Remote Monitoring for Wireless Sensor Based Irrigation System in Malawi Million MafutaA,1, Timothy ChadzaA, Harry GombachikaA, Graham AultB, Damien FrameB, Elijah BandaA A
Electrical Engineering Department, University of Malawi-Polytechnic, Blantyre, Malawi 1
[email protected]
B
Electronic and Electrical Engineering Department, University of Strathclyde, Glasgow, UK
Abstract— In the recent years there has been an increase in the application of Wireless Sensor Networks (WSNs) in agriculture, specifically aiding automatic application of water, chemicals and fertilizers to the field. Since WSNs are still under development stage, they are at times unreliable, power hungry, fragile and they easily lose communication. Intrinsically, any irrigation system based on WSNs requires close monitoring to guard against any horrendous mishaps. However, monitoring such an irrigation system which, usually, is located at a rural site can be expensive and time consuming. In this paper we developed an efficient, cost-effective and real-time wireless based remote monitoring mechanism for a WSN based Irrigation system situated in Manja Township within Blantyre city. The system archives data that include the soil moisture potential, link performance, electrical power levels, and valve status and subsequently sends the information as a text message over a cellular network to a remote monitoring site located at the Malawi Polytechnic. The remote station has a broadband wireless dongle which is interfaced to a MYSQL database via an open source FrontlineSMS. The information is graphically published on a web browser with the help of a PHP script. Preliminary results demonstrate that a WSN based irrigation system can be monitored remotely at a low cost and in real time. Keywords—Wireless Sensor Networks, FrontlineSMS, irrigation systems, text message, remote monitoring I. INTRODUCTION According to the 2008 Population and housing census conducted by the Government of Malawi [1], Malawi has a population of 13 million of which 83% heavily rely on agriculture for their livelihood and live in rural areas. Although Malawi is well endowed with several surface lakes, many areas are chronically water deficient especially during dry seasons and drought periods because of sporadic rainfall and high levels of evapotranspiration associated with the impact of climate change and high temperatures of the sub-Saharan region. Sporadic rainfall as a consequence of climate change, the continued population growth and the consequential chronic food shortages have compelled researchers to explore ways of implementing sustainable water management strategies that emphasize the water use efficiency and conservation. The most effective strategy is to improve irrigation efficiency. With the recent advent of effective Soil-Moisture Sensors, numerous ‘smart’ irrigation controllers have been designed based on feedback from soil moisture status. These controllers can be wired or wireless, advanced or basic, and real-time or historic. Regardless of what
form they may be, the objective is to improve crop water use. However, since crop water requirements vary from time to time and region to region due to the variations in climate and properties of the soil, the Irrigation Systems (ISs) must be flexible enough to adapt the environment in which they are deployed. One approach is to use Wireless Sensors Networks (WSNs). The ISs that are based on WSNs are considered to be efficient, cost-effective and flexible enough to adapt any environment [2], [3]. However, WSNs are still under development stage; as such, they are at times unreliable, fragile, power hungry and can easily lose communication [2] especially when they are deployed in a harsh environment. It is therefore necessary to remotely monitor the status of WSN mostly when the ISs are located at a rural site where frequent physical visits may become inevitable but costly and time consuming. The remote monitoring for ISs helps in (1) accessing status of irrigation valves in real-time so that if, for example, the system fails to terminate irrigation, then the personnel should rush to the field and rectify the problem; (2) identifying wireless link failures between sensor nodes in the field in order to quickly fix the fault and have a more robust IS; and (3) accessing the level of batteries for the solar powered sensor nodes to guard against total system failure in case of extended solar power absence. Therefore, in this paper we developed an efficient, cost-effective and real-time wireless based remote monitoring mechanism for a WSN based Irrigation system situated in Manja Township within Blantyre city, Malawi. The remainder of the paper is organized as follows: section II reviews literature for the remote monitoring in ISs; section III discusses the methodology used; section IV presents the results and discussion thereof; and finally section V concludes the paper and provides the recommendations and possible future work. II. LITERATURE REVIEW Attempts have been made in the recent past to introduce remote monitoring in ISs using various strategies ranging from the use of satellite to Wi-Fi connectivity. However, these strategies are expensive and hence not suitable for the developing world. WSNs have a great potential to make remote monitoring easy and less expensive than Programmable Logic Controller based ISs. Towards that end Chavez et al. [4] developed a remote irrigation monitoring and control system for continuous move systems. In their approach, they used a Single Board Computer (SBC) employing Linux operating system to control solenoids connected to individual or groups of nozzles based on prescribed application maps. A main control box housed the SBC connected to a sensor network radio, a Global Positioning System (GPS) unit, and an Ethernet radio thereby creating a wireless connection to a remote server. Furthermore, a C-software control program resided on the SBC to control the on/off time for each nozzle group using a ‘‘time on’’ application map developed remotely. The SBC then automatically populated a remote database on the server in real time and provided software applications to monitor and control the irrigation system through the Internet. However, the system is limited to line of sight transmission between the server and Ethernet radio in the main control box. Additionally, the remote server is limited to short distance from the monitored station. Besides, the use of Wireless Ethernet Bridge (WEB) complicates and hence makes the monitoring system expensive as it requires the SBC. The use of General Packet Radio Service (GPRS) module on top of the sensor node simplifies the monitoring system thereby making it less expensive. Furthermore, the sending of data via Short Message Service (SMS) can provide a non line of sight and unlimited distance option for a monitoring system located in polydisperse points and cross boundary.
On the other hand, Kalpana et al. [5] deployed a WSN for remote monitoring of crop field. They designed and implemented a WSN that could monitor the air temperature, humidity, light intensity both from a crop field and remote places. Their system comprised nodes equipped with small size application specific sensors and radio frequency modules. The sensor data was transmitted via radio frequency link to the centrally localized computer terminal for data logging and analysis. However, just like [4], they employed a computer on site for logging of data and analysis which is costly. However, this can effectively be replaced by a GPRS enabled sensor node. The objective of this paper is to design an efficient, cost-effective and real-time wireless based remote monitoring mechanism for a WSN based IS. The aim is to demonstrate how WSNs can be employed in remote monitoring of ISs at a low cost so that it can be used by small scale farmers in the developing world. III.
METHODOLOGY
An experimental irrigation field was set-up in Manja Township in the city of Blantyre, Malawi. Wireless sensor nodes were deployed to capture the level of moisture in the root zone of maize crop and accordingly the IS had to switch ON solenoid valves to irrigate the field based on the irrigation scheduling strategy. Sensor data which included soil humidity, battery level and status of solenoid valves, was archived at intervals of 30 minutes and sent to a remote server via a cellular network. The purpose of the remote monitoring system was to report sensor data and any mishaps of the IS to a remote server located at the Malawi Polytechnic in the city of Blantyre where the irrigation management personnel was based. The major components used to achieve this objective are discussed in the system architecture below. Later in this section, data collection and analysis tools will be presented. A. System Architecture The system had four levels as depicted in figure 1: (1) wireless sensor node level; (2) coordinator and actuator node level; (3) gateway level; and (4) remote monitoring station. Levels one through three were all being powered by solar energy which made the system suitable for deployment in rural remote areas where there is limited grid power. The description of each of these components is presented next.
Fig. 1 System architecture
1) Wireless Sensor Node Level: The wireless sensor node used in this experiment was a Waspmote node. It has a built-in microcontroller and can integrate various physical parameter sensors. The sensed data is sent wirelessly to a centrally localized processing node (coordinator) with the help of XBee transceiver which resides on top of the sensor node. The nodes are then able to create a wireless network and can transfer information between each other. Four sensor nodes were deployed in the field to report the moisture status to the coordinator node described next. 2) Coordinator and Actuator Node Level: Just like the sensor node described above, the coordinator and actuator node used was a Waspmote node. The two main functions of this node were to (i) aggregate data from all four sensor nodes deployed in the field; and (ii) actuate solenoid valves based on the data received and the irrigation scheduling strategy. To achieve these functional requirements, a software programme was embedded in this node. In addition, this node was used to archive battery levels of all the nodes, wireless link performance between nodes, and status of the valves. This information was very vital to the remote management personnel and was being sent to the gateway every 30 minutes. 3) Gateway Level: In addition to carrying out sensor node duties as discussed in (1) above, the gateway level node was responsible for communicating with the cellular network by sending SMSs to a remote monitoring station. Specifically, the gateway node was equipped with a GPRS module in addition to the XBee module which was used to receive SMS data from the coordinator node every 30 minutes. If there were any mishaps in the system, for example system failing to terminate irrigation or battery level of any of the nodes going low or failure of network links, then instead of reporting to the remote server the gateway node was sending a specific alarm message to the mobile number for the irrigation management personnel. In this way, the personnel is alerted in real-time and can quickly go to the field and rectify the problem. 4) Remote Monitoring Station: Figure 2 is a conceptual model of the remote station depicting how data emanating from the Broadband dongle was processed and analysed graphically. Firstly, the data from the IS was received directly by the broadband dongle housed in the remote monitoring station. FrontlineSMS interface then delegated the data storing capabilities from the dongle to a MYSQL database.
Fig. 2 Remote monitoring station
Notably, not all the characters in the raw data stored in MYSQL database as depicted in figure 3 were in the right format and syntax. As such, a PHP script was prompted to create a new database where processed data was stored ready to be graphed. A separate PHP script was then interfaced with PHPlot library to plot the data in the new database.
Fig. 3 MYSQL database snapshot
B. Data Collection and Analysis The data used in this study was collected using FrontlineSMS, a free open source software used to distribute and collect information via text messages. It enables users to connect a range of mobile devices to a computer to send and receive SMS text messages and works without an internet connection by connecting a device such as a cell phone or GSM modem with a local phone number. A GPRS module was connected to the sensor node to send SMSs to a specific mobile number connected to a remote mobile broadband wireless dongle. The FrontlineSMS then collected data that was sent to a local MYSQL database. MYSQL was installed using Windows Apache MYSQL Package (WAMP) server, a Windows web development environment that allows creation of web applications with Apache2, Practical Extraction and Report Language (PHP) and a MySQL database. With the integration of PHP and HyperText Markup Language (HTML), dynamic pages were created and graphical analysis was easily achieved using PHPLOT. PHPlot is a graph library for dynamic scientific, business, and stock-market charts and creating of pie charts, bar graphs, line graphs, point graphs, etc. from a simple PHP application. The web pages were automatically refreshed at some predetermined time and each time a refresh was initiated, PHPLOT library crosschecked any new data in the MYSQL database before plotting the data using PHP scripts. The data was thus monitored in real-time as will be outline in next section and any mishap was quickly noted and acted upon. IV. RESULTS AND DISCUSSION In this study various IS metrics were monitored to assess the WSN performance, particularly battery level, WSN received signal level, solenoid valves status, soil moisture potential, and soil temperature. Furthermore, system failure indicators were promptly reported to the management personnel. A. Battery Level Firstly the WSN performance was assessed in terms of battery level of the sensor nodes against time. Figure 4 shows the results obtained from the PHPlot.
Fig. 4 Battery level against time
The results show that the battery level is high during day time when there is solar power to charge the batteries, but it is depleted at night as generally expected. The graphs show that on two instances the coordinator’s battery was critically depleted which called for the intervention of the management personnel. It was cushy for the management personnel to scrutinize this mishap in real time through the developed remote monitoring system. B. WSN Received Signal Level Secondly, the WSN performance in terms of received signal strength was monitored remotely. Figure 5 shows the graph of the received signal strength against time as displayed on the PHPlot interface. The results show that the system can remotely monitor the status of the network performance in terms of RSSI. Since these graphs are displayed on the web browser, it is easy to monitor status of the network from any location on the globe as long as one is under the coverage of a cellular network.
Fig. 5 Received Signal Strength against time
C. Solenoid Valves Status Thirdly, the status of the irrigation valves was monitored. Figure 6 shows the respective graphs sampled from the PHPlot. The results depict the state of the IS in regards to irrigation or non irrigation mode. From the Figure, a HIGH (1) indicates that the valve was open thereby allowing the IS to irrigate, and on the other hand, a LOW (0) denotes that the valve was closed thereby putting the IS in an idle mode. It can be observed that the IS was irrigating based on the state of soil moisture and not on predetermined time intervals.
Fig. 6 Valve status against time
D. Soil Moisture Potential Fourthly, the soil moisture potential was also monitored remotely to determine the condition of the field and consequently observe whether the valves’ status discussed in the previous subsection corresponded accordingly. Figure 7 shows the resulting graphs drawn from the PHPlot library interface. The results show the soil moisture potential for the four nodes plotted against time. Examining the moisture potential together with the status of the solenoids, the results show that the field was irrigated based on the condition of the soil. This suggested that we could optimise the use of water commensurate with the predetermined requirements. In other words, the moisture potential is essential for the determination of whether the valves ought to close for irrigation to take place or not. The remote monitoring personnel would be able to note the trend of the moisture potential and thereby establishing how it is being depleted by the crops and replenished by the IS.
Fig. 7 Soil moisture potential against time
V. CONCLUSION AND FUTURE WORK In this paper we presented the experimental results for the remote monitoring of an IS based on WSN. The results have shown that several performance metrics can be monitored cost effectively using a GPRS enabled WSN node and using freely available open source tools for instance FrontlineSMS, MYSQL, PHP and PHPlot. Furthermore, we outlined how this remote monitoring can be attained in real-time. The use of cellular network enhances the remote monitoring at low cost since an SMS charge is extremely low and the cellular network coverage is broad.
This remote monitoring is not only limited to irrigation systems, but can also be an anchor for other systems that deploy WSNs such as solar power systems, machine health, landslide detection, structural, water/wastewater, greenhouse, and forest fire detection. However, for unreliable cellular networks, this technique can prove futile for critical applications like healthcare due to the delays or complete failure of SMSs. We therefore recommend users of such applications to pursue other alternatives like dedicated telephone lines or Wi-Fi. The future work proposed in this study is to consider large scale deployment of WSNs and to include reporting of more performance parameters such as sensor node temperature, level of water in the tank and powering requirements for the solenoid valves and latching circuits. ACKNOWLEDGMENT The authors would like to thank the Community Rural Electrification and Development (CRED) project funded by the Scottish Government through the University of Strathclyde (UoS) for providing the equipment deployed in this projected in the city of Blantyre, Malawi. REFERENCES [1] GoM (Government of Malawi). (2008). “2008 Population and Housing Census”. Ministry of Economic Planning and Development. National Statistical Office, Zomba. [2] J. Balendonck, J. Hemming, B.V. Tuijl, L. Incrocci, A. Pardossi, & P. Marzialetti, (2008). “Sensors and Wireless Sensor Networks for Irrigation Management under Deficit Conditions (FLOW-AID)”. Retrieved November 2, 2010, from FLOW - AID: http://www.flowaid.wur.nl/NR/rdonlyres/DA8B2ECC-9EEE-4A99-A1BAE0F7EDCEB087/73978/2008BalendonckFLOWAIDAgEng.pdf [3] S. Fazackerley, & R. Lawrence (2010). “Reducing turfgrass water consumption using sensor nodes and an adaptive irrigation controller”. Sensors Applications Symposium (SAS), 23-25 Feb 2010 IEEE, (pp. 90-94). doi: 10.1109/SAS.2010.5439386. [4] J.L.Cha´vez, F.J. Pierce, T.V. Elliott, & R.G. Evans,”A Remote Irrigation Monitoring and Control System for continuous move systems. Part A:description and development,” Retrieved April 1, 2012, from United States Department of Agriculture: http://ddr.nal.usda.gov/bitstream/10113/41436/1/IND44310127.pdf [5] S. Kalpana, J.R Chander, & S.M Ahmed, “Wireless Sensor Network for Remote Monitoring of Crop Field,” IJAEST – International Journal Of Advanced Engineering Sciences And Technologies, Vol No. 8, Issue No. 2, pp 210 – 213, 2011.
