1. SAS_EN - A Swarm of Agents for a Sustainable
Environment
Ruvindee Rupasinghe, Sanjaya Ratnayake, Anuruddha Ranatunga, Amila Sajayahan, Asoka S.
Karunananda
Faculty of nformation technology,
University of Moratuwa
Abstract: sustainability. It is evident that environmental
Environment consists of heterogeneous entities sustainability cannot be ensured as individual entities, yet
which are working collaboratively to keep the environment needs corporation from heterogeneous entities/agents
sustainable. The general concept of “environmental which have diverse interests, and physically distributed in
sustainability” refers to the necessary balance between
human wants and needs and the capacity of the natural
much lactation and also operate under different conditions.
systems of the earth. Thus to keep the balance of environment For instance, collaboration among entities that emits
the communication among photosynthesis, co2 emission, carbon dioxide, executes photosynthesis and releases
environmental conditions, nutrient conditions and nutrient oxygen must be communicating each other for ensuring
deficiencies of the plant are important. Nevertheless it has sustainability in the environment. Obviously, if the entity
become a major issue in maintaining those conditions within responsible for emission of Co2 can be communicated with
a controlled environment. the entity for photosynthesis and increases the rate of
Therefore the communication among resource photosynthesis, level of Co2 can be reduced. In order to
entities that are involved in a certain task of the environment
increase the rate of photosynthesis, the plant as an entity
and reaching consensus for protecting and ensuring the
sustainability of a given environment is highly important. requires a higher level of nutrients [10]. As such the
Thus project implements SAS_EN, which collaboratively photosynthesis entity cannot increase it processing without
works for a sustainable environment in a Hydroponics the consent from the plant. Therefore, we identify
Greenhouse environment. environmental sustainability as a distributed problem
The system has been implemented using agent involving many entities/agents. In this sense, any human
technology. There are distinct agents dedicated for each tasks activity such as cultivation, industries, urbanization and
of the environmental sustainability and they collaboratively even housing in the environment must be concerned with
work together to achieve a common goal. Hydroponics, which at least three aspects. Those are space/location utilization,
grows in a controlled environment, has been used for testing
and evaluating the solution. The test results have shown its
environmental resource as the input to the activity and the
potential of using SAS_EN for solving the distributed emission of waste products to the environment.
problems of the environment. Thus the project identifies the environmental
sustainability as a collaboration of many entities in the
I. INTRODUCTION environment and identifies an agent based solution is the
Natural environment has been the infrastructure most appropriate way to maintain the balance of the
for all activities of living beings in the world. As such all environment [9]. It consists of four agents namely plant
living beings occupy and consume environmental agent, ventilation agent, humidity agent and nutrient agent.
resources while releasing the waste products to the same Hydroponics environment have been used as the test
environment [15]. Undoubtedly, environment is the most environment for implementing the solution. The external
precise resource to be effetely utilized and managed. environmental factors and concentration of water medium
Nevertheless when compare with other living is controlled and monitored in order to achieve the
beings, humans release harmful waste products to the maximum yield from the plantation.
environment as a part of their consumption of Rest of the paper is organized as follows. The
environmental resources [1]. In one sense, environment section two depicts the existing approaches of balancing
space has been utilized by the man kind by destroying the the environmental sustainability while section three depicts
natural balance created by the existence of other living the adopted approach of using a multi agent system for the
beings. On the other hand, humans consume limited environmental sustainability. The section four describes
environment resources without allowing regeneration of the design and the section five is about the implementation
such resources through natural cyclic process [11, 14, 6]. approach. The section six is about user scenarios and the
For instance, level of carbon dioxide due to industries can features of the system. The paper finally deals with the
be reduced and pump oxygen in to the environment evaluation and further works.
through process of photosynthesis in plants [16].
However, despite the ever increasing II. EXSISTING APPROACHES
technological advancements in the modern world, the use Currently there are several existing approaches for
of the modern technology for ensuring a sustainable controlling and monitoring the environment. Some of these
environment is very limited. Therefore, SAS_EN intends existing approaches include grow rooms; grow boxes and
to provide a technological solution for environment automation of individual components. Further

2. environmental management systems and environmental source products which attempt to automate a hydroponics
information systems also have been implemented. environment. Hmeter project is one such approach [8].
A. Environmental information systems When consider all these approaches, common thing is that
The high availability and variety of large each of them tries to bring some kind of automation in to
environmental data sets have open new opportunities for the environment. However, all of them follow a reactive
data mining and useful cross-referencing of disparate paradigm, where the system waits till the environment
environmental data sets. Taking the advantage of these change for making any action. Anyway, if the automation
opportunities, environmental information systems will system could act pro-actively, it will be a much efficient
need to operate effectively in a distributed, open system compared to their reactive counterparts. Especially,
environment. The paper describes the New Zealand since the environmental changes are unpredictable and
Distributed Information System (NZDIS) software happens so dynamically, whatever the system which
architecture for environmental information systems. It controls the environment should be intelligent and flexible
optimizes extensibility, openness, and flexible query enough to adopt the system accordingly in quick time.
processing, by using collaborating software agents that However, existing reactive paradigm is not good enough
communicate by means of a standard declarative agent for this purpose.
communication language. The metadata of environmental III. SAS_EN
data sources are stored as part of agent ontologies, which In addressing the distributed problem of environmental
represent information models of the domain of the data sustainability, we propose to use the Modern Agent
repository. The agents and ontological framework are Technology. The intelligent agents represent one of the
designed as much as possible to take advantage of standard promising technologies for building complex software
object-oriented technology, such as CORBA, UML, and systems [7]. Multi-agent systems (MASs) are systems
OQL, in order to enhance the openness and accessibility of composed by at least two autonomous agents that are able
the system [13]. to interact with one another.
The agents may have a global goal to solve, or
B. Grow rooms
they may have their own goals to pursue [12]. The agent-
A grows room, is a room of any size where plants are based approach could be applied with success in the
grown under controlled conditions (Source: Wikipedia). In environmental management [2]. As such we intend to
grow rooms, artificial light is used mostly. Here, these develop a several autonomous Agents that are responsible
artificial light sources could come from three main types. for different aspects of a given environment thereby
Those are, the high intensity discharge lamps, compact forming a Multi Agent System for ensuring environment
fluorescent lamps and traditional fluorescent lamps. Since sustainability. Since a controlled environment was
these artificial lighting sources could generate a lots of necessary for implementing the solution hydroponics, the
heat, there should be a proper ventilation system installed technology of growing plants on a nutrient solution that
for the grow room. Otherwise there is a possibility of supply all nutrient elements needed for optimum plant
overheating the room and the over consuming of the
essential resources like Oxygen and Carbon Dioxide.
C. Grow boxes
Grow boxes are a smaller version of grow room which is
mostly used for domestic plantations. Here, a grow box is a
self contained environment for growing any kind of a plant
that could be grown hydroponically. Since this is a totally
enclosed environment fans have to be used for intake and
exhausting air from the system. Apart from that there will
be odor controlling systems and air conditioning systems
for the better management of the plant environment. Grow
boxes are highly useful when you are not having patch of
land for growing plant or living in an urban Flats.
D. Automation components
There are a number of automation products available in the
market for automating various aspects of a hydroponics
environment. These products include nutrient dosing
equipments, remote automation systems, wireless cameras
designed for hydroponics environments, automated water
sensors and pumps etc. Also there are few other open Fig 1: Top level architecture of SAS_EN

3. growth, in a greenhouse environment was used [4, 3]. The SAS_EN mainly uses four sensors to collect water
nutrient solution of the plantation is the basis for all level, Fertilizer level, EC and PH value. Each sensor
hydroponics systems. This contains all the mineral control takes the input signals and then sends to the ebox.
elements needed for the growth of the plant. After getting the sensor data, Logical unit is responsible to
control MCU and electric valve system which controls the
IV. DESIGN
above mentioned levels. All the control units are
The system consists of four agents namely Plant
implemented within the hydroponics box (Figure 2). The
agent, humidity agent, ventilation agent and nutrient agent.
implementation details of the SAS_EN control units are
Every agent communicates with other agents to keep the
described below.
green house environment balanced. When the temperature
goes up, the humidity agent notifies it to the ventilation
Water level controlling Unit
agent and then the ventilation agent reduces the ventilation
so that the humidity can goes up. Mean time the humidity The water level sensor and the water pump control are
agent can increase the humidity by increasing the kept inside the Hydroponics box and it measures the water
evaporation in the environment. Thus through the level of the hydroponic solution. The water pump control
communication between all the agents the environmental unit consists of electric valves which are controlled by the
conditions are kept balanced. ebox. When the water level changes it directly sends the
The figure 1 depicts the top level architecture of control signals to the logical unit and the logical unit
the system with the micro controller units (MCU), sensors, processes the water level with the given log file data and
agents and the ebox. The Agents are developed with a little using MCU unit and automatically controls the water level
knowledge base and ability communicate with appropriate by adding water to the Hydroponics box. Every time the
Agents and also able to send signals coming from Ebox to system is responsible to keep water level in required
control devices such as ventilation fans and lights. The maximum level. After reaching to the maximum water
features and the responsibilities of the agents and other
connected devices are as follows.
The plant agent is responsible for identifying the
abnormalities in the plant and the nutrient deficiencies.
The color changes happen in the leaves and the
abnormalities of the nuts and the disorders in harvest can
be identified by the plant agent. Thus with the
identification of the deficiencies the plant agent can
communicate with nutrient agent and other relevant
entities to solve the issues in the plant.
The Nutrient agent is responsible for controlling
the nutrient conditions of the plant. Thus it is able to
control the EC (Electric Conductivity), PH and nutrient
concentrations of the hydroponics solution. It controls the
concentration of the solution by adding water and Albert
solution.
The ventilation of the atmosphere is also an
important factor for a successful growth of a plant.
Therefore the ventilation agent controls the ventilation
states in the greenhouse. Thus it controls the air flow Fig 2: the architecture of hydroponics box
within the green house environment. it controls the
ventilation conditions of the plantation using the level the water level sensor sends control signals to ebox.
ventilation fan. Then it closes the electric water pumping valve.
The Humidity agent controls the humidity and
temperature states of the greenhouse. The humidity agent Air pumping Unit
and the ventilation agent maintain a direct communication Air pumping which is controlled by the ebox is another
between themselves in order to balance the atmosphere of special feature of SAS_EN. Normally it is pumping the air
the green house. When the humidity increases in the green periodically. The time is controlled by the ebox according
house environment, the ventilation agent gets activated and to the conditions of the plant and the water medium. In
increases the air flow of the greenhouse. Thus the addition to that the air pumping unit supports to mix the
ventilation and humidity agent work conversely to balance fertilizer around the hydroponics box. Whenever the
the environmental conditions of the greenhouse. fertilizer is added to the water medium the air pumping
unit automatically starts and mix up the hydroponics
V. IMPLEMENTATION

4. culture. The suitable time to add fertilizers and to start the
pumping is decided by the Plant Agent. A. Features of SAS_EN
The agents inherit the common features of software agents.
EC and PH level controlling Unit: They have specific behaviors to handle various
atmospheric conditions. Among others, following are the
The Albert solution is used as the main fertilizer for
major features of SAS_EN.
the hydroponic solution. Thus the EC and PH levels
System has autonomous monitoring and control
change according to the fertilizer level. In this case we
because agents operate with little or no intervention from
should get those three inputs together to control each other.
users, and apply control strategies accordingly. For
The current condition of the hydroponic medium is sent to
instance, if the greenhouse is used in tropical environment,
the Nutrient agent and it processes each value together.
natural ventilation will be heavily used, without operating
Therefore the concentration of the solution is controlled by
exhaust fans.
monitoring the time of water valve open. The Nutrient
Agents are necessarily able to communicate to share
agent periodically checks the EC and PH level and if there
knowledge and cooperate in solving problems. For
is a change compared to the ontology then the fertilizer
instance, on the request of Humidity Agent, the Ventilation
valves open for relevant time period. The concentration of
Agent may want to operate fast, but Nutrient Agent’s
the water medium is checked periodically to check the
concern about water level must be negotiated first. It is
imbalances of the solution. Further the pH levels in the
able to reach consents to increase utility. For example,
solution are controlled by adding suitable amount of
Plant Agent wants to increase the duration of
diluted acids. Thus the nutrient agent directly
photosynthesis to get a target high yield, and may request
communicates with the MCU then MCU controls the
Light Sensor Agent to release light for a longer period [5].
valves.
Collaboration among Agents is a means of
reducing workload of a single Agent. For instance,
Data analyze of Logical Unit: Humidity Agent can assist to reduce work load of
Logical Unit gets inputs from the sensor unit and Ventilation Agent, if the intensity of light is reduced in the
process those using data stored in ontology. Then Logical green house.
unit gives control signals to MCU to control each levels of Agents can work on the basis of previous
the Hydroponics box. The agents periodically experiences stored in the database, without solving similar
communicate with each other to analyze the conditions of problems from the scratch. Agents can also adapt to
the green house environment. situations, which are not experienced before. Parameters
required to grow different plants at different environmental
Further the reporting module of SAS_EN provides a conditions can be learned by the Agents.
daily report indicating the daily conditions of the It should be noted that the SAS_EN comprises
plantations and how SAS_EN has reacted to them. And Agent that are responsible for monitor and control
also through the SAS_EN log it can derive to certain environmental parameters such as Temperature, Humidity,
conclusions and provide custom reports. Ebox works as the Ventilation, Water, Nutrients, etc. The Agents developed
central processor and the communication medium between for this purpose can be used for modeling a sustainable
each Agent in the greenhouse and the PDA and takes up environment under given condition.
the task of information analysis, controlling and decision
making. VII. TESTING AND EVALUATION
Light Sensor Agent controls the light intensity of the
greenhouse. Micro control unit (MCU) works as the The system was tested in a hydroponics plantation. The
Bluetooth controller and sensor data collator and receiver. plants were grown in the water which was controlled by
the SAS_EN hydroponics box. The box was able to keep
VI. USER SCENARIOS AND THE FEATURES OF the required conditions of the solution continuously. We
SAS_EN tested the system for a week and after a week’s period the
plantation was healthy and looked good.
The main users of SAS_EN are farmers and the
gardeners. The farmers can get the real time control TABLE 1: WAY SAS_EN REACT TO VARIOUS CONDITIONS
environment details through the web application provided. Scenario Result Comments
Further if required, farmer can give instructions to change
Reduce the water Increase the water level solved Good
the conditions and also pre program certain conditions. level
They are given the capability of viewing the data received
from each agent, current conditions of the Hydroponics Deactivated Light Communicated fine OK
sensor
greenhouse environment and giving orders to the system.
Further the Researchers are given the capability of Raining Heavily OK OK
viewing the statistical records on the Hydroponics
greenhouse environment through the internet. Reduction of pH It was increased Ok
level

5. Test Environment Survey results
The test environment was a hydroponics plantation in Also a survey was conducted to analyze the effectiveness
a home garden. The plants were grown in the water which of the system. Ultimate objective of the project is to
was controlled by our hydroponics box. In order to ensure generate a maximum yield using minimum resources. In
that the system is working properly, both component order to test the system we used a same farm with two
testing and system testing was undertaken. nurseries where one is managed with embedded system
and other one is managed manually. We obtained
In the first scenario we reduce the water level and
following performance analysis from the farmers (Table
switch on the system. System identified that and pore the
2).Also table 3 depicts the live deployment results. Even
water up to the required level. Similar tests were conducted
though, above mention functions has taken nearly 10 min
to ensure component testing. So we reduce the light,
for 20 m2 plantation there is no liner relationship between
change EC / PH, change nutrient level. The results
time and area. The radiant reduces when area increases.
obtained during each scenario are depicted in table 1.
VI. CONCLUSION
The system was specially designed as an embedded system It is evident that use of the technology for
because in some plantations user frequent visits are rare. In environmental sustainability must have a mechanism for
order to simulate those situations the system was sensing the environmental parameters and response
continuously run for a week. Mean while when the accordingly. Therefore the system needed an autonomous
conditions were changed the Application reacted to those monitoring and control environment. Thus software agents
were able to provide this capability for SAS_EN to behave
TABLE 2: THE RESULTS OBTAINED FROM FIVE FARMERS
in different environmental conditions.
Performance Managed by the Managed manually SAS_EN can be implemented as an embedded
evaluation embedded solution application and it has already shown its potential to solve
criteria
distributed problems. The SAS_EN can be customized to
Yield Yield was high. Also Yield is average. There any application involving environmental parameters cost
there were only few were many perishable effectively. In future, we expect to enhance the system by
perishable items. (1.5% foods. (2.5% yield was
yield was perishable ) perishable ) enabling a more cooperative living among several
environments. This would allow different plantations to
Labour hours 2 labour hours per day 16 labour hours per day communicate with each other in case of adverse
environmental scenarios. For example, when one particular
Money Initial cost, electricity Labour cost, Rework cost system has been caught up in decease, it can communicate
cost this situation to other environments enabling them to take
Reliability Reliable Reliable necessary precautions pro-actively.
Response Very High Average ACKNOWLEDGEMENT
time
Usability Very High Average All the faculty members and our colleagues
deserve to be appreciated for contributing to the success of
situations fine. this project in various ways. Also the authors of the
references we have used throughout this project are highly
Also SAS_EN was tested for special scenarios such as appreciated.
multiple conditions going adverse. For example when
water level goes down and the day light fade SAS_EN
could tolerate such situations as well. REFERENCES
[1] A. Najam, D.Runnels, M. Halle Environment and globalization, five
TABLE 3: THE DEPLOYEMENT RESULTS OF SAS_EN propositions, International Institute for sustainable development,
Ministry of Foreign affairs of Denmark, 2007
Function Performance Evaluation [2] B. Malheiro and E. Oliveira, “Environmental Decision Support: a
Multi-Agent Approach”, Proceedings of Autonomous Agents 97,
Adjust EC and PH Levels 10 min for 20 m2 plantation ACM Press, 1997, pp. 540-541.
[3] C.W., Marr, Hydroponic Systems, Greenhouse Vegetable Production
Adjust Light 3 min for 3 bulbs
[4] D. Mohsen, easy gardening with hydroponics for teachers, gardners
Adjust Water Level 6 min for 20 m2 plantation and home hobbyist
[5] D.T, Bourgues, S., Wit, R., D., Auby, I., Effects on plant
Adjust Ventilation 15 min for 20 m2 plantation photosynthesis, carbon sources and ammonium availability on
nitrogen fixation rates in the rhizosphere of Zostera nolti, welsh,
Adjust Albert Solution 10 min for 20 m2 plantation [6], G. Bünemann , mineral nutrition and fruit quality of temperature
zone fruit trees, Symposium on Mineral Nutrition and Fruit Quality
of Temperate Zone Fruit Trees, 1980
[7] G. Weiss, Multiagent Systems: A Modern Introduction to Distributed
Artificial Intelligence, Cambridge, MIT Press, MA, 1999.
[8] http://hmeter.sf.net

6. [9] I.N. Athanasiadis & P.A. Mitkas, Applying agent technology in
environmental management systems under real-time constraints, In
Second Biennial Meeting of the Int'l Environmental Modelling and
Software Society: Complexity and Integrated Resources
Management, Osnabrück, Germany, pp. 531-536, 2004.
[10] M. FAUST, interaction between nutrient uptake and photosynthesis,
Symposium on Mineral Nutrition and Fruit Quality of Temperate
Zone Fruit Trees (1979)
[11] M. K. hill Understanding environmental pollution, A Primer, second
edition, , Cambridge University Press, 2004.
[12] M. Oprea, C.Nichita: Applying Agent Technology in Water
Pollution Monitoring Systems. 8th International Symposium on
Symbolic and Numeric Algorithms for Scientific Computing, 26-29
September 2006, 233-238
[13] M. Purvis, S. Cranefield, and M. Nowostawski, “A distributed
architecture for environmental information systems”, Environmental
Software Systems- Environmental Information and Decision Support,
Kluwer Academic Publishers, 2000, pp. 49–56.
[14] R. J. Jackson, M.D, C. Kochtitzky, S. Watch, Creating a healthy
environment, The impact of the built environment on public health
Clearinghouse monograph series
[15] R. P. Mills, R. H Cate at el, The living environment, core
curriculum, The University of the State of the New York, The state
education department,
[16]W. Cheng, , D. A. Sims, Y. Q. Luo, J. S. Coleman, and D. W.
Johnson. Photosynthesis, respiration, and net primary production of
sunflower stands in ambient and elevated atmospheric CO 2
concentrations: An invariant NPP: GPP ratio? Global Change Biology
6:931-942.