputs a digital value fromwhich illuminance, or the ambient light level, in lux is derivedusing an empirical formula to approximate the human eyeresponse.

G. Sensors Power Supply and Reverse Battery Protection

The power consumption in sleep mode for all the sensorsdoes not allow a long battery utilization period. This is whya separate power supply was developed and included in thedesign. The chip used offers an output disconnected from theinput, high efficiency while using small amounts of power, arange up to 140 mA at +3.3 V, from an 1.8 V input, anda current consumption in shutdown mode, which is lowerthan 1 µA. All these characteristics maximize the lifetimeof the battery in mobile applications. A CR123A 3 V lithiumbattery represents the main power supply. A reverse protectionis implemented for accomplishing safe operation even whenchanging the battery. This type of battery has a capacity of1500 mAh and is only slightly influenced by temperaturevariations and by loads [11].

H. PCB

The device’s PCB (Fig. 3) is double sided, all componentsbeing populated on the top layer; the bottom layer is usedonly for traces and for the ground plane. The componentsthat make up the user interface and that can be accessedby the users, namely the LCD, the buttons and the LED,

FOLEA AND MOIS: LOW-POWER WIRELESS SENSOR FOR ONLINE AMBIENT MONITORING 745

Fig. 3. Wireless sensor printed circuit board.

Fig. 4. Software application flow diagram.

are accessible from outside. Special attention was paid tocreating a PCB as compact as possible, ensuring that all ofthe sensors are exposed correctly and that ease of access forconnecting programming and configuration cables is provided.

III. SOFTWARE ARCHITECTURE

The block diagram of the main routines that make up thewireless sensor firmware is presented in Figure 4.

Its main component is represented by the main loop, whereall the sensors are powered up and read, and data messagesare sent using UDP. When it is first used, the wirelesssensor needs to be configured. This action is initiated bypowering up or resetting the device with the button on the

interface (Fig. 1) pressed. Configuration is performed throughthe serial interface, by using an RS232 cable. The menu allowsentering and displaying the parameters needed for the correctoperation of the measurement system. These consist of theperiod between measurements, which can be set to have avalue between one minute and 60 minutes; the informationfor connecting to wireless LANs, namely the channel used,the SSIDs and passwords; the data server information, whichincludes the server port, IP, gateway and the subnet mask; thenode IP and the CO2 sensor’s data. The latter is composedof filter value, altitude value and number of days for auto-calibration using fresh air, if this feature is activated. The nextstep is the configuration of the RN-131C/G wireless moduleby the core microcontroller using the data previously set andsaved in the EEPROM. Communication with this componentis performed serially, using the UART. WiFly commands thatmake the Wi-Fi module automatically connect to a specificaccess point and act as a pipe sending serial information overUDP, when reset, are sent. After these actions are completed,the period values are set in such a way that a first measurementis taken when entering the main application loop. The buttonon the interface has a single functionality here, namely thedisplay of the last values read by the attached sensors. Forminimizing the power consumption, the LCD, the sensors andthe Wi-Fi module are powered only when they need to performactions. After each measurement, the Wi-Fi module is wokenup, and specially formatted messages are sent to the previouslyset IP address.

The other important action performed inside the applicationmain loop is the CO2 sensor auto-calibration, taking placeat previously set time intervals (after days of continuousoperation), using fresh air. During this action, the sensor’s“fresh air” concentration value, considered to be 400 ppm,is replaced with the minimum recorded value, provided thefact that it had sensed fresh air at some point in time.

A. Usage Scenarios

Every ambiental sensor is associated to an AP and canmeasure the received signal strength indicator (RSSI) fordetermining if the network is proper for communication withlow energy costs. It is possible that from time to time thesensor scans the network, trying to determine whether an APis closer than the one to which it is associated. If the resultis favourable, the sensor will associate to a closer AP in caseit has the required security data stored in the EEPROM. Thisscenario also applies when the sensor tries to associate to anAP for a predetermined number of times and fails, leading tothe conclusion that the access point is no longer active

One of the most common applications employing wirelesssensors is represented by wireless sensor networks (WSNs).These consist of a large number of sensor nodes, commu-nicating in a wireless fashion among each other or to anexternal base-station [12]. The first field where they hadbeen used and where the potential is huge is represented byenvironmental monitoring, this being the primary purpose ofdeploying sensor networks [12]. Other domains include, butare not limited to military, health, home and other commercial

746 IEEE SENSORS JOURNAL, VOL. 15, NO. 2, FEBRUARY 2015

Fig. 5. Sensors in a wireless sensor network.

Fig. 6. Data visualization using web application.

applications [13], [14]. By satisfying the requirements foruse in WSNs, namely low cost, low power consumption,multifunctionality, small dimensions and wireless communica-tion capabilities, the sensor presented in this paper representsa wireless sensor node (Fig. 5).

Recently, a new generation of digital systems, called cyber-physical systems (CPSs) [15], emerged. These use a widerange of sensors for collecting information about the physicalworld and exploit the information collected by WSNs to bridgereal and cyber spaces [16]. Furthermore, the vision of Internetof Things calls for connectivity not only to consumer electron-ics and home appliances, but also to small battery powereddevices which cannot be recharged [17]. The device presentedin this paper can operate as an active component in CPSs orin the IoT. In this direction and for validating the proposedsolution, an application running on a personal computer wasdeveloped. This gets data from the device and sends them toa web application server for public display (Fig. 6).

The site presenting the measured data is www.xively.com,a “Public Cloud for the Internet of Things,” which displaysdata from sensors connected to the Internet from around theworld [3]. The software for displaying the data is unchained,residing on the Internet. The data from the sensors areprocessed by a LabVIEWTM application running on a PC or a

Fig. 7. Front panel and block diagram of the application implemented inLabVIEWTM.

server and are sent to the Xively web-site. For being correctlyinterpreted by the web-site, the data have to be bundledinto an EEML (Extended Environments Markup Language)script. The advantage of using an application running on aPC consists in the ability to read data from multiple devicesand to send a reduced number of packets to the Internetwithout performing a large number of accesses. For a low costsolution, the server and the application that runs on the servercan be omitted, including the enclosing of the data in an properformat into the sensor. This scenario has a major disadvantage,the use of the TCP/IP protocol, which leads to an increase inthe overall power consumption. An advantage of the solutionpresented in Figure 7 is the fact that data preprocessing takesplace in the application, the firmware being simplified, and theconnections with the Internet being reduced.

B. Data Transmission

The Wi-Fi standard was chosen for communication becausethe number of sensors used in the scenario of indoor envi-ronmental monitoring is not large and there is no need forcomplex routing protocols. The access point, or the router,covers, in this case, the entire area of the house and thewireless sensor nodes can associate and send messages directlyto it. Furthermore, environmental sensors do not have criticalreal time constraints which can be met only by protocolssuch as ISA100 or WirelessHART. The major advantage ofusing Wi-Fi technology consists in the use of the existinginfrastructure, which can be found in almost every home,where Internet connectivity or digital television is present.The major disadvantage lies in the increased power con-sumption, which directly influences the node lifetime. How-ever, as the next section will show, this drawback can beovercome.

The protocol chosen for data transmission is UDP, insteadof TCP/IP, offering lower package sizes, increased speeds, low

FOLEA AND MOIS: LOW-POWER WIRELESS SENSOR FOR ONLINE AMBIENT MONITORING 747

TABLE I

MESSAGE COMPONENTS

latency and connectionless communication. The messages con-sist of fixed-size numerical codes, which are called operationcodes or opcodes. These describe the settings of the sensor ormeasurement results. As it can be seen in Table I, the opcodesrepresent pairs of hexadecimal numbers associated to a specificfunction, with the first element being the function code, andthe second its associated value. For avoiding overhearingproblems, an unidirectional scheme was chosen, the deviceonly sending short messages separated by the specified inter-val [18]. Because household environmental information is notsensitive from the security and privacy points of view, standardWPA2 encryption is used. The possibility of changing thesecurity protocol, depending on the one used in the wirelesscomputer network to which the sensor node connects, is alsoavailable.

IV. POWER CONSUMPTION ESTIMATION

The entire system has a CR123A 3 V battery as the mainpower supply. This is the reason why several mechanisms forensuring the low-power operation of the device, were imple-mented. They lead to the achievement of a period betweenone and three years of operation using a single commercialoff-the-shelf battery. The configuration menu allows for valuesbetween one minute and 60 minutes to be set as the periodbetween two consecutive measurement and data transmissionactions. The system is a duty cycled one, spending mostof the time in sleep mode. The ratio between wakeup andsleep times can take values between 1:8 and 1:500. Thisalternation leads to a power consumption between one and twohundred microwatts. Wakeup time lasts for only a few seconds,depending on the value of the CO2 sensor’s digital filter,

Fig. 8. Complete wakeup periods.

Fig. 9. Detail of the normal measurement and data transmission.

period in which the consumed current lies between 14 mAand 26 mA (medium values), depending on the completedtasks: read data from the attached sensors (temperature andhumidity, CO2, pressure and light) or send data.

Three complete wakeup cycles are shown in Figure 8:1. boot, Wi-Fi and CO2 sensor setup (the other sensors donot require a setup action), measurement of temperature,humidity, CO2, pressure and light intensity, and data trans-mission; 2. awakening at the pressing of the user buttonand displaying data on LCD; and 3. awakening from thesleep period, measurement of the five physical quantities anddata transmission via Wi-Fi. The first actions are executed in28 seconds, the average current consumption being 19.89 mA,the second period is 10 seconds long, with an average currentof 14.01 mA, while the last one depends on the value of thedigital filter (for 10, wakeup time is 13 seconds), with a currentof 25.68 mA. For further reducing the power consumption,a DC/DC converter, which can be turned off, was used. Thisway, during sleep, when only the PSoC microcontroller and theDC/DC source are active, the entire system consumes 10 µA.Another choice motivated by the battery lifetime requirementsis represented by the unidirectional communication schemeand by the use of short opcodes.

The device only sends short messages between previouslyset time intervals, after which it goes to sleep mode.

Figure 9 presents a single wakeup period, consisting intemperature, humidity, CO2, pressure and light intensity

748 IEEE SENSORS JOURNAL, VOL. 15, NO. 2, FEBRUARY 2015

Fig. 10. Measurement setup diagram.

TABLE II

BATTERY LIFETIME

measurements, association to the access point and messagetransmission. In sleep mode, the power consumption is lowerthan 30 µW and for a complete cycle (measurement, trans-mission – association and transmission of results), the powerconsumption is between 42 and 78 mW.

These power consumption measurements were performedusing the setup presented in Figure 10, consisting of anINA138 circuit (current shunt monitor) and an integratingfunction for computing the consumed current during a wakeupperiod. The energy consumed during a complete cycle wascomputed and then, based on the sleep/wakeup ratio and on thetotal capacity of the battery, the node lifetime was estimated.

The operating time of a measurement system with wakeuptime of 7 seconds (the filter is set 2) and with varying sleepintervals is presented in Table II. Several devices were testedfor a measurement cycle of 1 minute, each of them performingaround 30,000 measurements and transmissions with the samesmall 3V battery. When using a measurement cycle of onehour (1 hour between two consecutive measurements andtransmissions), 30,000 cycles stretch over more than threeyears. The devices are currently tested for long operationperiods, by using a sleep time of 60 minutes. The experimentalsetup consisted of a system similar with the one presented inFigure 5, where several sensors send UDP messages, whichalso include the battery voltages, to a server posting the dataon www.xively.com. The results are comparable with the onespresented in [7], where the authors estimate a node lifetimeof almost three years for a smart gas monitoring system,organized as a IEEE 802.15.4/ZigBee network, in a cluster-tree configuration.

The battery chosen for being used by the device is aCR-123A, which is not a high performance one. However,it provides the advantage of reduced dimensions and can bebought at a relatively low price. If the design included atype C or D battery instead of the CR123A one, the sensornode’s lifetime would be doubled or even tripled with thecost of an increase in volume of the entire system. Thebattery voltage for a sensor node during testing is shown in

Fig. 11. Battery voltage over time.

Figure 11. It has the same characteristic as the one givenby the manufacturer for a given load. It can be replaced bya photovoltaic cell and a supercapacitor [19]. By using theinformation from the light intensity sensor and by periodicallychecking the voltage on the capacitor, the device can computethe right moment for data transmission. The device is designedin such a way that it can operate properly on a voltage startingfrom 2.0 V, making energy harvesting viable.

V. CONCLUSION

The development of a compact battery-powered system,that monitors the temperature, relative humidity, the carbondioxide level, the absolute pressure and the intensity of light inindoor spaces, and that sends the measurement data using theexistent wireless infrastructure based on the IEEE 802.11b/gstandards, was presented. Its power consumption was testedin a real environment, with a rate of one transmission perminute, indicating a battery lifetime close to one month.Further, tests and simulations revealed that the system canoperate continuously for up to three years without requestingbattery replacement. The device automatically self-calibratesthe attached CO2 sensor and offers the possibility of operationwithout maintenance for a long time. It can be used in a widerange of monitoring applications as a component in a WSN, inthe IoT or in a cyber-physical system. The replacement of thebattery with an accumulator, a photovoltaic cell and a chargingcircuit represents the subject of future work. By carefullyselecting the board components and sensors, a reasonableprice of the developed system was achieved even for smallquantities.

ACKNOWLEDGEMENT

The authors would like to thank Synchro CompS.R.L, Craiova, Romania, and especially Mr. Vio Biscu, forsupporting this research.

REFERENCES

[1] United States Environmental Protection Agency. (1991). “Indoor air factsno. 4 (revised) sick building syndrome,” Air and Radiation (6609J),Research and Development (MD-56), Tech. Rep. [Online]. Available:http://www.epa.gov/iaq/pdfs/sick_building_factsheet.pdf

[2] S. Sharma, V. N. Mishra, R. Dwivedi, and R. R. Das, “Quantification ofindividual gases/odors using dynamic responses of gas sensor array withASM feature technique,” IEEE Sensors J., vol. 14, no. 4, pp. 1006–1011,Apr. 2014.

[3] Xively. Xively Is the Public Cloud Specifically Built for the Internet ofThings. [Online]. Available: https://xively.com/whats_xively/

FOLEA AND MOIS: LOW-POWER WIRELESS SENSOR FOR ONLINE AMBIENT MONITORING 749

[4] H. Yang, Y. Qin, G. Feng, and H. Ci, “Online monitoring of geologicalCO2 storage and leakage based on wireless sensor networks,” IEEESensors J., vol. 13, no. 2, pp. 556–562, Feb. 2013.

[5] PointSix. WiFi 2000 ppm CO2 and Temperature Transmitter3008-40-V6. Point Six Wireless, Data Sheet. [Online]. Available:http://www.pointsix.com/PDFs/3008-40-V6.pdf

[6] Enocean Alliance. Self-Powered CO2 Sensor Moves Into VolumeProduction. [Online]. Available: http://www.enocean-alliance.org/en/gss-seamless-sensing-co2-sensor-moves-into-volume-production/

[7] V. Jelicic, M. Magno, D. Brunelli, G. Paci, and L. Benini, “Context-adaptive multimodal wireless sensor network for energy-efficient gasmonitoring,” IEEE Sensors J., vol. 13, no. 1, pp. 328–338, Jan. 2013.

[8] Cypress Semiconductor. (May 21, 2014). Programmable System-on-Chip (PSoC). PSoC®3: CY8C32 Family Data Sheet, document 001-56955. [Online]. Available: http://www.cypress.com/?docID=49257

[9] Roving Networks. (2012). RN-131G & RN-131C 802.11 b/g WirelessLAN Module. [Online]. Available: http://www.rovingnetworks.com

[10] S. S. Shrestha, “Performance evaluation of carbon-dioxide sensorsused in building HVAC applications,” Ph.D. dissertation, Dept. Mech.Eng., Iowa State Univ., Ames, IA, USA, 2009. [Online]. Available:http://lib.dr.iastate.edu/etd/10507

[11] S. Folea, G. Mois, L. Miclea, and D. Ursutiu, “Battery lifetime testingusing LabVIEW,” in Proc. 9th Int. Conf. Remote Eng. Virtual Instrum.(REV), Jul. 2012, pp. 1–6.

[12] D. Larios, J. Barbancho, G. Rodríguez, J. Sevillano, F. Molina, andC. León, “Energy efficient wireless sensor network communicationsbased on computational intelligent data fusion for environmental moni-toring,” IET Commun., vol. 6, no. 14, pp. 2189–2197, Sep. 2012.

[13] J. Ko, C. Lu, M. B. Srivastava, J. A. Stankovic, A. Terzis, and M. Welsh,“Wireless sensor networks for healthcare,” Proc. IEEE, vol. 98, no. 11,pp. 1947–1960, Nov. 2010.

[14] C. H. See, K. V. Horoshenkov, R. A. Abd-Alhameed, Y. F. Hu, andS. Tait, “A low power wireless sensor network for gully pot monitoringin urban catchments,” IEEE Sensors J., vol. 12, no. 5, pp. 1545–1553,May 2012.

[15] T. Sanislav and L. Miclea, “An agent-oriented approach for cyber-physical system with dependability features,” in Proc. IEEE Int. Conf.Autom. Quality Testing Robot. (AQTR), May 2012, pp. 356–361.

[16] F.-J. Wu, Y.-F. Kao, and Y.-C. Tseng, “From wireless sensor networkstowards cyber physical systems,” Pervasive Mobile Comput., vol. 7,no. 4, pp. 397–413, 2011.

[17] S. Tozlu, M. Senel, W. Mao, and A. Keshavarzian, “Wi-Fi enabledsensors for internet of things: A practical approach,” IEEE Commun.Mag., vol. 50, no. 6, pp. 134–143, Jun. 2012.

[18] C.-S. Lee, D.-H. Kim, and J.-D. Kim, “An energy efficient active RFIDprotocol to avoid overhearing problem,” IEEE Sensors J., vol. 14, no. 1,pp. 15–24, Jan. 2014.

[19] A. Pandharipande and S. Li, “Light-harvesting wireless sensorsfor indoor lighting control,” IEEE Sensors J., vol. 13, no. 12,pp. 4599–4606, Dec. 2013.

Silviu C. Folea (M’08) received the degree incontrol systems and the Ph.D. degree from theTechnical University of Cluj-Napoca (TUC-N),Cluj-Napoca, Romania, in 1995 and 2005, respec-tively. He is currently an Associate Professorwith the Department of Automation, TUC-N.His research interests include hardware and soft-ware embedded systems, reconfigurable systems,data acquisition, wireless networks, and low-powersensors.

He has authored nine books and book chapters,edited one book, and authored about 84 conference and journal publications;he was involved in over 33 research contracts, and two U.S. patents resultedfrom the research contracts he participated in.

George Mois (M’14) received the Degree in controlsystems and the Ph.D. degree from the TechnicalUniversity of Cluj-Napoca (TUC-N), Cluj-Napoca,Romania, in 2008 and 2011, respectively.

He is currently a Lecturer with the Department ofAutomation, TUC-N. His research interests includeembedded system design, digital design, Field-Programmable Gate Array (FPGA)-based systems,and fault-tolerant and error-tolerant systems.