Wireless Agricultural Sensor Network

A low cost wireless sensor network to monitor an agricultural space

M.L.G. Polpitiya

lalintha@ieee.org

W.K.S.S. Prasanna

wksudeera@gmail.com

D.P. Chandima

chandima@elect.mrt.ac.lk G.R. Raban

raehanar@ieee.org

D.T.S. Perera

startharusha@gmail.com

Department of Electrical Engineering

University of Moratuwa Moratuwa, Sri Lanka

U.K.D.L. Udawatta

lanka@ieee.org

Abstract – Enhanced agricultural methods can help to improve low productivity agriculture sectors of developing countries. In Sri Lanka, where the country is experiencing a rapid pace of development, the agriculture sector is beginning to appreciate the benefits that can be gained by merging agriculture with engineering. The use of electronic, communication, and information technologies has proven to result in productivity improvements in agriculture, but most developing countries are yet to adopt new technology due to financial and social restraints. Maintaining optimum levels in the microclimate of a crop is a key factor that affects the quality and quantity of its yield. By controlling the said microclimate, crops can be managed more effectively. The monitoring system that controls the surrounding environment must have the flexibility of addressing the diverse requirements of different cultivations, must be simple and reliable enough to operate, and most importantly must be affordable for the average farmer. In this paper, we present a method to develop a cost effective wireless sensor network that enables monitoring of an agricultural space from a single location, while enabling it to adapt to various types of cultivations. It also investigates the feasibility of establishing and promoting affordable agriculture electronics in Sri Lanka. Keywords – Humanitarian technology; wireless sensor network; controlled environment agriculture; low-cost technology; self-healing networks; object oriented concepts.

Statistics indicate that the productivity of the agriculture sector is very low. The Sri Lankan agriculture sector employs nearly one-third of the total workforce. But its contribution to the Gross Domestic Product (GDP) is a mere 12.8% [1]. A market exists for methods and equipment that can improve this situation as the government and private businesses are increasingly seeking ways to improve agricultural productivity. In order to design and implement a low-cost sensor network, it is required to first understand the present situation of the local agriculture industry. A series of discussions were held with experts in agricultural training institutions to find out the challenges currently faced by the agriculture sector, the resources available for farmers and the accessibility of those resources. A key drawback identified was the lack of sufficient information and data available for farmers engaged in intensive and protected agriculture. Most of the farmers were relying on their past experience or a rule of thumb, which sometimes may not be accurate enough to reap the optimum benefits. Intensive and protected agriculture are the new trends in agriculture which can guarantee a high yield, and presently there is a need to improve the productivity of these sectors economically. The Wireless Sensor Network (WSN) presented in this paper is designed to monitor an entire crop or a group of different crops from a single location. By constantly monitoring the microclimate of a crop, it can be effectively

I. INTRODUCTION

The field of agriculture electronics has rapidly expanded since its emergence during the 1960s. It now finds many uses in the fields of intensive agriculture, precision agriculture, protected agriculture, and agriculture automation. Integrating electronics into agriculture has resulted in increased levels of productivity. The Sri Lankan agriculture sector is beginning to incorporate electronics in its processes. One of the greatest challenges faced by the modern Sri Lankan farmer is the lack of affordable products that will ensure a reasonable rate of return.

C – Coordinator, R – Router, E – End Device

Figure 1. Network hierarchy

controlled [2]. Another function of this WSN is gathering data. Farmers need to tend to their crops exactly at times when it is crucial, with minimum resources. If data can be gathered long enough, it can be processed to create useful information, addressing the lack of information faced by farmers today. This WSN can be further extended to be used in research crops, agriculture automation, and even to build a national database for information on geographically based crops.

II. OBJECTIVE AND SCOPE

Our aim is to design a low cost wireless sensor network that can make agricultural processes more efficient. It is designed specifically for cultivations operating under a controlled III. METHODOLOGY

A. Hardware Implementation

The WSN is designed to sense physical parameters such as ambient temperature, humidity, pH level, and light level, transmit that data to a centralized computer, and process that data to retrieve useful information. Its basic function is to monitor an entire crop or a group of crops, and help in controlling its surrounding environment.

The network consists of nodes with integrated sensors placed strategically covering the entire area of the crop. The network topology employed in this sensor network is mesh as seen in Figure 1. A mesh network is the most effective topology for this type of WSN in terms of flexibility [4]. Three environment such as a greenhouse. The network senses physical parameters and transmits that information to a centralised computer, thus enabling monitoring of the entire farm from a single location. The target group for the implementation of the end product is small to medium scale commercial farmers operating in greenhouses.

Greenhouses today are driven by agricultural concepts such as Intensive Agriculture and Protected Agriculture. Intensive agriculture or intensive farming is an agricultural process that yields a high amount of crop relative to the utilized land area. It is characterized by its high input of labour, capital and other resources. Intensive agriculture is the primary method of food production in many developed nations [2]. Although it is not widely utilized in Sri Lanka, it is neither an entirely alien concept to us. Ancient farmers used terracing and paddy fields to cultivate their rice crops, which were basic intensive farming techniques.

Protected agriculture is another widespread trend in

agriculture. It advocates growing and maintaining plants in a protected environment such as a greenhouse, where this cultivation can be kept safe from adverse environmental conditions. When aerial and soil conditions are controlled in a protective environment, it is called Controlled Environment Agriculture (CEA). “Optimum growth conditions for agricultural produce” has been a popular research topic for some time now. Information on this is widely available for farmers [3].

Figure 3. Schematic diagram of a coordinator

network layers facilitate the mesh topology;

1) Coordinator 2) Router 3) End Device End devices placed at the far end of the network establish a wireless connection with the nearest router it can detect. The router communicates with other routers to find the shortest path to the coordinator. A router can support up to a maximum number of 8 end devices [5]. End devices and routers can input sensor data as well as perform their specific communication function. The coordinator is a single wireless module interfaced through a microcontroller to a USB port that connects to a computer, Figure 31.

A schematic diagram of a node is seen in Figure 21. A node essentially consists of analog and digital sensors, the microcontroller, and a communication interface.

Physical parameters are sensed as analog or digital data. Analog data is converted to digital for wireless communication. The conversion is done by a microcontroller using its in-built analog to digital data conversion facility. Figure 2. Schematic diagram of a node in the sensor network

1

Abbreviations; Tx- Transmission, Rx- Receive, HMI– Human Machine Interface, UART- Universal Asynchronous Receiver/Transmitter, SPI- Serial Peripheral Interface, DO/DI– Digital Inputs/Outputs, EEPROM– Electrically Erasable Programmable Read Only Memory.

Figure 4. Hardware implementation of a node

Inputs to the microcontroller can be increased by the use of multiplexers.

Wireless modules used in this network consume a very low amount of power, and therefore can be conveniently powered for an extended period of time through a battery or a solar panel depending on the application. These modules operate within the ISM 2.4 GHz frequency band with a range of 40 m for indoor applications, and 120m (line-of-sight) for outdoor applications [5].

Wireless modules can be programmed for a specific function. This network requires modules to perform three functions; End Device, Router, and Coordinator. Once programmed, the modules can be configured to be used in various types of network topologies.

A special feature of this network is that it is self-healing; if a node stops functioning, the network will automatically reroute itself without the dis-functioning node. Fault tolerance and reliability are other significant properties that increase the

accuracy of this network.

Data received by the end devices are communicated to the coordinator through routers serially. A cyclic process is adopted to read data from sensors at each node. The

microcontroller controls how the sensors read data, and how

the cycle is repeated. A similar process is used when the end devices are communicating with a router, and the router with the coordinator. The coordinator is directly connected to a centralized Personal Computer (PC) through a USB port. Data

communicated to the coordinator is stored in the PC for future reference. The hardware implementation of the node as seen in Figure 4 was tested in the In-Service Training Institute at the Gannoruwa Agricultural Research Centre in Sri Lanka. Figure Figure 5. Testing a node in a hydroponic system at the

Gannoruwa In-Service Training Institute

5 depicts the testing of the pH sensor input at the institution’s hydroponic cultivation.

B. Software Implementation

The WSN Graphical User Interface (GUI) Program was developed to assist the user of the sensor network to obtain real-time data of the required environmental factors. The GUI program as seen in Figure 6 provides the following facilities; • Real-time data plotting of each sensor of each node (Figure 7)

• Analyse real-time data, by comparing generated simultaneous graphs

•• Ability to store and process data for research purposes • Platform independence and enhanced user friendliness Capability to configure wireless modules

The Rational Unified Process (RUP) is used as the system design methodology. Its goal is to ensure the design of high quality software that meets the needs of its end-users, within a predictable timeline and budget [6]. In Wireless Sensor

Simulator development, the most used approach is Object-oriented approach particularly for interactive system development. Object Oriented Concepts ease a great deal of work by providing technologies such as Object Oriented

Analysis (OOA) and Object Oriented Designing (OOD). Multi-threading concepts were used for real time analysis of the

simultaneous graphs [7]. The user interface design is an essential part of the overall system design process as it is the layer that end users are directly in contact with. A console program is developed to assist an implementation engineer or technician on configuring

purposes. It provides facilities to manipulate raw data, and to test receiving and transmitting ends of the wireless modules. The program in the microcontroller facilitates the operation of each node and routing function of the network. The program is developed providing the following facilities;

• Processing of raw data of sensors (analog to digital

conversion) • Display real-time processed data and status on an LCD screen

• Routing data through wireless modules

• Indicating status through indicator LEDs (Power, Tx/Rx, Sensor inputs)

A number of development tools are used for the software

implementation;

• Development Environment

- Ubuntu 10.10 (Maverick Meerkat) - QT Creator IDE and tools

- C++ code editor (QT Creator)

• Compilers

- g++ compiler

- mikroC compiler v8.2.0.0

- Microsoft HTML Help Workshop - Setup Factory 8.0

Figure 7. Real-time data plotting in the software interface

The end user of this network will be receiving a complete user friendly software program. It is designed to be installed in the computer to which the coordinator is connected. The following recommended system requirements can fully utilize the capabilities of the WSN by providing the best experience. - Dual 2GHz CPU - 2+ GB RAM

- 10 MB of available hard-disk space

- Windows 7 x 32 or Windows Vista Service Pack 2 x 32

IV. RESULTS

Results presented in this paper are two-fold; results assessing the need for a WSN in the Sri Lankan agricultural sector and results pertaining to the implementation of the WSN.

A demand exists in the current Sri Lankan agriculture

sector for devices that can automate agricultural practices.

There are many wireless sensor network designers and manufacturers around the world. But local farmers face difficulties in purchasing them because of their high price. A

typical network will cost over 1,000,000 LKR. Mid to small scale farmers cannot afford this. There is an opportunity for affordable wireless agricultural sensor networks that can address the problems of the Sri Lankan farmer.

A. Major Findings

Currently there is approximately 6,000,000 m2 of commercial greenhouses operating in Sri Lanka. This includes large-scale commercial greenhouse operators as well as small scale local farmers. Average capital cost to construct a typical greenhouse is around 5000 LKR/m2. If we assume that an average greenhouse span is around 400 m2, a commercial farmer will have to invest LKR 2,000,000 to get his farm up and running. Our product is priced at LKR 50,000 on average. Therefore, it is feasible to assume that a farmer who invests LKR 2 million as a start-up will consider implementing our WSN in their greenhouse.

This WSN can be modified to determine the optimum level of input for each resource [3]. This eliminates the possibility of resource wastage.

Figure 6. Graphical User Interface (GUI) of the

software program

Requirements of crops change with the stage of growth as shown in Table I for tomato [8]. Four distinctive stages of growth are identified for a tomato crop, and the ambient temperature requirement of each varies therefore, provisions must be provided to modify control values put into the system, so that the requirements for each stage of growth can be met. Requirements may also vary according to the height of the crop. Therefore, positioning of sensors is very important. Lower canopy temperature tends to be higher than that of the upper canopy. This characteristic must be taken into account when implementing control measures.

The ideal point to measure soil parameters is mid root level.

The top root level closer to the ground level mostly has matured roots. As we go deep into the soil, younger and tender roots appear. Mid-level roots are found to be the most active. It

is the most critical area for nutrient absorption. Therefore, soil parameters must be measured at mid-root-level.

Methods that are currently in use for Controlled Environment Agriculture;

• Misters are used to control temperature. A valve can be

opened for the mister to spray water. • Drippers control the amount of water and fertilizer.

TABLE I. TEMPERATURE REQUIREMENTS OF DIFFERENT STAGES OF

GROWTH FOR TOMATO

Temperature (oC)

Stages Minimum Optimum

Range

Maximum

Seed germination 11 16 – 29 34 Seedling growth 18 21 – 24 32 Fruit set 18 20 – 24 30 Colour development 10 20 – 24 30 Temperatures below 21 °C can cause fruit abortion Plant tissues are damaged below 10 °C and above 38 °C

TABLE II - OPTIMUM TEMPERATURE AND PH VALUES FOR TOMATO AND

BELL PEPPER

Cultivation Optimum Range (oC)

Temperature pH Tomato 21 – 24 5.5 – 6.8 Bell Pepper

21 – 27 (day)

16 – 21 (night)

6 – 6.8

Control mechanisms can be implemented to automate these actions such as opening a valve, dimmer action, etc. We have identified this as the next stage of development.

B. Optimum Conditions

As mentioned before, optimum environmental conditions for growth is dependent on many factors;

• Stage of growth of the plant

• Growing media • Type of greenhouse • Variety of the cultivation

• Height of the plant

Table II lists optimum temperature and pH values for tomato and bell pepper [8], [9]. It is generally not recommended to have more than one type of crop in a single greenhouse. This facilitates better management of crops.

C. Network Performance

The following performance characteristics were observed in the implemented sensor network;

• Microcontroller operating at a voltage of 5V, and a 20MHz crystal oscillator

- Power consumption: 0.45 W

• Microcontroller operating at a voltage of 3.3 V with 4 MHz Clock

• - Wireless Module

Data Receive: 40mA x 3.3 V - Data Transmit: 40mA x 3.3 V

- Average power consumption for communication: 132mW - Sleep mode Current: 0.5 mA

- Average Total power ( for essential tasks) consumed by a Node: 0.25 W

- 110 hours with sleep mode assist (Using a common 9V battery of 550 mAh)

V. FUTURE ADVANCEMENTS We have identified many opportunities to develop the overall system. Studies are being carried out to find new methods of having a less power consuming sensor network, which is a major challenge in WSN applications. In order to increase functional flexibility a sleep command can be introduced for remote devices. This will be advantageous for the user in numerous ways; user will be able to “wake-up” or activate a selected number of nodes and gather data without interrupting the whole network.

In order to ensure proper utilization of energy, 3.3 Voltage

level at reduced clock frequency was used. We are further

analyzing new methods in NanoWatt Technology introduced for Microcontrollers. Power budgets have been prepared and this gave us the opportunity to identify dump loads along with a priority list to dispatch. Operation of the remote device can be altered considering the availability of the power source. Coordination between available power level and operations may help to keep the system live to perform its basic tasks for an extended period of time [10]. Also, introducing solar power and other power scavenging techniques could keep devices active for months without any need for maintenance [11]. Further, we are focusing on ways to achieve long distance communication. An option is to have local networks and use GSM networks to reach a central location. This would be a solution for widely spread agricultural lands.

The next step would be controlling the environmental conditions as suitable for each crop [12]. A confined

environment like a Greenhouse could achieve optimum

conditions with our proposed monitoring system and mechanisms to control the parameters. Further research is

conduct on the control of actuators to rectify the environment in a confined space with closed loop control system. Much work needs to be carried out to identify the proper implementation with the help of Cultivators and Crop researchers. VI. CONCLUSION

The scope of this project is limited to crop-based intensive agricultural applications. This scope is timely because currently the government is assisting small scale vegetable producers in adopting intensive farming practices.

Electronic technology is yet to be fully utilized in Sri Lankan agriculture. Due to this fact, our farmers have failed to exploit the maximum benefits out of the limited resources available to them. This WSN aims to cater to the tech-savvy new generation farmers who are aware of the importance of sustainability in agriculture.

Implementation of a wireless sensor network as described in this paper was successful. We were able to build a mesh network with the self-healing property successfully. Data transmission through the network was proved to be efficient and reliable. The software application was user friendly, needing an average knowledge about handling software programs to navigate through it.

The greenhouse complex in the In-Service Training

Institute of Gannoruwa is an ideal place to implement a pilot

project of this WSN. It is a government institute closely working with local farmers which provides them with training on emerging trends in agriculture, information sharing on improved methods and technologies, and it has a far reaching network that includes farmers from all over the country. By implementing a WSN in an agricultural training institute which already has an established network, we have a channel to share our WSN technology with local farmers, and educate them

about the benefits they can gain from it. The WSN can even be further improved by getting feedback from end users.

Having large scale integrated crop management systems will help a tropical country like Sri Lanka to identify the spread of various diseases, insect attacks, and sudden changes of environment and also to predict upcoming severe weather conditions. Disease spreading and insect attacks are severe threats for cultivations. Identifying and predicting the possibility of infection in early stages is critical for protecting the yield’s quality and quantity. Large-scale integration and

number of sensors from many a source has proven to be very beneficial in the field of agriculture. A web based common platform to share data would be beneficial for many.

There is a market for low cost wireless sensor networks for agricultural applications in Sri Lanka. This network as a product has great potential to be commercialized. With adequate funding and technical assistance, we can produce a superior product that can serve our nation.

ACKNOWLEDGEMENT

Our sincere gratitude goes out to Mr G.D. Amarasinghe, Agriculture Officer at the In-Service Training Institute in Gannoruwa, and Dr. Keerthi Wickramasinghe, ROIC, Horticultural Crops Research & Development Institute (HORDI), Gannoruwa for their generous support and assistance provided in doing background research. Lastly, we thank many individuals, friends and colleagues who have not been mentioned here personally in making this educational process a success.

REFERENCES

[1] Department of Census and Statistics Sri Lanka, Statistical Data Sheet,

2011. [2] S. Chaudhary, V. Sorathia, and Z. Laliwala, \"Architecture of sensor

based agricultural information system for effective planning of farm activities,\" In Proceedings of IEEE International Conference on Services Computing (SCC 2004), pp. 93-100, September 2004.

[3] R. Beckwith, D. Teibel, and P. Bowen, \"Report from the field: results

from an agricultural wireless sensor network,\" In Proceedings of 29th Annual IEEE International Conference on Local Computer Networks, pp. 471-478, November 2004.

[5] K. Konstantinos, X. Apostolos, K. Panagiotis, and S. George, \"Topology

Optimization in Wireless Sensor Networks for Precision Agriculture Applications,\" In Proceedings of International Conference on Sensor Technologies and Applications, (SensorComm 2007), pp.526-530, October 2007.

[4] M. Hirafuji, S. Ninomiya, T. Kiura, T. Fukatsu, H. Hu, H. Yuoichi, K.

Tanaka, K. Sugahara, T. Watanabe, T. Kameoka, A. Hashimoto,R. Ito, R. Ehsani, and H. Shimamura, \"Field Server Projects,\" In Proceedings of International Symposium on Applications and the Internet Workshops (SAINT Workshops 2007), pp.75, January 2007.

[6] International, Digi. XBee Series 2 OEM RF Modules - Product Manual .

Lindon, UT : Digi International, 2007. [7] H. Huang, C. Gill, and C. Lu, \"MCFlow: A Real-Time Multi-core

Aware Middleware for Dependent Task Graphs,\" In Proceeding of 2012 IEEE 18th International Conference on Embedded and Real-Time Computing Systems and Applications (RTCSA), pp.104-113, August 2012. [8] Albrecht, C.; Hagenau, R.; Doring, A.; , \"Cooperative software

multithreading to enhance utilization of embedded processors for network applications,\" In Proceedings of 12th Euromicro Conference on Parallel, Distributed and Network-Based Processing, 300- 307, February 2004. [9] W. M. K. B. Wahundeniya, R. Ramanan, C. Wickramatunga, and

W.A.P. Weerakkodi, Peradeniya. Comparison of growth and yield performance of tomato varieties under controlled environment conditions.: 2007. [10] R.A.G. Ranawaka, P.A.I.S. de Alwis, W.A.P. Weerakkody, K.P.S.B.

Premalal, and U.S.K. Abeysinghe, \"Hydroponics bell pepper production and plant nutrients under tropical climate,\" In Proceedingd of Bibliotheca Fragmenta Agronomica (IX ESA Congress). Pp. 277-278, 2006.

[11] C. Lin; Y. He, and N. Xiong, \"An Energy-Efficient Dynamic Power

Management in Wireless Sensor Networks,\" In Proceedings of the Fifth International Symposium on Parallel and Distributed Computing (ISPDC '06), pp.148-154, July 2006.

[12] E. M. Yeatman, \"Energy scavenging for wireless sensor nodes,\" In

Proceedings of 2nd International Workshop on Advances in Sensors and Interface ( IWASI 2007), pp.1-4, June 2007.