Environmental Sensor Network

Transcription

1 Environmental Sensor Network The New Jersey Governor s School of Engineering and Technology 2014 Arthur Belkin Wayne Hills High School Yunhee Kang Ridge High School Shweta Dutta South Brunswick High School Hope McGovern Haddonfield Memorial High School Matthew Scalamandre Moorestown Friends School Abstract In order to demonstrate how technology can be used to facilitate human interaction with the environment, the research team designed and built a proof-of-concept mesh network to collect data about the home environment and display it to a base station. Simplifying human to environment interaction makes life easier in a world where many people, devices, and tasks require a person s attention. The network connects two sensors that report data to Arduino platforms. These platforms communicate with XBee radios, which in turn talk to a base laptop that displays the data to the user of the network. A mesh network that reports conditions within a house is a viable way of monitoring and regulating an area; this eases human to environment interactions. Subsequent work might implement technology for the mesh network to automatically respond to changes in the environment, as opposed to only reporting them back to the user. 1. Introduction Modern technology is intrinsically interactive. While it does connect humans to virtual reality and to each other, it can also be used to connect humans to their environment; such interactions can be both monitored and regulated. Sensor network technology has been previously used to communicate data from remote locations quickly and effectively. For example, a satellite-based remote sensor network allows researchers at the University of Hawaii to analyze the eruptions of volcanoes on remote islands off the coast of Antarctica within mere hours of the explosion 1. Figure 1. A Nest Temperature Thermostat 2 Sensors can also be utilized in more quotidian situations; Nest home thermostats include temperature, humidity, light, and activity sensors in the unit in order to construct a personalized heating and cooling schedule. The Nest system automatically adjusts the temperature of a home according to activity, season, ambient temperature, and recorded preferences. It also communicates data about a user s energy utilization and can be controlled via a smartphone application 3. These systems provided the initial inspiration for the project; however, the group was determined

2 2 THE GOVERNOR S SCHOOL OF ENGINEERING AND TECHNOLOGY 2014 to create a simpler and cheaper in-home monitoring system it dubbed a Smart-Home. A Smart- Home is a house that has multiple sensors set up to detect the condition of appliances and objects throughout the building and report them to a central computer. As a proof-of-concept of how technology can be used to organize environmental data and facilitate regulation, the group created a wireless sensor network to perform and regulate various household tasks. Sensors were used to monitor the progress of a washing machine cycle and the ambient light level. 2. Background 2.1 Open-Source Movement Programmers and hardware developers alike take part in the ever-changing open-source movement. The concept of open-source refers to the freedom to distribute code and hardware for reuse and modification. The operating system GNU, which recursively stands for GNU s Not UNIX, is one of the more well-known products of this movement, with UNIX referring to the more restrictive operating system controlled by AT&T and the University of California, Berkeley 4. Today, platforms such as Arduino allow for easy integration of hardware and software into anything from personal projects to full-scale replicable and marketable products. Open-source allows amateur coders and builders, start-up companies, or any person with an idea to contribute to the world of technology, putting some power into the hands of consumers who otherwise must rely on the giant technology companies of today. Essential to the project, the opensource movement allows for effective development of proof-of-concept designs. 2.2 Arduino Arduino is an open-source computing platform based on a simple microcontroller and an Integrated Development Environment (IDE). The physical board, known as the microcontroller, has multiple pins and outputs which can be used to build circuits. Unlike other programmable circuit boards, Arduino requires only a USB cable to load new code onto the microcontroller. Additionally, the simplicity of Arduino s IDE - which has a language based on C++ - makes the programming easy to learn and manipulate 5. Other advantages of Arduino include its relatively cost-effective platform (with the more expensive modules valued at $50), cross-platform capabilities as it can run on Windows, Macintosh OSX, and Linux operating systems, and the open-source nature of its software and hardware Sensors Sensors are devices that allow humans or computers to detect changes in their environment. Most sensors are designed to respond to one particular type of stimulus, such as temperature or light, and output data that corresponds to the intensity of the event. The sensors used by Arduino output data as either a digital or analog signal. The two sensors used for this project were a light sensor and a piezoelectric sensor. The light sensor combines a silicon diode that converts incipient light into an electric current with a device which converts this current into a square wave with frequency directly proportional to the current, and therefore proportional to the light intensity. This signal can be read by the Arduino as a digital input 7. Piezoelectric sensors detect vibrations, shocks, and impacts. Such an impact causes the sensor to deform, which in turn produces a voltage. The voltage level corresponds to the degree of deformation the sensor experiences; this level can be read by the Arduino board 8. Thus, sensor networks allow technology to collect, analyze, and respond to environmental data otherwise inaccessible to humans. 2.4 XBee and Mesh Network XBee radios, manufactured by Digi International, Inc., are small chips that wirelessly transmit information and allow multiple Arduinos. to communicate with each other. The protocol, or digital rules for data exchange between the radios, is ZigBee, which is typically used by small, low power radios to transmit small amounts of data over a short range 9. Multiple XBee radios can be connected

3 ENVIRONMENTAL SENSOR NETWORK 3 to form a mesh network, providing a more extensive range with each XBee acting as a node. Much like the networking system used by cellular devices, XBees act as radio towers to relay wireless information. For multiple radios to be able to send and receive information, they must be set to Application Programming Interface (API) mode 10, as opposed to Attention (AT) mode, which only lets two radios communicate 11. API mode sends and receives data in frames, a technical term for radio messages. Radios can be configured into two connection arrangements: full mesh and partial mesh topologies. In the former case, all radios are interconnected; in the latter, some nodes in the network have restricted communicating ability with respect to the other nodes 12. A full mesh network was chosen to connect all of the XBee radios. 3.1 Sensor Selection 3. Methods The research group decided to create a network of multiple sensors to monitor in-home activities as a proof-of-concept of how technology can collect and react to ambient data. A piezoelectric sensor converts motion into electrical voltage, which can be monitored in the home environment to determine whether a washing machine is running. A light sensor can be used to determine the level of light in an environment. Substantial documentation exists for both of the sensors, and they are both extremely cost-effective. 3.2 Arduino Platform Extensive information on the Arduino microcontroller and its corresponding programming language is readily available online from the creators of the Arduino board as well as from amateur developers. Therefore, it was not necessary to fabricate a wholly new design for an Arduino-based mesh network. The open-source nature of the board encourages developers and coders to share their projects on the internet. In addition to official documentation for using the selected sensors on the Arduino platform, many examples of sensor-interfacing Arduino projects were available Figure 2. Piezoelectric Sensor: SEN Figure 3. Light to Frequency Conversion Sensor: Model TSL235R 14 to the research team to use and modify. It was also necessary for the researchers to familiarize themselves with the Arduino programming language itself, which is based on C++. This project uses two Arduino Uno boards and one Arduino Mega board. 3.3 Circuit Development and Testing The light sensor returns the intensity of the light it senses with the units µw/cm 2, while the piezoelectric sensor returns a number from 0 to 1023, with 0 being the lowest and 1023 being the highest. Safety factors, particularly resistors, were incorpo-

4 4 THE GOVERNOR S SCHOOL OF ENGINEERING AND TECHNOLOGY 2014 Figure 4. Arduino Uno: Model R315 Figure 5. Arduino Mega: Model 2560 R216 rated into each design to shield the Arduino from voltage spikes; the piezoelectic sensor generates a potential difference when it senses a vibration that deforms it. Each circuit was then prototyped on a breadboard. The sensors were thoroughly tested before being incorporated into the final network. To confirm that each sensor worked properly, they were assessed in different conditions. It was determined that the piezoelectric sensor required a very secure attachment to a vibrating object in order to accurately measure the oscillation level. The light sensor was tested under varying intensities to determine its sensitivity. Perfecting data passing from one sensor to a separate laptop required troubleshooting and incremental development; for example, instead of creating the full mesh network and testing it, the research team started by passing sensor information from one Arduino to another and then to a computer, fixing hardware and software challenges as they arose. 3.4 XBee Radios and Creation of Mesh Network XBee radios constitute the nodes of the mesh network. A diagram of the mesh network is seen in Figure 6. Using X CTU software, the four XBee radios were configured with one as the coordinator and the others as routers. This setup allows for simplified packaging of sensor messages by the Arduinos, as the destination address is easily set to the coordinator XBee radio that is connected to the base station for data display. Each Arduino platform must package the messages, often referred to as a frame for the XBee radios to both transmit and read the information. The frame contains fields the XBee radios will recognize, with a message containing a start byte, which alerts the radio that a valid message transmission is beginning. Next in the frame are frame details, which specify attributes such as the length of the message and the type of message being sent. A frame ID asks for a specific return sequence confirming that a message was sent successfully, left empty for this project as the network created is too small to require this check. This is followed by two addresses, one 64-bit and another 16-bit. The 64-bit address is unique to all devices in the world and is called a destination address, whereas the 16-bit destination network address references the coordinator XBee radio, unique to the mesh network created. A broadcast radius was set, declaring the maximum number of nodes for the frame to pass through before reaching its destination. Setting the radius to 0 allows the data to reach the receiver in however many nodes are necessary, and this is the radius used in the mesh network. Transmission options can also be set allowing for data encryption, unnecessary for a low-security smart-home network. Finally, the actual sensor data formatted by the Arduino was added to the frame, followed by a checksum byte which ensures the entire data packet was received. filler filler filler filler filler filler filler

5 Figure 6. Concept of Operations: Mesh Network ENVIRONMENTAL SENSOR NETWORK 5

6 6 THE GOVERNOR S SCHOOL OF ENGINEERING AND TECHNOLOGY 2014 Figure 7. XBee Radio: Model XBEE Formatting Sensor Data for User Data is sent from Arduino nodes to a host computer via XBee radios. However, this data must be formatted accordingly in order to be understood by a human user. For this task, a program was created using the Python programming language to capture data from the Serial monitor, format it, and display it in an organized fashion. PySerial is a Python application programming interface module used to create an application to store and display data. The Arduino platform that is connected to the piezoelectric sensor transmits one byte of sensor information: a 0 indicating no vibration, or a 1 indicating vibration. These values are only sent if the state changed from the last sensor reading, otherwise no data is sent. If a 0 is sent to the coordinator XBee, then PySerial will display a message telling the user that the washing machine/dryer cycle is done. The Arduino connected to the light sensor sends the intensity of the light in µw/cm 2, which is converted by PySerial to a low to high intensity scale from 1 to 10 and displayed in a manner easily understood by the user Results and Discussion The research team built a mesh network that successfully communicates environmental data to a host computer. Sensor data is relayed from one Arduino board to another by way of XBee radio signals, which pack information into data frames. Messages are sent to the one XBee attached to a host laptop and are displayed on the base station monitor through a Python program. Arduino code for the two sensor nodes is included in the Appendix B. The objective of the project was to demonstrate the usefulness of a mesh network in facilitating human interaction with the environment. However, the result was limited by factors including the reliability and range of the available sensors and the practicality of a mesh network on the scale to which it was built. The design would have to be modified to be implemented on a larger scale. Significant difficulty was experienced in applying a conceptual understanding of the mesh network to the challenge of physically creating working code to send messages between nodes on the network. The majority of difficulties the team encountered was in writing the programs to run the project. Two distinct codes were needed to power the network: one to send and receive sensor data and one to display the data on the host laptop. Reading the piezoelectric sensor was straightforward as it outputs an analog voltage signal that the Arduino can read easily. However, it was difficult to acquire usable data from the washing machine, as the values returned were too low. The piezoelectric sensor was tested in various configurations, but it was unable to detect any meaningful events until a weight was placed on the sensor. The weight vibrated in concert with the washing machine, intensifying the stimulus. The program used by the Arduino is also designed to increase accuracy by averaging the readings over a period of one second. The light sensor presented a different set of problems. It was easy to set up physically; however, reading the light sensor proved more difficult as the sensor output a square wave that cannot be natively interpreted by an Arduino. The problem was resolved by finding sample code online that treated each peak of the square wave as a digital input and introduced an interrupt that kept count of the number of inputs that occurred each second 19. This allows the Arduino to calculate the intensity of incipient

7 ENVIRONMENTAL SENSOR NETWORK 7 light with acceptable accuracy. At higher intensities, the interrupts occur with enough frequency to overload the Arduino s processor, causing it to return incorrect values; however, this problem is not relevant at the intensities of light for which the proof-of-concept network is intended to be used. The team also encountered difficulties in programming the message passing system. Our main challenge was formatting the data into a message that the XBee radios would understand. The group decided that the problem of differentiating between two sensors would most easily be solved by having the message be of different lengths for each sensor. For the piezoelectric sensor, we simplified the response into one byte, whereas for the light sensor we stored the value into two bytes. The light sensor returns a value that is often greater than 255, and therefore must be stored in two separate bytes. However, in dark environments, the sensor returns a value less than 255. If the XBee attempted to send this data, the base station would misinterpret the message as coming from a piezoelectric sensor. To resolve this problem, the code was modified to force the Arduino to parse the data into two variables before sending it. This approach forces the radio to send two values in the message, which preserves the base station s ability to discern between the two sensors. Setting up the XBee radios required the use of X CTU software, which allows for easy installation of firmware. One XBee was made a coordinator, while the others were made into routers. Originally, the data was hard-coded as a long series of write method statements that wrote information to the Serial port one byte at a time, problematic as the frame was being sent in parts and received in parts as viewed through X CTU. While testing the network, the X CTU software was sending message requests and receiving partial data, but this issue was fixed in formatting the package of data as a receive request in a byte array. Using a for loop, we filled an array of bytes one byte at a time with the correct frame format as discussed in section 3.5. In this loop, the checksum byte is filled with a 0, as it has to be calculated before being put in the byte. A second loop calculates the checksum, summing the values from the fourth byte to the last non-checksum byte, dividing by 256, and then taking the remainder when dividing by 256 yet again. This byte is inserted as the last byte of the data, and then the package is transmitted using the write method that accepts a byte array and its size as arguments. Through trial and error, the team recognized the need to send the bytes in hex instead of decimal, as the hex number system is what the X CTU software reads the messages in. Finally, PySerial was implemented to read the values sent in hex and translate them into console messages. 5. Conclusion Based on the research completed, a mesh network has the ability to simplify human interaction with the environment. This type of network allows technology to act as an interface between the real world and humans, reading and interpreting raw data and formatting it so that it is easy to understand and respond to. The team s network consisted of two sensors and a base station, and the setup can easily be expanded to include more sensors and report other environmental information to the user. The group s network successfully communicated data about the home environment, specifically data about the light level in a room and about the functioning of a washing machine. Possible additions and improvements that could be made to the created mesh network include adding multiple sensors to one Arduino platform, adding more nodes to the network to collect data from other locations, developing a Graphical User Interface (GUI) to display formatted sensor data and integrating it into a smartphone application, and creating actuators that allow the network to not only transmit data, but also react to the conditions sensed by taking physical action, e.g. opening and closing window blinds in response to light levels. Further research might look into making mesh networks secure, in order to avoid others from reading data transmitted, or possibly broadcasting malicious signals that compromise the performance of the network. As the open source movement continues to expand and include programmers and hardware-builders

8 8 THE GOVERNOR S SCHOOL OF ENGINEERING AND TECHNOLOGY 2014 alike, the future application of mesh networks also grows. Mesh networks can also be used in outdoor settings where people might have trouble checking certain conditions manually due to physical limitations. A sensor placed in such areas could provide a live feed of such conditions, connected in a full mesh creating a local network that does not require the internet to send and receive data. Such a network would require sensors that are weather resistant and robust code that ensures network security, both easily developed and obtained from the open source movement. Mesh networks, when combined with the advantages of open source, have the potential to facilitate human to environment interactions. 6. Acknowledgements We would like to acknowledge our project mentors Josef Grossmann, Joe Mirizio, Ryan Flynn, and Edward Kahn from Lockheed Martin who have generously provided their time along with the necessary resources and information to successfully complete our proof-of-concept mesh network. The team would also like to acknowledge the New Jersey Governor s School of Engineering and Technology s Program Director Dr. Ilene Rosen, along with the Assistant Program Director Jean Patrick Antoine and the entire GSET staff for presenting such an incredible opportunity to the team, key to the coordination and organization of such a comprehensive and exciting summer program. We also extend thanks to our Residential Teaching Assistant Maya Saltzman for overseeing the project, making sure we had the required tools and workspace. A final thank you goes to all of the sponsors of the New Jersey Governor s School of Engineering and Technology, specifically Rutgers University, The State of New Jersey, Morgan Stanley, Lockheed Martin, Silverline Windows, Jersey South Industries, Inc., The Provident Bank Foundation, and Novo Nordisk. 7. References [1] Schmidt, Laurie J. "Sensing Remote Volcanoes." NASA Earth Observatory. National Aeronautics Space Administration, 13 July Web. 11 July [2] Nest Temperature Thermostat. Digital image. Amazon. N.p., n.d. Web. 11 July [3] "Home." Nest. Nest Labs, n.d. Web. 13 July [4] Henderson, Harry. "Open-Source Movement." Science Online. Facts On File News Services, n.d. Web. 10 July [5] "Arduino Introduction." Arduino - Introduction. Arduino, n.d. Web. 13 July [6] "What Is an Arduino?" SparkFun. SparkFun Electronics, 26 Feb Web. 8 July [7] Light-To-Frequency Converter Datasheet. N.p.: Texas Advanced Optoelectronic Solutions, Sept PDF. [8] Piezoelectric Sound Components. N.p.: Murata Manufacturing Co., Ltd., n.d. PDF. [9] "XBee Buying Guide." SparkFun. SparkFun Electronics, n.d. Web. 19 July [10] "What Is API (Application Programming Interface) Mode and How Does It Work?" Digi. Digi International Inc., n.d. Web. 19 July [11] "The AT Command Set." Digi. Digi International Inc., n.d. Web. 19 July [12] Brinton, Stephen. Mesh Networks. N.p.: Gordon College, PPT. [13] Piezoelectric Sensor. Digital image. SparkFun. SparkFun Electronics, n.d. Web. 15 July [14] Light to Frequency Conversion Sensor. Digital image. SparkFun. SparkFun Electronics, n.d. Web. 15 July [15] Arduino Uno. Digital image. Tested. N.p., n.d. Web. 15 July [16] Arduino Mega. Digital image. Electroschematics. N.p., n.d. Web. 15 July [17] XBee S2. Digital image. Entesla. N.p., n.d. Web. 15 July [18] Liechti, Chris. "Welcome to PySerial s Documentation." PySerial. N.p., n.d. Web. 19 July [19] Tillaart, Rob. "Arduino Playground - Sensor TSL235R." Arduino Playground. Arduino, 30 Nov Web. 10 July Appendix A A1. Arduino Code for Light Sensor

9 ENVIRONMENTAL SENSOR NETWORK 9 no A2. Arduino Code for Piezoelectric Sensor