Digital Villages: A Data-Driven Approach to Precision Agriculture in

Small Farms

Ram Fishman

, Moushumi Ghosh

, Amit Mishra

, Shmuel Shomrat

, Meshi Laks

, Roy Mayer

Aakash Jog

, Eyal Ben Dor

and Yosi Shacham-Diamand

1,4

School of Public Policy, Faculty of Social Sciences, Tel Aviv University, Tel Aviv, Israel

TAU/TIET Food Security CoE, Thapar Institute of Engineering and Technology, Patiala, Punjab, India

Remote Sensing Lab, Faculty of Exact Sciences, Tel Aviv University, Tel Aviv, Israel

School of EE, Faculty of Engineering, Tel Aviv University, Tel Aviv, Israel

{meshi_laks, roymayer, aakashjog96}@gmail.com

Keywords: Precision Agriculture, Sensor Network, Field-deployable Sensors, Satellite Multispectral Imaging.

Abstract: An approach for system monitoring of smallholder farms. The system will be based on low-cost mobile units

(i.e. IoTs, phones) collecting and transmitting data directly from the farms. The IoT information will be

merged with available and free access satellite data to form near real-time thematic images to the end-users.

It will serve people with low technical literacy who are working with smallholders in developing countries.

The novelty of using an integrated interdisciplinary behavioral-technological approach that builds on our

respective disciplinary expertise, and the ability to pilot and implement at scale through partnerships, on the

ground, allowing gaining new insights into smallholder cultivation and revolutionizing agricultural extension

in the developing world. To achieve that goal of Holistic Integrated Precision Agriculture Network (HIPAN)

three networks have been established in experimental farms in India: wireless network for “on-the-ground”

sensing, virtual network with satellite multispectral imaging-based data and social network collecting the

farmers’ inputs. The three networks are fused together and the data is processed using a cloud supported data

analysis; the results are visually transferred to the farmers as well as to organizations and companies for their

benefit.

1 INTRODUCTION

The concept of the digital village provides modern

technology-based solutions for intelligent crop

farming. It assists smart irrigation (Atieno, L. V., &

Moturi, C. A. 2014, Manami, A., Harshitha, H., &

Mohan, R., 2018, October), efficient electricity

supply (Mackenzie, D., 2019), E-healthcare (Ella, S.,

& Andari, R. N. (2018, October) water quality

monitoring and management (Manoharan, A. M., &

Rathinasabapathy, V. 2018). Digitally supported

precision agriculture will improve living conditions

in rural areas and reduce population migration to

urban areas (Shukla, P. Y., 2016). Precision

agriculture is a wide field the includes identifying

field phenotyping (Großkinsky, D. K et al., 2015),

manually or assisted by robotics (R Shamshiri et al,

2018), data collection and transfer using information

and communication (ICT) methods (Koutsouris, A.

2010, Griffith, C., 2013), and data analysis using

various methods, for example, deep learning and

artificial intelligence (Wolfert, S. 2017, Kamilaris,

2017). Recent papers are discussing the issue of

precision agriculture in the rural areas (Walter, A.,

Finger, R., Huber, R., & Buchmann, N. (2017), Wolf,

S. A., & Wood, S. D. 1997 and Salemink, K., Strijker,

D., & Bosworth, G. 2017) highlight its importance.

We propose to develop, field-test and implement

a novel, integrated approach to monitoring

smallholder farms that is based on the rapidly

increasing ability of low-cost mobile phones and

sensing devices to collect and transmit data directly

from farms. We will adopt these technologies to be

usable at scale by extension agents of low technical

literacy who are working with smallholders in

developing countries.

The novelty of using an integrated

interdisciplinary behavioral-technological approach

that builds on our respective disciplinary expertise,

and the ability to pilot and implement at scale through

partnerships, on the ground, in developing countries

could give our approach the potential to gain new

Fishman, R., Ghosh, M., Mishra, A., Shomrat, S., Laks, M., Mayer, R., Jog, A., Ben Dor, E. and Shacham-Diamand, Y.

Digital Villages: A Data-Driven Approach to Precision Agriculture in Small Farms.

DOI: 10.5220/0009373101610166

In Proceedings of the 9th International Conference on Sensor Networks (SENSORNETS 2020), pages 161-166

ISBN: 978-989-758-403-9; ISSN: 2184-4380

161

research insight into smallholder cultivation and to

revolutionize agricultural extension in the developing

world. Innovative sensing technologies are already

widely used in industrialized farms, and there is

growing interest in applying them in developing

countries. However, our approach stands out in:

(1) A human-centered design process, in the field of

“crowd sourcing or citizen science , of a tool that

is useable by extension agents with limited

technical literacy and will overcome social,

cultural, economic (cost) and behavioral

limitations to the applicability, at scale.

(2) Integrating data on farmers (crop assessments,

decisions, practices, investments and sales) and

crops (biophysical data gathered from sensors).

To date, both types of data remain disparate,

severely limiting inter-disciplinary research.

An ability to pilot, adapt and deploy the same

product at the global scale through existing

partnerships with strong partners in India, Nepal and

Kenya.

2 THE DIGITAL VILLAGE

The digital village project (Fig. 1) includes several

components:

 Development of the digital platform.

 Integration and development of new and existing

sensing technologies.

 Piloting and adapting in real field conditions in

ten villages in Punjab.

 Data analysis to derive novel research insights on

smallholder agriculture.

Figure 1: The digital village concept (Illustration).

The idea is to develop a holistic remote and local

sensing methods using both spectral imaging and

Micro and Nano sensor based digitalized information

and communication technology (ICT) for the farmer

use, taking data from the farmland to the cloud

providing accurate near-real time to the logistics

analysis. This kind of a system, as well as other many

options, are tailored to the huge varieties of

agriculture. This approach, which we may call,

Holistic (or Wireless) Integrated Precision

Agriculture Network (HIPAN or WIPAN), can be

based on existing cellular networks as well as on

proprietary networks forming an “Internet of Things”

network for agriculture and food.

The project methodology (Fig. 2) encompasses

data display for the respective field parameters: soil

properties (pH, texture, temperature, organic matter,

nitrogen, salinity, phosphorous content, pesticides

and water properties), plant parameters, pre-planting

& post-planting, irrigation requirements and post

harvesting parameters and vegetation stress (i.e.

color, shape, volatiles etc.). The proposed technology

is being developed to present current status as well as

to predict possible scenarios such as dehydration,

transpiration heat (or cold) damage or pathogen

attacks based on the change detected using Micro

Systems Technologies, i.e. sensors, micro-electro-

mechanical systems (MEMS) supported by very large

scale (VLSI) application-specific integrated circuits

(ASIC). Close and far remote sensing means (from

the field, air and space domains) is being developed

to assess soil chemical and physical attributes using

simple and easy to use methodologies. Libraries of

spectral signals (LSS) resulted from the outcrop of the

former activity are being established over the cloud.

The LSS is processed by using deep learning

approach yielding recommendations to the farmers

regarding the best ways to maximize field production

on a local basis.

Predicted attacks can be verified using bio-signals

taken from the plant leaves, stem roots or sap. Based

on that, the pesticide and dosage could be decided and

displayed digitally for the respective field and crop.

This could help the farmer in various avenues

providing the analysis results from the lab to the field

directly for the farmers’ use. It can also help the

government synchronizing their activity taking care

of stress and attacks appearing in numerous places.

The MEMS-based sensor will require data analysis

for the respective crop and field area, which could be

processed to prepare the standard bands. Within the

context of Indian agriculture practices, the proposed

technology offers an advantage in both the status

monitoring collecting data for both short- and long-

range status report and also on the response strategy.

The Digital village environment includes both

microelectronics-based sensors and networks and

hyper and multispectral imaging data. The

The Digital village

Precision Agriculture

Digital Dairy

E-Healthcare

Waste Water Management

E-transport

WSN4PA 2020 - Special Session on Wireless Sensor Networks for Precise Agriculture

162

information is typically in the visible and near-

infrared parts of the spectrum. The reflected and

emitted light provides information about the soil and

plants’ chemical and physical status. This information

is used as a proxy for wet chemistry analyses of both

soil and vegetation and can be used from drones

and/or free access satellite data. Information such as

visible change due to drought, nutrient deficiency,

pathogens, weeds or pests can be processed rapidly

providing the farmer early warning regarding the crop

yield or information about maturation and harvest

readiness. This information has a high economic

value.

There are a few challenges in hyperspectral

imaging: a) identifying the signal. i.e. identifying its

location and separating it from the noise, b) taking

care of its high dimensionality, i.e. every vector data

has many components generated by the imaged

bands, and c) generating “big data” and “deep

learning” algorithms analyzing the images.

The digital village uses databases as well as

algorithms, for their analysis. A Soil Spectral Library

(SSL which is the LSS of soil) for Punjab is being

constructed in order to foster precise monitoring of

the soil’s properties in the field for improved

agriculture practices without being dependent on an

extensive laboratory-based labor. Second, to develop

machine learning algorithms for hyperspectral

analysis; The goal is to develop the base for the proper

algorithms making the Internet Of Things and the

associated multispectral analysis useful. For example,

deep learning and machine learning algorithms can be

useful to automate the overall system reducing the

need for manual intervention providing the farmer

with the desired information faster and in a more

reliable and cost-effective way. The data is processed

externally whereas the final product, the thematic

maps, with practical instructions. will be uploaded

and delivered to the farmer’s mobile screen.

The model incorporates novel concepts such as

Deep learning and the “Intelligent cloud” introducing

concepts such as smart gateways and machine

learning in the cloud leveraging the capabilities of the

cloud. The cloud is getting more intelligent and is

capable to learn from the data stored in the cloud. It

can make predictions and analyze the situation better.

This platform is capable to perform tasks accurately

and efficiently. Artificial intelligence systems are

working better on cloud computing servers. This is

due to the low cost of operations, better reliability,

and the availability of huge processing power

analyzing a large amount of data. The database and

algorithms are being written in a modular way and in

open architectures.

The IoT network includes the sensors and the

gateway in a mesh architecture. The overall system is

adaptive, flexible and smart. It has few tasks: a)

taking care of the data, filtering it, storing it and

transmitting it, b) providing general information such

as position and time, c) monitoring system

performance generating a status report, d) checking

the sensors deciding what to do when a sensor fails,

and e) providing the desired cybersecurity layers.

Fog, co-located with the Smart Gateway can perform

the pre-processing tasks generating interrupts and

status report to the system on the way to the cloud.

We assume that only pre-processed data is required to

be stored in the cloud, providing relatively low

latency communication and more context-awareness.

3 OVERVIEW

The research includes the following steps where the

whole concept is being evaluated:

1. Run a feasibility study defining the desired

crop where the holistic approach of IoT +

smart gateways and novel algorithms (i.e.

deep learning, AI, etc.) is being applied.

2. Run a feasibility study defining the key

parameters required for that crop (including

soil water and climate), prioritized them and

define the economic value of the HIPAN

approach: the intrinsic contribution to the

overall yield, the extrinsic impact on the

marketplace up to the customer taking into

consideration the whole food chain

3. Assess the social impact of the HIPAN

approach including the resources needed for

training, application and management.

4. Assess the potential economic value building

industry in India based on the HIPAN

approach, both national and international.

5. Identifying a national (or international)

partner for large scale manufacturing.

4 METHODOLOGY

The Digital village methodology takes action from

the plant to the cloud using wireless networks of

sensors, gateways, ground and satellite multispectral

data and social networks. (Fig. 2)

Digital Villages: A Data-Driven Approach to Precision Agriculture in Small Farms

163

Figure 2: The digital village methodology.

A key strength of our approach lies in the ability

to continuously pilot, evaluate and adapt the platform

by leveraging on the partnerships between the

universities, the research institutes and the farmers.

Agricultural extension agents will be employed in the

pilot “digital village” getting careful training by

faculty and students who are supervising the

fieldwork. The agents, supported by students, collect

“non-digital” data complementing the “digital” data

from the sensors.

5 DATA ANALYSIS AND

EVALUATION OF PILOTS

After completion of developing the prototype, we will

implement data collection on the scale of a few

hundred farmers in the three field sites over the course

of one agricultural annual cycle. We demonstrate the

potential of our approach which includes also an

assessment of the correlation between farmers’ percep-

tions of crop conditions and actual conditions captured

by sensing technologies, and whether any differences

between perceived and actual status may lead farmers

to respond in agronomically inappropriate ways. To

date, hardly anything is known about those questions.

5.1 Sensor Networks

The sensor periodically measures physical and

chemical parameters of field such as temperature,

relative humidity, pH value etc. Those sensors are

placed in selected locations in the field and

configured in a mesh topology. The Sensor Gateway

(SG) is designed to transmit all the data collected by

sensors throughout the monitored site to the

communication server and through it - to the Web

application. The SG acts as an access point for

sensors, and other wireless sensors. The SG is the

"root node" of the sensors network, where all the

measurements are uploaded to. The SG also connects

to the communication server, further uploading

measurements from the sensors through an internet

connection, whether cellular or LAN based. The data

is merged with remote sensing information from

drones and satellites.

5.2 Satellite based Networks

For that purpose, field, drone and satellite sensors will

be used covering the optical range (0.4-2.5nm) as

well in some parts we might use imager covering the

“thermal” long-wavelength infrared (LIR) range (7.5-

11nm). The optical sensors are multi and hyperspec-

tral ones for field assessment operating either as a

single point measurement or in the imaging mode.

The point spectrometer is used to construct the SSL

and the imaging for establishing the thematic maps on

a spatial basis. To that end, we will use sensors on

drones (for local monitoring) and from satellites (e.g.

LANDSAT and Sentinel-2) that characterize by an

effective temporal coverage. We also construct the

foundation to process the forthcoming hyperspectral

sensors from orbit that will be available soon to expert

users.

5.3 Social Networks

In parallel to the technology-based networks social

networks are also established verifying the actual

status of various parameters in the field as seen via

the human perception of the farmers and other human

factors. This information is used to adjust the overall

usability of this approach for the benefit of the

farmers and the a whole agricultural system as

governed by local and state government and non-

government organizations.

6 HOLISTIC INTEGRATED

PRECISION AGRICULTURE

NETWORK (HIPAN)

The HIPAN is actually an Internet of Things (IoT)

system tailored for agriculture. As such it is

composed of the following units:

6.1 The Hardware Unit

6.1.1 The End Unit at the Field

1. Sensor/actuator,

2. Front end electronics – analogue unit,

analogue to digital converter (ADC),

WSN4PA 2020 - Special Session on Wireless Sensor Networks for Precise Agriculture

164

3. Digital unit – processor, memory,

4. Communication unit,

5. Power supply – power harvester, distribution

and monitoring unit.

6.1.2 Network Gates

Smart gateways, there are few options depending on

the network type (i.e. star, mesh) and how we collect

the data. We currently use the following approach:

1. For connecting the sensors between themselves

and the gate we use the amateur band at 433 MHz

(non-licensed band),

2. For connecting to the gateways an outside we use

licensed bands (i.e. WIFI, Cellular, etc.).

6.1.3 Multispectral Imaging Units

1. A full range ASD spectrometer (400-2500nm) (or

equivalent) as the motherhood sensor with a point

spectrometer of Ocean Optics (400-2500nm) as a

daughter’s sensor shall be required to build the

LSS.

2. Assemblies to Cellphone cameras for the standard

measurement of the soil and crop images.

3. A gate to ESA and NASA data satellite data basis.

4. SoilPRo® assembly for spectral measurement of

the undisturbed soil surface.

6.2 The Software Unit

6.2.1 Communication Protocols

Following is the list of principles used in the digital

village:

1. Connecting he sensors between themselves and

the gate using the amateur band at 433 MHz

(non-licensed or amateur band),

2. Connecting to the gateways and to the outside

using licensed bands (i.e. Wi-Fi, Cellular, etc.)

3. Data processing algorithms: improving the signal

to noise ratio, data compression, spectral based

model for a given attribute etc.

4. Data security protocols – RSA, blockchain etc.

5. Data analysis algorithms - preprocessing to

identify critical problems before sending the data

to the cloud for further analysis,

6. General software: control, monitoring and

uploading / refreshing protocols. Data transfer

from the field to the cloud and back

6.2.2 Additional Units

Signal processing, data encryption etc.

6.3 Auxiliary Outcome

The digital village system provides the following

tools:

 Chemical and physical analytical data of soil and

plants using traditional laboratory methods

Establishing a SSL for Punjab. Within this action,

soil samples from all over Punjab will be collected

and documented in the field under a common

protocol and achieve in one place for future

analyses. The soil will undergo traditional wet

chemistry analysis as well as a standard soil

spectral measurement. This process is adaptive:

as much more samples will be collected, the SSL

will be improved. In the long term, the Punjab

SSL will be merged with the World SSL.

 Extracting spectral base models for any attribute

by deep learning approach (first attributes are:

Organic Matter, Salinity, Soil water retention, pH,

and Texture)

 Developing a simple field apparatus for the

farmer. Developing the Satellite base application;

Applying the SSL on the current and future

satellite sensors.

7 CONCLUSIONS

This “digital village” concept combines the Internet

of Things concept with satellite-based information

generating a localized low-cost source of data and

analysis tools for the farmers. It is expected to reduce

costs significantly in precision agriculture monitoring

allowing small farm, which is the most abundant

concept in underdeveloped and developing countries,

to succeed. However, the digital village concept,

where modern digital media is harnessed for the

benefit of the farmer requires significant research and

tune-ups. We describe here a preliminary study being

carried in India under the Tel Aviv University/Thapar

Institute of Engineering and Technology Food

Security Center of Excellence (T

FSSoC). This

project is paving the way to more specific

applications with very specific targets:

 The application: variety on the field, variety in the

soil, plant or animal, post harvesting and storage,

transport and delivery etc.

 The technology: The system complexity, the huge

number of possible sensors, IoT systems,

communication protocols etc.

Therefore, here are a few important objectives:

 The most important agriculture products to apply

the HIPAN approach,

Digital Villages: A Data-Driven Approach to Precision Agriculture in Small Farms

165

 The most successful technologies to be applied,

 Define the test case and building a prototype,

testing its feasibility in the farm.

 Define the best practices to engage farmers with the

HIPAN approach

In future phases of this research, we will seek to

scale up gradually to additional sites in Punjab and

India and conduct a “Big Data” analysis to derive

novel research insights on smallholder agriculture.

We will also conduct a randomized evaluation of the

impacts of the recommendations on productivity.

ACKNOWLEDGEMENTS

Thanks to the TAU/TiET Food Security Center of

Excellence grant for the “Digital village” project,

2019. This research was partially supported by the

Israel Science Foundation (grant no. 1616/17). We

would like to acknowledge the Boris Mints Institute

for Strategic Policy Solutions to Global Challenges,

the Department of Public Policy and the Manna

Centre for Food Security, Tel Aviv University for

their support under the program "Plant based heat

stress whole-cell-biosensor" (grant no. 590351) 2017.

Also thanks to the Boris Mints Institute for their

support of the NITSAN Sustainable development

laboratory. Finally, thanks to our industrial partners:

Cartasense Inc. and Unispectral Inc. for supporting

the project infrastructure

REFERENCES

Atieno, L. V., & Moturi, C. A. (2014). Implementation of

Digital Village Projects in Developing Countries-Case

of Kenya. British J. of Applied Science & Technology,

4(5), 793.

Ella, S., & Andari, R. N. (2018, October). Developing a

Smart Village Model for Village Development in

Indonesia. In 2018 International Conference on ICT for

Smart Society (ICISS) (pp. 1-6). IEEE.

Griffith, C., Heydon, G., Lamb, D., Lefort, L., Taylor, K.,

Trotter, M., & Wark, T. (2013). Smart Farming:

Leveraging the impact of broadband and the digital

economy. New England: CSIRO and University of New

England.

Großkinsky, D. K., Pieruschka, R., Svensgaard, J., Rascher,

U., Christensen, S., Schurr, U., & Roitsch, T. (2015).

Phenotyping in the fields: dissecting the genetics of

quantitative traits and digital farming. New Phytologist,

207(4), 950-952.

Kamilaris, A., Kartakoullis, A., & Prenafeta-Boldú, F. X.

(2017). A review on the practice of big data analysis in

agriculture. Computers and Electronics in Agriculture,

143, 23-37.

Koutsouris, A. (2010, July). The emergence of the intra-

rural digital divide: A critical review of the adoption of

ICTs in rural areas and the farming community. In 9th

European IFSA Symposium (Vol. 4, No. 7).

Mackenzie, D. (2019). IEEE Smart Village: Sustainable

Development Is a Global Mission. IEEE Systems, Man,

and Cybernetics Magazine, 5(3), 39-41.

Manami, A., Harshitha, H., & Mohan, R. (2018, October).

IoT based Smart Village. In TENCON 2018-2018 IEEE

Region 10 Conference (pp. 1219-1223). IEEE.

Manoharan, A. M., & Rathinasabapathy, V. (2018, August).

Smart Water Quality Monitoring and Metering Using

Lora for Smart Villages. In 2018 2nd International

Conference on Smart Grid and Smart Cities

(ICSGSC) (pp. 57-61). IEEE.

Salemink, K., Strijker, D., & Bosworth, G. (2017). Rural

development in the digital age: A systematic literature

review on unequal ICT availability, adoption, and use

in rural areas. Journal of Rural Studies, 54, 360-371.

Shamshiri, R., Weltzien, C., Hameed, I. A., J Yule, I., E

Grift, T., Balasundram, S. K., & Chowdhary, G. (2018).

R&D in agricultural robotics: A perspective of digital

farming.

Shukla, P. Y. (2016). The Indian smart village: Foundation

for growing India. International Journal of Applied

Research, 2(3), 72-74.

Walter, A., Finger, R., Huber, R., & Buchmann, N. (2017).

Opinion: Smart farming is key to developing

sustainable agriculture. Proceedings of the National

Academy of Sciences, 114(24), 6148-6150.

Wolf, S. A., & Wood, S. D. (1997). Precision Farming:

Environmental Legitimation, Commodification of

Information, and Industrial Coordination 1. Rural

sociology, 62(2), 180-206.

Wolfert, S., Ge, L., Verdouw, C., & Bogaardt, M. J. (2017).

Big data in smart farming–a review. Agricultural

Systems, 153, 69-80.

WSN4PA 2020 - Special Session on Wireless Sensor Networks for Precise Agriculture

166