IoLT Smart Pot: An IoT-Cloud Solution for Monitoring Plant Growth in

Greenhouses

J. Hadabas

, M. Hovari

, I. Vass

and A. Kertesz

Software Engineering Department, University of Szeged, Hungary

Institute of Plant Biology, Biological Research Centre, Hungary

Keywords:

Internet of Things, Cloud Computing, Plant Phenotyping, Gateway.

Abstract:

According to a recent Beecham Research report, food production have to be increased by 70 percent till 2050

to feed 9.6 billion global population predicted by the United Nations Food and Agriculture Organisation.

Since Cloud Computing and the Internet of Things (IoT) have already opened new ways for revolutionizing

industrial processes, these technologies could be important for the farming industry. Smart farming has the

potential to improve productivity and reduce waste to transform agriculture. Plant phenotyping is an important

research ﬁeld that gained a high attention recently due to the need for complex monitoring of development

and stress responses of plants. However, the current phenotyping platforms are very expensive, and used in

large central infrastructures, which limit their widepread use. The newly emerging ICT technologies together

with the availability of low cost sensors and computing solutions paved the way towards the development of

affordable phenotyping solutions, which can be applied under standard greenhouse conditions. The Internet of

Living Things (IoLT) project has been launched to integrate IoT technological research with applied research

on speciﬁc, biological applications. In this paper we introduce our research results for developing a low cost

plant phenotyping platform for small sized plants, which is one of our goals in this project. The proposed

IoLT Smart Pot is capable of monitoring environmental parameters by sensors placed above the plant and into

the pot, managed by a Raspberry Pi board placed under the pot. We have also developed a private IoT-Cloud

gateway for receiving, storing, visualizing and downloading the monitored parameters sent by the pot devices.

We have performed the evaluation of our proposed platform both with simulated and real smart pots.

1 INTRODUCTION

The United Nations Food and Agriculture Organisa-

tion predicts that by 2050 global population will grow

to 9.6 billion. A recent, corresponding Beecham Re-

search report (Beecham Research, 2017) states that

food production have to respond to this growth by

70 percent increase till 2050. Agriculture also need

to be reformed since it is currently responsible for

a ﬁfth of greenhouse gas emissions and for 70 per-

cent of the worlds fresh water usage. They also ar-

gue that smart farming has the potential to improve

productivity and reduce waste by exploiting new ICT

technologies, such as the Internet of Things (IoT).

IoT represents a dynamic global network infrastruc-

ture with self conﬁguring capabilities (Sundmaeker et

al., 2010), in which things can interact and commu-

nicate among themselves and with the environment

through the Internet by exchanging sensor data, and

react autonomously to events and inﬂuence them by

triggering actions with or without direct human inter-

vention. Such systems can be utilized in many appli-

cation areas, thus they may have very different prop-

erties. According to recent reports in the IoT ﬁeld

(e.g. (Mahoney et al., 2011)), there will be 30 bil-

lion devices always online and more than 200 billion

devices discontinuously online by 2020. Such esti-

mations call for smart solutions that provide means to

interconnect and control these devices in an efﬁcient

way.

Cloud computing (Buyya et al., 2009) enables

ﬂexible resource provisions that have become hugely

popular for many businesses to take advantage of re-

sponding quickly to customers demands. There is

a growing number of cloud providers offering IoT-

speciﬁc services, since cloud computing has the po-

tential to serve IoT needs such as hiding data genera-

tion, processing and visualization tasks. With the help

of these virtualized solutions, user data can be stored

in a remote location and can be accessed from any-

144

Hadabas, J., Hovari, M., Vass, I. and Kertesz, A.

IoLT Smart Pot: An IoT-Cloud Solution for Monitoring Plant Growth in Greenhouses.

DOI: 10.5220/0007755801440152

In Proceedings of the 9th International Conference on Cloud Computing and Services Science (CLOSER 2019), pages 144-152

ISBN: 978-989-758-365-0

where.

Plant phenotyping covers high throughput ap-

proaches, which make possible to monitor the growth,

physiological parameters, and stress responses of

plants with high spatial and temporal resolution by us-

ing the combination of various remote sensing meth-

ods. Until recently typical plant phenotyping plat-

forms used very expensive instrumentation to moni-

tor several hundreds to few thousands of plants. Al-

though these large infrastructures are very powerful,

their high cost, in the range of few mEUR per plat-

form, limited their widespread, everyday use. Due

to the recent development in computer, sensor, and

IoT technology a promising alternative, called afford-

able phenotyping, started to develop, which applies

low cost sensors and computing solutions for moni-

toring fewer number of plants with high ﬂexibility in

standard greenhouse environment.

The goal of our research is the development of a

low cost plant phenotyping platform for small sized

plants, which enables the montoring of growth of

leaves and shoots in parallel with the monitoring of

environmental parameters, as well as the development

of an IoT Cloud platform capable of collecting, stor-

ing and visualizing environmental data.

The remainder of this paper is presented as fol-

lows: Section 2 introduces related works, and Sec-

tion 3 discusses the background, the research goals

and the applied idea. Section 4 presents our proposed

support gateway using IoT Cloud technologies, and

also shows its evaluation about its scalability and de-

vice management features. Finally, the contributions

are summarized in Section 5.

2 RELATED WORK

Nowadays, we can ﬁnd smart solutions in the com-

mercial world for many household areas, including

indoor plant monitoring, e.g. Xiaomi Flora (Sharma,

2018) and Parrot pot (Parrot Pot, 2018). They are

mostly capable of monitoring light, humidity and salt

content of the plant soil, and able to communicate

with nearby devices via Bluetooth. For professional

usage, there are only very few commercially available

platforms for affordable phenotyping (e.g. PhenoBox

(CzedikEysenberg et al., 2018)). However, they are

typically limitied to monitoring only a single plant.

Brogi et al. (Brogi et al., 2018) have developed

a hands-on lab activity for educational purposes by

monitoring a single plant. Their goal was to exem-

plify the use of IoT, Fog and Cloud technologies. We

exploit a similar idea for using these technologies, but

we propose a complex platform usable for real world

greenhouse application.

Dagar et al. (Dagar et al., 2018) proposed a model

of a simple smart farming architecture of IoT sensors

capable of collecting information on environmental

data and sending them over wireless networks to a

server. There are also generic solutions to monitor

IoT systems including agriculture applications, such

as the Kaa IoT Platform (Kaa project, 2018). It is

a commercial product that is able to perform sensor-

based ﬁeld and remote crop monitoring. It also has

an open source version called the Kaa Community

Edition. Such generic toolkits are quite complex and

heavy-weight, so they are not well suited to speciﬁc

needs.

Concerning generic IoT gateways, Kang et al.

(Kang et al., 2017) introduced the main types and

features of IoT gateways in a detailed study, which

presents the state-of-the-art and research directions in

this ﬁeld. This solution is also too generic for our

needs.

In contrast to these solutions, our approach aims

to provide a low-cost solution using the latest IoT and

Cloud techniques to enable a robust and scalable solu-

tion to be used for groups of plants with user friendly

management.

Figure 1: IoLT project tasks.

3 THE IOLT SMART POT

3.1 The Internet of Living Things

Project

The University of Szeged and the Biological Research

Centre of the Hungarian Academy of Sciences work

together to create a Network of Excellence called the

Internet of Living Things (IoLT) since 2017. This

project aims to integrate IoT technological research

with applied research on speciﬁc, biological IoT ap-

plications. The project will create an opensource IoLT

programming platform based on JavaScript. It will

IoLT Smart Pot: An IoT-Cloud Solution for Monitoring Plant Growth in Greenhouses

145

be able to execute applications on cheap, low capac-

ity IoT devices by providing easy to use program-

ming interfaces, thus enabling application develop-

ment for researchers of any discipline. Technologi-

cal developments will address the JavaScript executor

engine, software-hardware porting, programming en-

vironment, secure management algorithms and soft-

ware quality.

Figure 1 depicts the main development tasks (de-

noted by F1, F2 and F3) of our project. The IoLT

application areas (within F3) address (a) the develop-

ment of a smart pot for plants enabling complex plant

phenotyping using medium-high throughput charac-

terization of plant growth and physiological status,

(b) the development of a smart watch for perform-

ing actigraphy to investigate ultradian activity lev-

els of patients in psychosocial treatments, and (c)

the development of Lab-on-a-chip systems for en-

hanced microﬂuidic diagnostic technologies for high-

throughput cell analysis.

Figure 2: IoLT Smart Pot prototype.

3.2 Designing and Assembling the IoLT

Smart Pot

We have developed a low cost plant phenotyping plat-

form for Arabidopsis and other small sized plants.

The platform consists of a cluster of 12 pots (in 4x3

matrix) with individual plants, and uses a low-cost

computer based system to monitor plant growth and

environmental parameters. Plant growth is moni-

tored by using an RGB camera, located above the

plant cluster, as shown in Figure 2. The cluster is

also equipped with an LED-based illumination sys-

tem, which allows supplementing the natural light,

if needed. The usual frequency of image capture is

one hour, but it is possible to increase it up to one

minute for higher time resolution. Environmental pa-

rameters are monitored by sensors (light intensity, air

temperature, relative air humidity), placed above the

plants, as well as soil humidity sensors placed into

selected pots. The data are stored temporarily on a

memory card of the computer, and it can be trans-

ferred via WiFi connection to a database located in

a local server or in the cloud. Segmentation of plant

related green pixels and calculation of projected leaf

area is performed by a home-developed software. The

system was tested during a one-month growth period

with WT Arabidopsis plants. The used one hour im-

age capture frequency revealed a circadian change in

the projected leaf area due photoperiod dependent leaf

movements. The proposed IoLT Smart Pot system

allows to monitor the effect of various stress factors

(drought, nutrition, salt, heavy metals, etc.), as well

as behavior of various mutant lines.

4 IOT-CLOUD GATEWAY FOR

DATA MANAGEMENT

4.1 Overview of the Gateway

Application

The architecture of our proposed IoT-Cloud gateway

can be seen in Figure 3. It is composed of three ser-

vices. We have developed a (i) Node.js Webapp appli-

cation to provide a web-based graphical interface for

grouping and managing pots and users in the form of

projects. Users can register pots, and create projects

for a certain time interval, to which additional users

and already registered pots can be assigned. It is also

capable of visualizing the sensor values gathered from

one or more pots added to a project. In Figure 4,

we visualized experimental results of a cluster of pots

(called BRC Smartpot 1) of a project (titled Real BRC

CLOSER 2019 - 9th International Conference on Cloud Computing and Services Science

146

Figure 3: The architecture of the IoLT Smart Pot Gateway.

Figure 4: Historical sensor data visualization in the IoLT Smart Pot Gateway.

IoLT Smart Pot: An IoT-Cloud Solution for Monitoring Plant Growth in Greenhouses

147

Smartpot test (1 week)) depicting values of 7 sensor

types for a week of utilization. As we can see, if all

of the sensor types are selected on the graph, some

curves may overlap. We can adjust the time inter-

val (on the x-axis, while the y-axis denotes the ac-

tual sensor values), and switch on and off the sensor

types, and download the set of values according to

the deﬁned visualization parameters in CSV format.

The downloaded ﬁle can be used by the associated re-

searchers for further processing.

We have also developed a microservice called

Mosquitto MQTT Broker (ii), which is built on the

open-source Mosquitto tool (Mosquitto, 2018), using

a MongoDB (MongoDB, 2018) database to store the

received sensor values. The monitored 7 sensor types

of a pot are described by a JSON document, regu-

larly sent by the Smart Pot to the MQTT broker of

this microservice directly. The sensors of a pot is

managed by a python script implementing an MQTT

client. This script can be conﬁgured with a pot iden-

tiﬁer, sensor value sampling frequencies and picture

taking frequencies, hence each IoLT Smart Pot (which

is a cluster of pots in our case) is equipped with a cam-

era. The pictures taken are sent directly to our third

microservice called Apache Web Server (iii) via SFTP

connection.

The source code of our proposed gateway is avail-

able at (IoLT Smart Pot Gateway Source, 2019). Con-

cerning the implementation of these microservices,

we used the Docker container technology (Docker,

2018). Each microservice is placed in a Docker

container, and the three of them are composed to-

gether, since the Node.js Webapp reads the sensor

values stored in MongoDB and the pictures stored at

the Apache server. Finally, the composed microser-

vices are placed in a virtual machine (VM), in which

the container performance values are monitored by a

script (which we use for the evaluations in the next

section). The VM containing the microservices can

be placed to any cloud. In our case, it is instantiated

in the MTA Cloud (MTA Cloud, 2018) with a small

VM ﬂavor (having 1 virtual CPU core and 2 GB RAM

memory). The MTA Cloud is an OpenStack-based

national community cloud ﬁnanced by the Hungarian

Academy of Sciences in order to provide cloud ser-

vices for scientists from the academy.

4.2 Evaluation with Simulated Smart

Pots

In order to evaluate our proposed phenotyping plat-

form, ﬁrst we performed a detailed evaluation by

means of simulation. After some initial measure-

ments, we got to know the exact, real data value

{

"Project": "SampleProjekt",

"Soil-sensor II": "434.437",

"Full light intensity [lux]": 16901.38,

"Time": "2019-01-14 14:02:56",

"Humidity [%]": "41.1",

"Soil-sensor I": "594.940",

"IR light intensity [lux]": 15865.80,

"Temperature [C]": "21.8",

"Visible light intensity [lux]": 1035.58

}

Figure 5: Sample JSON message of 7 sensor values of a pot.

ranges for the installed sensors, therefore we de-

signed a simulated Smart Pot represented by python

scripts capable of sending generated sensor data via

the MQTT protocol. Figure 5 presents a generated

sample JSON ﬁle for the considered sensor types.

Figure 6: CPU measurement results for 250 pots.

Figure 7: Memory measurement results for 250 pots.

First, we created 250 simulated pots with scripts

that sent generated sensor data to our IoLT Smart

Pot Gateway service (deployed at MTA Cloud) for 30

minutes. We divided the total experiment time-frame

to the following periods:

• in the ﬁrst 10 minutes we applied sensor data gen-

eration frequency of 30 seconds (which means

that each pot sent a message of 7 sensor values

every 30 seconds);

CLOSER 2019 - 9th International Conference on Cloud Computing and Services Science

148

Figure 8: CPU measurement results for 50 pot clusters.

Figure 9: Memory measurement results for 50 pot clusters.

• in the second 10 minutes we applied sensor data

generation frequency of 10 seconds;

• in the following 5 minutes we applied sensor data

generation frequency of 2 seconds;

• and in the last 5 minutes we applied sensor data

generation frequency of 10 seconds, again.

The resource usage sampling by the monitoring

scripts were set to 10 seconds. They queried CPU and

memory resource utilization for all containers, and we

summed them to get the total resource consumption of

the composed service (thus of the whole VM). We can

see the measurement results for this initial round sim-

ulating 250 pots in Figure 6 and Figure 7. The x axis

denotes the timestamps of resource usage sampling,

while the y axis denotes the resource usage percent-

age. We can see that there are some spikes in both

resource usage percentages after the ﬁrst 10 minutes,

when we start to send more messages, and from the

20th minute the utilization is clearly rising. (Note that

the resource using sampling is less frequent than the

arrival rate of the messages.)

Next, we set the simulation parameters in a way

to mimic future, real world utilization. Our proposed

IoLT Smart Pot is basically a cluster of 12 pots, as

shown in Figure 2. To evaluate the scalability of

our gateway solution, we performed three simulation

measurements with 50, 100 and 250 clusters (com-

posed of 600, 1200 and 3000 pots respectively). In

all cases we performed the measurements for half an

Figure 10: CPU measurement results for 100 pot clusters.

Figure 11: Memory measurement results for 100 pot clus-

ters.

hour, and the simulated smart pot platform sent sensor

values with the following periods:

• in the ﬁrst 10 minutes we applied sensor data gen-

eration frequency of 5 minutes (which means that

each pot sent a message of 7 sensor values every

5 minutes: resulting 2 messages in this period per

pot);

• in the second 10 minutes we applied sensor data

generation frequency of 1 minute;

• and in the last 10 minutes we applied sensor data

generation frequency of 5 minutes, again.

Table 1: Comparison of the three evaluation rounds.

No. of clusters 50 100 250

No. of pots 600 1200 3000

Max. CPU util. (%) 13.42 30.91 39.29

Max. Mem. util. (%) 18.94 18.1 24.73

In the ﬁrst simulation for 50 clusters we set the

sampling of resource usage (processor and memory

usage) in every 10 seconds, while for the second and

the third one (100 and 250 clusters) we set it to 2 sec-

onds (to have a better resolution of resource loads).

We can see the measurement results for the ﬁrst

round simulating 50 clusters with 600 pots in Figure

8 and Figure 9 for 30 minutes. Here we can see that

the average CPU load varies between 1 and 2 percent,

IoLT Smart Pot: An IoT-Cloud Solution for Monitoring Plant Growth in Greenhouses

149

Figure 12: CPU measurement results for 250 pot clusters.

Figure 13: Memory measurement results for 250 pot clus-

ters.

and the memory usage ﬂuctuates between 15 and 19

percent. In this experiment we also observed that the

time of an actual data processing (receiving a message

and writing its contents to the database) and the time

of the resource usage sampling are rarely matched.

One matching example can be seen right after the 3rd

minute in Figure 8, which shows a spike with almost

14 percent of CPU utilization.

For the second round we doubled the number of

clusters to 100, and performed the simulation only for

5 minutes with detailed resource usage sampling of

2 seconds. We can see the measurement results for

this round simulating 100 clusters with 1200 pots in

Figure 10 and Figure 11. Now the results reveal a

periodic resource usage ﬂuctuation denoting the data

processing activities.

Finally, for the largest experiment we further in-

creased the number of pot clusters to 250 arriving to

a total number of 3000 simulated pots. For this third

round, we performed the simulation for 30 minutes,

again, with the same periods as deﬁned for the ﬁrst

round (of 50 clusters). We can see the measurement

results in Figure 12 and Figure 13. If we take a look at

the middle 10 minutes period we can see the periodic

resource usage spikes, as in the previous round.

To summarize our investigations, Table 1 com-

pares the maximum resource utilization values mea-

sured during the experiments. We can see that by

increasing the number of pots to be managed by the

gateway service, the utilization raises. As expected,

the CPU utilization was the highest in the third round

managing 3000 pots at a time with more than 40 per-

cent, and the memory utilization is also the highest

with almost 25 percent. These results show that we

can easily serve numerous phenotyping projects mon-

itoring up to thousands of pots with a single gateway

instance in a Cloud.

4.3 Evaluation with Real Pots

After the simulation experiments proved the usability

of our gateway service, we have tested the IoLT Smart

Pot platform with real world utilization. We placed 12

Arabidopsis plants in small pots to the prototype clus-

ter (as shown in Figure 2), and conﬁgured the smart

pot (its python scripts running in the Raspberry Pi

board under the cluster) to send the sensor values reg-

ularly to the IoT-Cloud gateway service. The wiring

of the smart pot prototype allowed us to consider the

whole cluster as a single IoT device, meaning that 7

sensors was placed in the cluster in total (for some

of the 12 pots). We performed the monitoring of the

growth of Arabidopsis plants under standard green-

house conditions for more than a month. RGB imag-

ing was performed every one hour, and sensor sam-

pling frequency was set to 5 minutes (resulting in one

JSON message per 5 minutes). Figure 14 shows a

query at the gateway web interface for one week of

monitoring the real IoLT Smart Pot prototype. Fig-

ure 15 shows the pictures taken on 2018.12.01., and

one month later, revealing the growth of the moni-

tored plants.

The biologists performed a post-processing of the

monitored data by downloading them from the gate-

way portal. Figure 16 depicts the time course of leaf

area growth of Arabidopsis plants. The red curve

shows the time dependence of projected leaf area in

an 8 days time window, while the blue curve in the

inset shows the same for the whole 29 days of the ex-

periment. The data represent the mean value for the

12 plants placed in Smart Pot cluster. The time course

of the projected leaf area revealed a cirkadian oscilla-

tion pattern due to periodic leaf movement (ﬂattening

in the dark and erection in the light period).

5 CONCLUSION

Agriculture takes a signiﬁcant role in greenhouse gas

emissions, therefore smart farming solutions have

started to be developed to improve productivity and

reduce waste by exploiting new ICT technologies,

such as the Internet of Things. Affordable pheno-

CLOSER 2019 - 9th International Conference on Cloud Computing and Services Science

150

Figure 14: Real world measurement data visualization in the IoLT Smart Pot Gateway.

Figure 15: Pictures of a Smart Pot cluster taken at 2018.12.01 (left), and 2019.01.01 (right).

typing represents a related research area that aims

to apply low cost sensors and computing solutions

for smart monitoring of plants with high ﬂexibility in

standard greenhouse environments.

To contribute to this aim, in this paper we pro-

posed the IoLT Smart Pot Platform, which is capa-

ble of monitoring environmental parameters with IoT

technology by placing sensors above the plants and

into the pots. We also developed a private IoT-Cloud

gateway for receiving, storing, visualizing and down-

loading the monitored parameters sent by the IoT de-

vices of the pots. We have performed the evaluation

of our proposed platform both with simulated and real

smart pots, and the results proved the scalability and

ﬂexibility of our platform.

Our future work will address further improve-

ments of the smart pot by attaching further sensors,

and we also plan to extend the IoT-Cloud gateway

IoLT Smart Pot: An IoT-Cloud Solution for Monitoring Plant Growth in Greenhouses

151

Figure 16: Post-processed environmental results of the real

experiment.

with additional services for postprocessing the moni-

tored environmental data.

ACKNOWLEDGEMENTS

The research leading to these results was supported

by the Hungarian Government and the European

Regional Development Fund under the grant num-

ber GINOP-2.3.2-15-2016-00037 (”Internet of Liv-

ing Things”), and by the Ministry of Human Ca-

pacities of Hungary under the grant number 20391-

3/2018/FEKUSTRAT. The authors also thank for the

usage of the MTA Cloud (MTA Cloud, 2018) that

helped us achieve the results published in this paper.

REFERENCES

Beecham Research. 2017. Smart Farming.

Smart Farming Sales Brochure, http://www.

beechamresearch.com/ﬁles/Beecham%20Research

%20%C2%BB%20Smart%20Farming%20(sales)

%202017%20.pdf

H. Sundmaeker, P. Guillemin, P. Friess, S. Woelfﬂe. 2010.

Vision and challenges for realising the Internet of

Things. CERP IoT, Cluster of European Research

Projects on the Internet of Things, CN: KK-31-10-

323-EN-C.

J. Mahoney and H. LeHong. 2011. The Inter-

net of Things is Coming. Gartner report,

https://www.gartner.com/doc/1799626/internet-

things-coming

Buyya, R., Yeo, C. S., Venugopal, S., Broberg, J., and

Brandic, I. 2009. Cloud computing and emerging

IT platforms: Vision, hype, and reality for delivering

computing as the 5th utility. Future Generation Comp.

Syst, 25(6):599–616.

P. Sharma. 2018. Xiaomi Smart Flower Pot Flora

Review, https://www.xiaomitoday.com/xiaomi-smart-

ﬂower-pot-ﬂora-review/, December, 2018.

Parrot Pot, https://www.parrot.com/global/connected-gar

den/parrot-pot, December, 2018.

CzedikEysenberg, A., Seitner, S., Gldener, U., Koemeda,

S., Jez, J., Colombini, M. and Djamei, A. 2018. The

’PhenoBox’, a ﬂexible, automated, opensource plant

phenotyping solution. New Phytol, 219: 808-823.

A. Brogi, S. Forti, A. Ibrahim and L. Rinaldi. 2018. Bonsai

in the Fog: An active learning lab with Fog comput-

ing. Third International Conference on Fog and Mo-

bile Edge Computing (FMEC), Barcelona, pp. 79-86.

R. Dagar, S. Som and S. K. Khatri. 2018. Smart Farm-

ing – IoT in Agriculture. International Confer-

ence on Inventive Research in Computing Applica-

tions (ICIRCA), Coimbatore, India, pp. 1052-1056.

Kaa project website, https://www.kaaproject.org/documen

tation/, January, 2019.

B. Kang, D. Kim, H. Choo. 2017. Internet of Everything:

A Large-Scale Autonomic IoT Gateway. IEEE Trans-

actions on Multi-Scale Computing Systems, vol. 3, no.

3, pp. 206–214.

Mosquitto website, https://mosquitto.org/, November,

2018.

MongoDB website, https://www.mongodb.com/what-is-

mongodb, September, 2018.

Docker Container Environment, https://www.docker.com/,

September, 2018.

MTA Cloud service, https://sztaki.cloud.mta.hu, Septem-

ber, 2018.

IoLT Smart Pot Gateway on GitHub, https://github.com/

sed-inf-u-szeged/IoLT-Smart-Pot-Gateway, January,

2019.

CLOSER 2019 - 9th International Conference on Cloud Computing and Services Science

152