Temperature and Illuminosity Management to monitor greenhouse conditions using SunSPOTs
Suthikshn Kumar
Dept of Information Science, PESIT Bangalore suthikshn.kumar@pes.edu

Ishaan Raghunandan
Dept of Information Science, PESIT Bangalore ishaan.raghunandan@gmail.com
Abstract: In modern greenhouses, several measurement points are required to trace down the local climate parameters in different parts of the enclosure efficiently.A Wireless Sensor Network (WSN) consisting of small-size wireless sensor nodes an attractive and cost-efficient option to build the required measurement system. The most important climatic conditions are the temperature and the light intensity within the greenhouse.This paper describes the implementation and configuration of the wireless sensor network using the Sun SPOT platform.It also uses a Time Series Algorithm to predict the temperature variation. Keywords: WSN,SunSPOT,TimeSeries,Java,Zigbee,Back Propagation 1.Introduction The most important factors for the quality and productivity of plant growth are temperature and light. Continuous monitoring of these environmental variables gives information to the agriculturalist to better understand, how each factor affects growth and how to manage maximal crop productiveness. The optimal greenhouse climate adjustment can enable us to improve productivity and to achieve remarkable energy savings - especially during the winter in northern countries. Increased number of measurement points should not dramatically increase the automation system cost. It should also be possible to easily change the location of the measurement points according to the particular needs, which depend on the specific plant, on the possible changes in the external weather or

Garima Srivastava
Dept of Information Science,PESIT garima.clairvoyant@gmail.com greenhouse structure and on the plant placement in the greenhouse. Wireless sensor network (WSN) can form a useful part of the automation system architecture in modern greenhouses. Wireless communication can be used to collect the measurements and to communicate between the centralized control and the actuators located to the different parts of the greenhouse. In advanced WSN solutions, some parts of the control system itself can also be implemented in a distributed manner to the network such that local control loops can be formed. Compared to the cabled systems, the installation of WSN is fast, cheap and easy. Moreover, it is easy to relocate the measurement points when needed by just moving sensor nodes from one location to another within a communication range of the coordinator device. If the greenhouse flora is high and dense, the small and light weight nodes can even be hanged up to the plants’ branches. WSN maintenance is also relatively cheap and easy. The only additional costs occur when the sensor nodes run out of batteries and the batteries need to be charged or replaced, but the lifespan of the battery can be several years if an efficient power saving algorithm is applied. This paper utilises SunSPOT i.e Sun Microsystem's Small Programmable Object Technology. The objectives of this paper are the implementation of the temperature and illuminosity monitoring system with the following modules :

2.

SunSPOT Architecture :

Sun Microsystems developed wireless sensor network (WSN) mote called SunSPOTs. SunSPOTs are built upon the IEEE 802.15.4 standard these are MIDP compliant devices which run Squawk Java Virtual Machine.The SunSPOT Device is a small, wireless, battery powered experimental platform. It is programmed almost entirely in Java, to allow regular programmers to create projects that used to require specialized embedded system development skills. The hardware platform includes a range of onboard built-in sensors as well as the ability to interface to external devices. This device comes with three sensors namely: accelerometer, light and temperature sensor. MIDLets are installed on these devices because of their MIDP compliant nature. There is a specific base station which is used to connect with the host computer and then through host applications remote SunSPOTs are used.

runs the host application. In this model, the host application has its own system-generated address, distinct from that of the base-station. Base-station accepts the multicast request in this mode. The main advantage of shared mode is that more than one host application can use the same base-station simultaneously. On the other hand in dedicated mode, it runs within the same JVM as the host application and only host application can use it. In this model, the host’s address is that of the basestation. 3. Implementation :

Figure : SunSPOT Architecture The physical arrangement of the basic connection arrangement between host computer and sensor nodes is showed in the figure below

This paper implements the system in the following four modules : Data collection : In this module the system gathers the information from various sensor nodes by opening a Radiogram connection.The communication between the base station and the SPOT deployed in the system is through OTA command Server. The base station is enabled to be a shared basestation so it can be identified by multiple SPOTs in order to transmit data from different locations. Each SPOT has a SPOT program running in which the sensor readings are stored in variables and then these values are packed together in a datagram and sent over a particular port. Data Analysis : In this module the collected data is analysed by storing the values recorded by the sensors into a database using a jdbc driver and MySql.The analysis module also runs three basic queries of Min_Val,Max_Val and Avg_Val where in the data is summarised in a particular way. This paper particularly aims to implement the Time Series Algorithm in order to predict the temperature values that are stored in the database. Time series analysis comprises methods for analyzing time series data in order to extract meaningful statistics and other characteristics of the data. Time series forecasting is the use of a model to forecast future events based on known past events to predict data points before they are measured Data Access and Control : In this module primarily the user is allowed to set various temperature and light value ranges

figure : Software Stack SunSPOT The host application can be J2SE program but target sensor nodes application need to be J2ME MIDLet for the Squawk VM. There are two modes for base station to run, shared mode or dedicated mode. In shared mode, two JVMs are launched on the host computer: one manages the base-station and another

for different plants so as to easily set the greenhouse to control the conditions. Whenever the set value ranges are exceeded or the values fall below it the system issues a warning condition automatically in order to notify the user.Another part of this module is used to give remote access to the conditions within the greenhouse. The application developed posts the values at a certain interval of time to a website that is designated for it. As a future enhancement this information can be periodically emailed or sent as an sms to the user Time Series Analysis Algorithm: In this module neural network model training of the network is performed using the Back-propagation algorithm that uses supervised learning. During the training process, the network adaptively changes its synaptic weights so as to reduce the total system error within specified tolerance. When such a condition is achieved, the synaptic weights of the network can be used to produce the outputs from those input features for which no training has been done. The neural networks trained to a satisfactory level with proper features and architectures perform better in prediction, pattern classification, feature extraction etc. (Steppe et al. 1996). The generalization of the network, which means producing appropriate outputs for those input samples not encountered during training process of the network is best described by training data size and number of synaptic weights

adding a ratio of it from the weight. This ratio impacts on the speed and quality of learning; it is called the learning rate. The sign of the gradient of a weight designates where the error is increasing; this is why the weight must be updated in the opposite direction. The phase 1 and 2 is repeated until the performance of the network is satisfactory. B. Modes of Learning There are essentially two modes of learning to choose from, one is on-line learning and the other is batch learning. Each propagation is followed immediately by a weight update in online learning.In batch learning, much propagation occurs before weight updating occurs. Batch learning needs more memory capacity, but on-line learning requires more updates. C. Algorithm Actual algorithm for a 3-layer network (only one hidden layer): 1. Initialize the weights in the network (often randomly) 2. Do 3. For each example e in the training set O = neural-net-output (network, e) ; forward pass T = teacher output for e 4. Calculate error (T - O) at the output units 5. Compute delta_wh for all weights from hidden layer to output layer ; backward pass 6. Compute delta_wi for all weights from input layer to hidden layer ; backward pass continued 7. Update the weights in the network 8. Until all examples classified correctly or stopping criterion satisfied 9. Return the networ

Back Propagation Algorithm A. Phases in Backpropagation Technique The back propagation learning algorithm can be divided into two phases: propagation and weight update. Phase 1: Propagation Each propagation involves the following steps: 1. Forward propagation of a training pattern's input is given through the neural network in order to generate the propagation's output activations. 2. Back propagation of the output activations propagation through the neural network using the training pattern's target in order to generate the deltas of all output and hidden neurons. Phase 2: Weight Update For each weight-synapse: 1. Multiply its input activation and output delta to get the gradient of the weight. 2. Bring the weight in the direction of the gradient by

4. Future Enhancements : The control module can be strengthened by interfacing control devices such as irrigational system to control moisture and light sources to control the illuminosity can be added. Also a detailed dependency of temperature on the moisture,humidity and light conditions can be manipulated to regulate the temperature within. The ventilation in the greenhouse could

also be included and controlled by this system by opening and closing various vents. The warning system could also be enhanced in order to sound alarms for uncontrollable conditions.A module for image processing could also be programmed wherein the live streaming of the conditions in the greenhouse could be reviewed by the user in order to make a better controlled decision. In addition to the above funcationalities this system could be integrated to the internet so that the control and the monitoring could be done via remote access. 5. Simulation Results :

The following graph shows the result of the output DataSet.

The moving average and the exponential moving average is calculated for the dynamic database. The graph below show the variation of MA and EMA over certain number of days labelled as indexes.

6.

Conclusion

The follwing figure graph shows the trained input dataset and its variations.

The WSN-based controller has allowed a considerable decrease in the number of changes in the control action and made possible a study of the compromise between quantity of transmission and control performance. The limit of the level crossing sampling has presented a great influence on the event based control performance where, for the greenhouse climate control problem, the system has provided promising results. The sensor readings being recorded for the temperature and illuminosity are tested and verified upto 98% accuracy. The concepts of multithreading and exception handling is effectively implemented. All the modules run independently without creating any interruption in the other process. The system is also efficient in terms of time variable. The communication of the datapacket is with very little delay. This monitoring system can be deployed in various application scenarios such as sports,medicinal procedures,chemical processes that are carried out in the industry. It also has a huge application in photography where the light intensity is an important characterstic to a photographer. The system is easily deployable. It can be implemented and used with ease and convinience.

7. REFERENCES [1] Sun™ Small Programmable Object Technology (Sun SPOT) Owner’s Manual Release 3.0, Sun Microsystems, Inc. 2007 [2] Sun Spot Developer’s Guide. Sun Microsystems. 2005 [3] Demo Sensor Board Library. Sun Microsystems. 2005 [4] Sergio Scaglia, „The Embedded Internet”, Addison-Wesley 2008 [5] James Gosling, „ The Java™ Language Specification Third Edition”, Addison-Wesley 2005 [6] Matijevics István, Simon János, „Advantages of Remote Greenhouse Laboratory for Distant Monitoring”, Proceedings of the Conference ICoSTAF 2008, pp 1-5, Szeged, Hungary, 2008 [7] R. Rojas: Neural Networks, Springer-Verlag, Berlin, 1996. [8] Y.Radhika and M.Shashi,” Atmospheric Temperature Prediction using Support Vector Machines,” International Journal of Computer Theory and Engineering, Vol. 1, No. 1, April 2009 1793-8201.

[9] M. Ali Akcayol, Can Cinar, Artificial neural network based modeling of heated catalytic converter performance, Applied Thermal Engineering 25 (2005) 2341-2350. [10] Mohsen Hayati, and Zahra Mohebi,” Application of Artificial Neural Networks for Temperature Forecasting,” World Academy of Science, Engineering and Technology 28 2007. [11] Brian A. Smith, Ronald W. McClendon, and Gerrit Hoogenboom,” Improving Air Temperature Prediction with Artificial Neural Networks” International Journal of Computational Intelligence 3;3 2007. [12] Arvind Sharma, Prof. Manish Manoria,” A Weather Forecasting System using concept of Soft Computing,” pp.12-20 (2006)