technical aspects

Greenhouse Asset Management Using Wireless Sensor-Actor Networks V. Lakshmi Narasimhan1, Alex A. Arvind2, and Ken Bever3

Department of Computer Science, 1University of Western Kentucky, Bowling Green, KY 42101, USA 2University of Northern British Columbia, Prince George, BC, Canada, 3Chief Technical Officer,

MIMOSA, Emails: [email protected], [email protected] Abstract: Greenhouse plays an increasingly important role in modern horticulture in order to meet the needs of the world’s growing and demand driven economy. The primary issue of greenhouse based horticulture is to manage the greenhouse environment optimally in order to comply with the economic and environmental requirements. This paper discusses the advantages of using asset management strategy along wireless sensor-actor network technology for such cost-effective and environmental friendly greenhouse management. Keywords: Greenhouse management, asset management, sensor-actor networks. 1. INTRODUCTION An agricultural greenhouse can be considered as a man-made system to emulate a suitable ecosystem in order to grow crops. Initially, greenhouse technology was driven mainly by the desire of the rich people for out-of-season fruits and flowers [18]. Later, due to widespread industrialization, commercial horticulture began to develop all over the world starting with Europe and the United States [18]. In modern horticulture, greenhouses play an increasingly important role to meet the demand-driven economy. The primary issue of greenhouse based horticulture is to manage the greenhouse environment optimally in order to comply with the economic and environmental requirements [17]. Although technological advancements offer innovative solutions for specific issues, achieving optimal management of overall greenhouse is generally difficult. The latter involves making complex decisions on investments, crop production processes and controlling the dynamically changing environmental conditions. Better decisions can often be made based on sufficient knowledge and insight of the system of study [2]. A greenhouse asset management system can provide knowledge and insight required to operate the greenhouse in an optimal level. Efficient control of the greenhouse environment requires a control system which is adaptive, accurate, and cost effective. Such a control system can be built using wireless sensor-actor network technology [19], which offers local processing and storage

capabilities, wireless communication, and comes in highly portable sizes. Greenhouse management often involves continuous monitoring and activation of different units, such as heating, cooling, lighting, water/soil gradient, etc., in order to maintain suitable climate for the plants. These activities sometimes require human presence in the greenhouse in order to observe the condition and activate appropriate units that may be undesirable for many reasons. With the advancement of computing and wireless sensor- actuator systems, it is quite possible to perform such tasks anywhere, any time [7]. This paper discusses the advantages of using asset management strategy and architecture along wireless sensor-actor network technology for effective greenhouse management. Particularly, the paper illustrates the benefits of deploying a sensor- actor network along with an integrated asset management system in a greenhouse environment. We illustrate 1) how a sensor network can facilitate greenhouse management, 2) how it can be integrated with the Internet for remote monitoring and operation, 3) how it can be enhanced with a database and software tools to exploit the collected data for effective management of greenhouse and 4) how capacity building (financial management, scalability, etc) in greenhouse environment can be effected. The rest of the paper is organized as follows: section 2 provides an overview of a typical greenhouse and their operational requirements. A brief description of the common conceptual object model is provided in section 3, followed by the development of a CCOM model for greenhouse sensor network in section 4. A typical Greenhouse sensor network information management architecture is described in section 5, followed by a case study in section 6. The conclusion summarizes the paper and offers pointers for further work in this area. 2. OVERVIEW OF THE GREENHOUSE ASSET ENVIRONMENT The main objective of an agricultural greenhouse is to have maximum productivity with minimum use of energy and other resources in order to be economical and environmentally friendly.

International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies

0-7695-2993-3/07 $25.00 © 2007 IEEE DOI

9

International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies

0-7695-2993-3/07 $25.00 © 2007 IEEE DOI 10.1109/UBICOMM.2007.43

9

From the cost perspective, greenhouse management involves primarily three types of costs: (i) initial infrastructure cost (setup cost) and (ii) material cost during crop cultivation (seed, water, fertilization, etc.), and (iii) climate and pest control cost. The later two costs can be collectively referred to as the operating cost and most of the present day investments are planned based on the operating cost [20]. Setup cost is one-time or long-term investment, whose value has direct effect on the operating cost. Material costs are generally fixed and low compared to the total cost. For example, in most cases, the cost of fertilization in greenhouse is between 1 to 2% of the gross value of product compared 10 to 20% for field crops [18]. Thus the major part of the operating cost goes to climate and pest control. Both setup cost and operating cost involves human cost and modern trend is to reduce the human cost by the way of computerized automation [20]. The objective of climate control is to protect the crops from external climate that is harmful to the crops and to provide the climate that is favorable for the growth of crops. Pest control system protects the crops from pests (insects, mites, diseases, weeds, and vertebrates) [24]. Since the artificial environment created by the greenhouse often favors the rapid development of pests and weeds [18], climate control and pest control need to be integrated suitably in order to achieve optimal control of both. Greenhouse asset includes the following: Humans who work on the greenhouse Greenhouse land and building Input materials such as seeds, fertilizers, water,

pesticides, biocontrol agents (predatory mites, pirate bugs, soil-dwelling mites, and parasitic insects), etc.

Control systems such as climate control and pest control

o Hardware – sensors, actors, connectors, interface boards, input and display panels, routers, computers, generators, transformers, etc.

o Software – communication, data filter and fusion, etc.

Crops that grow inside the greenhouse Output products such as, flowers, fruits,

vegetables, etc., and Data and information management system.

3. GREENHOUSE CONTROL SYSTEMS

Any control system has two main set of components, namely, a sensor or monitoring component to collect data and an actuator component to perform specific actions as a response to the collected data (or to execute a predetermined strategy). To understand the sensors and actuator components used for greenhouse climate control system, we briefly review the ecosystem requirement for plant growth. Photosynthesis is the most important factor in plant growth and it requires light, water, and carbon dioxide [18]. So, high transmission of light in the waveband from 400nm to 700nm is essential to maximize the photosynthesis rate. The amount of such radiation entering the greenhouse depends on the shape, structure, and material of the house and, the amount of absorption depends on the distribution and orientation of the leaves of the plants. Apart from natural daylight, artificial light is increasingly used to supplement or substitute prolonged daylights in greenhouses. So, lighting system is an important component in greenhouse climate control. Supplies in terms of water, fertilizer and any other nutrients have also been considered as parts of the greenhouse climate control system. Fertilizers typically provide, in varying proportions, plant nutrients such as nitrogen, phosphorus, and potassium, calcium, sulfur, magnesium, etc. Maintaining suitable temperature is another important factor in plant growth. Many factors could influence the temperature inside a greenhouse, such as, exterior temperature, sunlight, wind speed and its direction, etc. In order to maintain the required temperature, both heating and cooling systems are required. Cooling system may be employed through ventilation or, fan, and heating system may be implemented in terms of roof heating, fan heating, metal, pipe, water, steam, or air heating, decomposition of manure and compost, etc. Thus, cooling system and heating system are essential components of a climate control system. Water and temperature will cause change humidity within a greenhouse and it is necessary to maintain suitable relative humidity, which measures the amount of water vapor a gas is currently being held in comparison with the amount it can hold at a given temperature. The third component involved in photosynthesis is CO2 and it can be obtained by decomposition of organic matter, natural gas, propane, kerosene, respiration of human being, or pure CO2 from

1010

chemical industries. Since the late 60s, most greenhouse growers use either liquid compressed CO2 or CO2 produced by burning of liquid fuel or gas [18]; hence, carbon dioxide generator is another component in the climate control system. The pest control system also has sensors to monitor the crops for pest identification and eradication. Once a pest is identified, the pest control system is activated to deal with it. Many techniques are being used for pest control, which include, sanitation, screening, biological control, and chemical control [24]. Biological control techniques use living organisms to reduce the incidence of pest organisms. Chemical control techniques use chemicals to kill or disrupt their normal development. Summarizing, a typical greenhouse climate control system has the following subsystems: Lighting system, Cooling system, Heating system, Carbon dioxide generation system, Watering system and Fertilization system. These systems will be activated based mainly on the temperature, relative humidity, light, wind speed and direction, and water level inside the greenhouse. The devices used to measure these quantities are referred as sensors. Given that greenhouses are considerably large in size and that a federation of greenhouses are possible, a network of sensors are therefore essential to monitor and govern a greenhouse management system. 4. SENSOR-ACTOR NETWORK BASED GREENHOUSE CONTROL SYSTEMS Wireless sensor-actor or sensor-actuator network is a collection of sensor and actor nodes linked by a wireless medium to perform distributed sensing and acting tasks [19]. The sensor nodes collect typical information about the physical world (e.g., humidity and temperature) and communicate over a network environment to a computer system, which is called, a base station. Based on the information collected, the base station takes the decisions (in some cases the actors themselves take decisions) and then the actors perform appropriate actions upon the environment. This action allows users to sense and control the environment from anywhere, at any time. Further, the networked sensor information typically provides a better overall perspective of the environment than a single sensor- based source of information. With the advent of wireless technology, such sensors and actors can be connected over a wireless network system (e.g., Bluetooth, WiFiWide).

Wireless sensor-actor network has many advantageous over conventional control systems, which include the following: One of the major limitations of the current

automation system for greenhouse management is the amount of wiring involved in connecting sensors to actuating systems (actors) [21]. Wireless sensor-actor network technology eliminates this problem.

Many existing control systems for greenhouse maintain the environment at known set-points. Wireless sensor-actor based control system for greenhouse is amenable to have more flexible and adaptive control over the environment due to its communication, computing, controlling, and storage capabilities.

As the connection between effective maintenance management techniques and significant improvements in efficiency and profitability becomes evident, system health monitoring and prognostics are getting increased attention in modern enterprises [22]. Effective system health monitoring and prognostics requires critical data of the system environment. A suitably designed wireless sensor-actor network can collect and provide such data in a timely manner.

Since wireless sensor-actor systems are designed to accommodate addition and deletion of nodes with minimal effort, changing the system in accordance with the changes in demand is relatively easy to implement compared to conventional systems.

By allowing flexible coordination, wireless sensor-actor based control systems offer system-wide scalability at low cost.

The above advantages favor the use of wireless sensor-actor based system design for greenhouse control systems. A typical greenhouse contains several sensors for measuring humidity, temperature, pressure, carbon dioxide, light, motion, etc., and contains several actors, such as, air conditioners, floodlights, sprinkler and water management facility, plant decks, air blowers, fans, CO2 production units, etc. As shown in figure 1, these sensors and actors can communicate among themselves and to the base station, through wireless media. The base station is then connected to the Internet so that it can be accessed from anywhere, at any time. A test bed for monitoring the sensor readings from a remote location for conducting experiments on the growth

1111

rate of tobacco plants have been reported in [23]. If the base station is located farther away from the greenhouse, then small systems such as, Stargate nodes along with RFID tags [12] can be used as routers to communicate to the base station. The base station usually maintains the database and can also act as a web server so that the data can be entered and/or visualized from remote locations using handheld devices. A typical architecture of a wireless sensor-actor network based greenhouse climate control is shown in Figure 1. The placement of the sensor and actor nodes plays a crucial role in the design of the network. For example, different sensors in a given greenhouse can report different temperature at any time depending on their locations and, consequently, the location and number of sensor and actor nodes are to be determined by this requirement. The numbers and locations can influence the quality of sensor readings. Although there are techniques available to extract the true values from the observed values, the task of data collection and management becomes challenging when the participating nodes (either sensor or actuator) start to deviate from normal operation. As a consequence, an unusual sensor reading could trigger a system-wide diagnostic process, which can typically use historical data in order to identify the system status. A typical sensor- actor based control system operation is shown in Figure 2. The collection of sensor data, called sensor scheduling, involves such issues as time to collect data, duration of collection, and their accuracy, all of which depend on operational and economic requirements. 5. GREENHOUSE ASSET MANAGEMENT SYSTEM Asset management encompasses a set of processes, tools, performance measures, and related decision support system in order to establish an optimum way of manage assets and achieve the desired sustainable outcome [25]. In other words, asset management concerns the seeking of the best return on investment. This process calls for the maintenance of a detailed registry for all equipments as a common repository (called, the global asset registry) and historical information including operation, repair, and compliance records on the various assets. These information sets are typically held over an information management environment (typically over distributed databases or data warehouses) and they need to be integrated with suitable asset health monitoring, quality

control and risk management systems, before they are embedded into a suitable decision support system. 5.1 A Brief Description of CCOM for Sensor Network Modeling The objective of the Common Conceptual Object Model (CCOM) is to provide a formal representation of a system of interest in order to understand and communicate with it. It is a high level model, which is impendent of the implementation technology and associated constraints and therefore it must use notations that are widely known. Unified Modeling Language (UML) [10, 26] is a popular modeling environment, which is heavily used in software engineering and CCOM is a derivative that is tailored towards asset modeling problems. Briefly, CCOM provides a clean means to describe various entities in an enterprise and their inter-relationships. 5.2 A CCOM Model for Greenhouse Sensor Network Figure 3 describes the CCOM model for a greenhouse sensor network, wherein the term Enterprise is the representation for the Greenhouse as an ensemble entity, while Table 1 describes the various entities in a greenhouse. Figure 4 captures the characteristics of each component and the relationships between them.

Item Description Enterprise Describes the greenhouse as an

ensemble entity GHSite Describes each greenhouse site and

its characteristics GHSiteType Describes a greenhouse site types Asset Describes assets AssetType Describes asset functionalities Sensor Describes sensors SensorType Describes sensor types and their

characteristics SensorLocation Describes sensor location Actor Describes actors ActorType Describes actor types and their

characteristics ActorLocation Describes actor location AssetModel Describes asset models and their

characteristics Asset Manufacturer

Describes asset manufacturers

AssetAgent Describes asset agent (for maintenance at a given GHSite)

GHSiteDB Describes greenhouse database at a given GHSite

Table 1: Description of Entities in a Greenhouse

1212

6. ARCHITECTURE FOR GREENHOUSE INFORMATION MANAGEMENT The sensor network information management architecture is presented in Figure 4, where the Input Section manages data acquisition (including cleaning and formatting of sensor data) and sensor network management (including wireless connectivity management). The Asset Registry contains three modules, namely, a Data warehouse to store cleaned sensor data, a Catalog Service to access data sets and a Data Management service to facilitate federated data warehousing and federated catalog maintenance and management. The architecture is designed to conform with the principles of the Model Driven Architecture [4] and, in addition, in conformance with modern Standards on Open System Architecture (for Condition based Maintenance) [9, 14]. The Information Exploitation Section has several modules containing a number of tools to support decision making [6], report generation, financial status generation and visualization (see [3, 5, 11] for the kinds of information that could generated). The Output Section facilitates query management, query optimization and MIMOSA/ISO complaint data exporting facilities. The Data Acquisition performs Sensor-actor network management, Sensor output, Wireless connectivity & Feature extraction. The Data Storage performs Distributed storage, Metadata management, Classification of data sets using MIMOSA/ISO compliant schemas. The Exploitation Section performs Report generation (Asset registry view), Annotation service (Asset quality view) and Interaction view generation. The Tools and Techniques performs Horticulture data analysis tools, Decision support systems (what-if questions are analyzed here) and Asset financial status generator. The Catalog Service provides (federated) catalog and Metadata repository 7. CONCLUSIONS This paper describes a wireless sensor network environment for managing agricultural greenhouses. A CCOM model of the greenhouse system has been developed. In addition, a four-layer information management architecture containing modules that collects and cleans data, an integrated asset registry,

tools & technologies for information exploitation and output management to handle customers has been detailed in this paper. We are currently working on following up on the MIMOSA Standards [8, 13, 15, 16] and ISO Standards [1] and enhancing them to suit greenhouse information management. The issues to be addressed include, conditional monitoring and facilitation for diagnostics and prognostics of various asset types inside the greenhouse, including plants, machinery and others. We intend to develop operational IT systems, in addition to developing international Standards in these areas. 8. REFERENCES 1. ISO 13374-1:2003, Condition monitoring and

diagnostics of machines – Data processing, communication and presentation – Part 1: General guidelines.

2. Using Mineset for Knowledge Discovery Barry G. Becker IEEE Computer Graphics and Applications, July/August 1997

3. Updates and view maintenance in soft real-time database systems. Ben Kao, K. Y. Lam, Brad Adelberg, Reynold Cheng, Tony Lee. Proceedings of the eighth international conference on Information and knowledge management.

4. Model-Driven Architecture: Vision, Standards and Emerging Technologies. John D. Poole. ECOOP 2001.

5. Benyahia I, Potvin, J-Y. (1998) Decision support for vehicle dispatching using genetic programming. IEEE Transactions on Systems, Man and Cybernetics, Part A. 28(3), 306-314.

6. Guo B, (2003) Knowledge representation and uncertainty management: applying Bayesian belief networks to a safety assessment expert system. Proceedings of the 2003 International Conference on Natural Language Processing and Knowledge Engineering. pp 114-119.

7. Dreyfus L P, Ahire S L, Ebrahimpour M. (2004) The impact of just-in-time implementation and ISO 9000 certification on total quality management. IEEE Transactions on Engineering Management. (51)2, 125-141.

8. Machinery Information Management Open System Alliance, http://www.mimosa.org

9. Open System Architecture for Enterprise Application Integration (OSA-EAI) Version 3.0d, http://www.mimosa.org

10. OSA-EAI Common Conceptual Object Model (CCOM) V3.0 http://www.mimosa.org

11-26 Please get these from the authors

1313

Figure 2: Operation of a Sensor-Actor Network

Receive Data from Sensors

Analyze it (with old Data)

Decide Actions

Send Commands to Actors

Internet Base Station Remote

User Greenhouse Sensor

Actor

Figure 1. A Wireless Sensor-actor Network Based Control System

Asset Registry

Information Exploitation Section

Input Section

Sensor Network

Data Acquisition

Catalog Service

Data Storage (Data Warehouse)

Output Section

Query optimization and Management

Standards (MIMOSA/ISO) Compliant Data Exporter

Visualization Environment

Decision Support Systems

Tools and Techniques

Report Generation

Data Management

Actors

Fig.4: Greenhouse Information Management Architecture

Enterprise EntId UserId

GHSiteType TerrainId GeoId Mobile +/-

GHSite

Asset AssetId

GHType SiteId

AssetType FunctionId

Sensor SensorId SensorAge

SensorType FunctionId Metric

SensorLocation Location (x,y,z) NearestactorRelativity NearestSensorRelativity

AssetModel GHSiteDB entId userId

AssetManufacturer ManId

AssetAgent

Actor ActuatorId ActuatorAge

ActorType FunctionId Metric

ActorLocation Location (x,y,z) NearestSensorRelativity NearestActorRelativity

*

1

*

*

* 1

* *

Figure 3: CCOM model for a green house sensor network

1414

05674751.pdf