5
IoT and Machine Learning-Based Approaches for Real Time Environment Parameters Monitoring in Agriculture: An Empirical Review

Parijata Majumdar1 and Sanjoy Mitra2*

1Department of Computer Science and Engineering, National Institute of Technology, Agartala, India
2Department of Computer Science and Engineering, Tripura Institute of Technology, Agartala, India

Abstract

Agriculture monitoring is a promising domain for the economy as it is the primary contributor of job market and food production. Farmers are facing challenges in reducing water consumption and formulating the best irrigation schedules due to discontinuous monsoon, changing weather conditions for improvising crops yield and soil fertility. IoT-based decision making gives real time insight of weather parameters based on cost-effective sensor data acquisition and intelligent processing that reduces manual labor and saves time in Agriculture. Here in this chapter, we present an empirical review on real time visualization and on demand access of weather parameters even from remote locations and intelligent processing using IoT-based solutions like Machine Learning (ML). The ever-augmenting technologies like Machine Learning paved the way for identifying and adapting changes of crop design and irrigation patterns taking into account multi-dimensional large variety of weather data to accurately predict climate conditions suitable for crop irrigation. Hence, this chapter offers a detailed review of IoT-based Machine Learning solutions for precision Agriculture depending on weather and irrigation schedules. This chapter also highlights security solution based on Machine learning capable of handling illegal data access by intruders during cloud data storage.

Keywords: IoT, wireless sensors, machine learning, smart agriculture, weather, irrigation

5.1 Introduction

India is primarily an agricultural economy. Farmers need to control irrigation, as intermittent monsoon droughts and flooding lead either to low or excessive irrigation, which decreases soil fertility and interferes with crop yield. Due to unequal water supply, lixidation, denitrification and rolling out of mud the amount of fertilizer is also wasted. Traditional irrigation methods such as furrows, water sprinklers cannot predict the exact water needed and water supply duration. Emerging concepts of Internet of Things (IoT) and wireless sensor networks allow smart agricultural monitoring, processing and storage of real-time weather data. The wireless sensors used for weather data analysis help us to understand climate change’s influence on crop growth and reduce the problem of instability and parameter variability due to distant location in real time. Such a study of soil moisture and weather restrictions may also be rendered in order to control irrigation schedules minimizing loss of water. Unlike conventional methods, the weather data implanted in small microcontrollers are exchanged in real time by using appropriate communication protocols from near or remote areas which lead to precise, economic time-saving monitoring of agriculture.

IoT can solve in real time the problem of soil conditions, the content of soil moisture, temperature and humidity levels, use of fertilizers, cultivation of crops to optimize water usage, which is unaddressed by conventional farming techniques for sustainable agriculture growth.

5.2 Machine Learning (ML)-Based IoT Solution

IoT-centered ML solutions include the autonomous framework for local farmers based on the sensor data, which is highly scalable, simple to use, reliable and offers flexibility. Machine learning can provide a range of suggestions and insights into the decisions of farmers by analyzing statistically and recognizing links between these sensor-generated Big data. Machine learning helps farmers by determining the most suitable type of crop to be grown based on location, weather and soil type etc. all over the world.

Figure 5.1 shows IoT-based solution for smart farming based on irrigation and weather parameters.

Schematic illustration of IoT-based intelligent agriculture monitoring.

Figure 5.1 IoT-based intelligent agriculture monitoring.

5.3 Motivation of the Work

Solutions focused on IoT provide versatile, easy to use, robust and adaptable autonomous system for local farmers based on sensor generated data. For effective decision making in light of the problem of management of irrigation schedules reducing water and fertilizer wastage by environmental parameter acquisition in real time, several IoT based approaches are used in irrigation and weather monitoring perspective as these two factors are pivotal for precision Agriculture management.

5.4 Literature Review of IoT-Based Weather and Irrigation Monitoring for Precision Agriculture

In order to guarantee a high degree of agricultural precision, IoT brings automation through the irrigation methods such as furrow, soil, dripping, capillary and sprinklers by allowing sensor communication according to different weather parameters such as temperature, humidity, wind speed, wind direction, pressure, gases, UV radiation, etc. On processing sensor data, actuator is enabled to take irrigation controlling measures, even in remote locations. Via accurate prediction of weather parameters using cost effective sensors and reduced manual work by farmers, IoT enables intelligent decision-making to increase crop yield.

Table 5.1 gives an overview of IoT-based irrigation methods with claimed advantages and shortcomings. Pivotal environment parameters, unmonitored parameters and mode of the work are also shown.

Table 5.2 gives an overview of Communication protocols used to communicate sensor data to the cloud with Security attack perception in IoT layered architecture used by different methods of IoT-based irrigation monitoring.

Table 5.3 details about the Sensors with measuring range, micro-controllers with communication interfaces and data storage used by different methods of IoT-based irrigation monitoring.

Table 5.4 gives an overview of IoT-based weather parameter methods with claimed advantages and shortcomings. Pivotal environment parameters, unmonitored parameters and mode of the work are also shown.

Table 5.5 gives an overview of Communication protocols used to communicate sensor data to the cloud with security attack perception in IoT-layered architecture used by different methods of IoT-based weather parameter monitoring.

Table 5.6 details about the sensors with measuring range, micro-controllers with communication interfaces and data storage used by different methods of IoT-based weather parameter monitoring.

5.5 Literature Review of Machine Learning-Based Weather and Irrigation Monitoring for Precision Agriculture

The IoT solution based on machine learning is a close loop intelligent irrigation control system, which makes decisions and gathers knowledge from sensor-generated weather parameters. A broad variety of weather data is used to adjust irrigation schedules according to changing weather conditions by applying artificial neural networks, classification and regression algorithms. For mathematical modeling fuzzy logic is also used where accuracy of user dependent fuzzy logic must be measured by long data surveillance periods of unpredictable weather parameters. The ideal weather parameters is set as a threshold value to be compared with changing weather parameters for agriculture precision to improvise crop yield based on current climate conditions in order to continuously update irrigation schedules. The nomenclature of various ML algorithms used in Agriculture monitoring are Support vector machines (SVM), Artificial neural network (ANN), Support vector regression (SVR), Linear discriminant analysis (LDA), Radial basis function (RBF), Decision tree classifier (DTC), Genetic algorithm (GA), Fuzzy logic (FL), Random forest (RF), Reinforcement learning (RL), etc.

Table 5.1 Review of different methods on IoT-based smart irrigation monitoring.

Method Claimed advantages Shortcoming Mode of work Monitored parameters Unmonitored parameters
[1] Automated inexpensive irrigation system. More parameters are required for precision in monitoring. Hardware field. Soil moisture, gas, light intensity. Temperature, humidity, wind speed, direction, rain and solar energy, pH.
[2] Consistent monitoring of current crop condition. Low cost sensors have to be used for cost effective irrigation. Hardware field. Temperature, wind speed, humidity, wind direction, solar radiation, rain. Soil moisture, soil pH, wind speed.
[3] Weather parameters are compared with past data to increase adaptability. Sensors data accuracy is needed to be analyzed. Hardware prototype. Soil moisture, Temperature, humidity, air pollutants. Soil pH, wind speed and direction, rain, solar energy.
[4] Soil condition is monitored for regulating water supply using Travelling Salesman. Sensors data accuracy and power consumption has to be analyzed. Hardware field. Soil pH, soil moisture. Temperature, humidity, wind speed and direction, rain, solar energy.
[5] Crop transpiration is also used with water balance to decide irrigation schedules. Accuracy of sensor data, power issues are not unaddressed. Hardware field. Soil moisture, temperature, humidity. Wind speed and direction, rain, solar energy.

Table 5.2 Details of communication protocols used with security vulnerabilities.

Method Protocol & N/W Frequency Data rate Range IoT layer with security attack
[1] WiFi–LAN. 92.4–5 GHz 54–600 Mb/s, 6.75 Gb/s 100 m Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
[2] Zigbee–LAN, GSM–LAN. 868–915 MHz, 2.4 GHz 850–1,900 MHz, 250 kb/s, 80–384 Kb/s 10–50 m,5–30 km MAC layer-DoS, jamming, eaves dropping, User Tracking. Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
[3] WiFi–LAN, GSM—LAN. 2.4–5 GHz, 850–1,900 MHz 54–600 Mb/s,6.75 Gb/s, 80–384 Kb/s 100 m,5–30 km Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
[4] Zigbee—LAN. 868–915 MHz 250 kb/s 10–50 m MAC layer-DoS, jamming, man in the middle, spoofing, routing attacks.
[5] GPRS—LAN. 850–1,900 MHz 80–384 Kb/s 5–30 km Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.

Table 5.3 Details of sensors, micro-controllers and data storage used by different methods.

Method Sensor & measuring range Micro-controller & communication interface Data storage
[1] Undefined Soil moisture, CO2, SO2, light intensity sensor. Raspberry Pi 3 with Serial Interface, Display Serial Interface. Undefined database.
[2] SHT 15-1,800 w/m2, Solar Pyranometer, wind speed sensor measuring wind speed from 3 to 125 mph. PIC18F2620 with Serial Interface, GSM Modem or Zigbee module, RS 232 Interface. Undefined database.
[3] DHT11-55–150 °C, 20–90%, MQ2-200–10,000 ppm, MQ135-10–300 ppb, MQ131-100 ppb/50 ppb, MQ9-500–10,000 ppm. Undefined microcontroller. Undefined server.
[4] Unnamed Soil moisture, temperature, humidity, sunshine, radiation, wind speed, rain fall sensor Unnamed central controller. Cloud database.
[5] – MCU ARM M3 kernel STM32F103 with I2C, SPI, UART. Weather database.

Table 5.7 gives an overview of irrigation algorithms with claimed advantages, shortcomings, Pivotal environment parameters and overlooked parameters. Mode of the work is also shown.

Table 5.4 Review of different IoT-based smart weather parameters monitoring.

Method Claimed advantages Shortcomings Mode of the work Monitored parameters Unmonitored parameters
[6] Economic and low power monitoring. More parameters have to be compared for precision in monitoring. Hardware field. Temperature humidity, wind direction Wind speed, direction, pressure, gas, soil moisture, UV radiation.
[7] Cost effective and adaptable monitoring. Accuracy of the sensor data and power consumption has to be taken care off. Laboratory Proto type. Temperature, humidity, wind, UV radiation, rain. Wind speed, direction, pressure, sunset, moon phase, UV radiation.
[8] Cost effective, flexible monitoring. System without net connectivity using renewable energy sources for rural area. Laboratory Proto type. Wind speed, wind direction and rainfall Temperature, atmospheric pressure, humidity, UV radiation.
[9] Cost effective monitoring. Accuracy of sensor data is needed & high power consumption. Laboratory Proto type. Temperature, humidity, pressure, rainfall, light intensity. Wind speed, wind direction and rainfall, UV radiation.
[10] Cost effective monitoring. Long-time weather forecasting for coverage of larger area has to be done. Simulation based. Temperature pressure, humidity, wind speed. Wind direction and rainfall, UV radiation.
[11] Flexible cost effective data acquisition in real time. Low power consumption and range of sensors has to be increased. Laboratory Proto type. Temperature, humidity, pressure, altitude. Wind speed, wind direction and rainfall.
[12] Real-time cost-effective monitoring. Data analysis from multiple sensors has to be done while minimizing power usage. Hardware field. Temperature, relative humidity, soil pH. Wind speed, wind direction, atmospheric pressure.

Table 5.5 Details of communication protocols used with security vulnerabilities.

Method Protocol and Network Frequency Data rate Range IoT layer with security attack
[6] Xbee-LAN WiFi-LAN. 868–915 MHz, 2.4 GHz, 2.4–5 GHz. 250 kb/s, 54–600 Mb/s, 6.75 Gb/s 10–50 m, up to 100 m MAC Layer-DoS, jamming, eaves dropping, user tracking. Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
[7] nRF24l01+-WAN 2.4 GHz 250 kbps–2 Mbps 100 m. Perception Layer-DoS, jamming, eaves dropping, User Tracking.
[8] Wi-Fi-LAN, GSM-LAN 2.4–5 GHz,850–1,900 MHz 54–600 Mb/s, 6.75 Gb/s, 80–384 Kb/s Up to 100 m, 5–30 km Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
[9] WiFi-LAN 2.4–5 GHz 54–600 Mb/s, 6.75 Gb/s Up to 100 m Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
[10] Zigbee-LAN 868–915 MHz2.4 GHz 250 kb/s 10–50 m MAC Layer-DoS, jamming, eaves dropping, user tracking.
[11] WiFi-LAN 2.4–5 GHz 54–600 Mb/s, 6.75 Gb/s Up to 100 m Transport Layer-DoS, jamming, man in the middle, spoofing, routing attacks. Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
[12] GPRS -LAN 850–1900 MHz 80–384 Kb/s 5–30 km Network access and physical layer-DoS, jamming, man in the middle attacks, spoofing, routing attacks.

Table 5.6 Details of sensors, micro-controllers and data storage.

Method Sensor and measuring range Micro-controller and communication interface Data storage
[6] Unnamed temperature, moisture, humidity, light, wind speed sensors. Raspberry Pi 2 model B. Cloud-MySQL database
[7] FC-37 −0 to 1,024, ML8511-280–390 nm, H21A1—ON and OFF, DHT-11-55–150°C, 20–95%. Arduino UNO ATmega8L with I2C, SPI, UART. –
[8] DHT22-10–40 °C, 0–99.9% RH, BMP180-300–1,100 hPa. Raspberry Pi with I2C, SPI, SERIAL, Arduino with I2C, SPI, UART. MySQL database
[9] – ESP8266-EX with I2C. Thing speak cloud
[10] BMP085-30 to 110 kPa , SHT21-0–100% RH. ArduinoUno with I2C, SPI, UART. –
[11] DHT11-55–150 °C, 20–95%, BMP180-300–1,100 hPa. Raspberry Pi B+ with I2C, SPI, SERIAL. –
[12] Unnamed Temperature, Relative Humidity, Soil pH Sensors. Raspberry-Pi 3. AWS

Table 5.8 gives an overview of communication protocols used to communicate sensor data to the cloud with security attack perception in IoT layered architecture used by different methods of machine learning based irrigation monitoring.

Table 5.7 Review of different machine learning-based smart irrigation monitoring.

Method Claimed advantages Shortcomings Mode of work Monitored parameter Unmonitored parameters
SVR and RF [13] Interconnected Sensor nodes used for economic monitoring. System needs to be trained on big data and has to be protected from climate adversities. Hardware field. Soil moisture, humidity, temperature. Wind speed, wind direction, pressure, gas, soil pH, UV radiation.
SVR and K-MEANS clustering [14] High precision with low error prediction of soil moisture. Soil moisture difference has to be compared for longer duration to improve accuracy. Hardware field. Soil moisture, humidity, solar radiation, temperature Wind speed, wind direction, pressure, gas, soil pH.
FL [15] Decentralized database with Block chain security. Accurate prediction of likelihood of a specific disease by measuring more parameters. Hardware field. Soil moisture, light intensity, humidity, air temperature. Wind speed, wind direction, pressure, gas, soil pH.
FL [16] Monitoring based on renewable energy. Flexible system is needed to suit each farm size. Hardware field. Soil moisture, humidity, temperature, water level. Wind speed, wind direction, pressure, gas, soil pH, UV rays.
RBF and FL [17] Real-time monitoring with minimum error. Inaccurate classification and precision is user dependent. Hardware field. Soil moisture, temperature, humidity, CO2 and UV intensity Wind speed, wind direction, pressure, soil pH.
LDA and SVM [18] Precision, economic, low powered implementation. Big data to be handled saving memory and with more parameters. Hardware field. Soil type, temperature, moisture, humidity, gas, UV intensity. Wind speed, wind direction.
Regression [19] Easily accessible, reliable parameter implementation. Overfitting of unconstrained parameters to remember the training data. Hardware field. Soil moisture, rain and temperature. Wind speed, wind direction, gas, solar radiation.
Feed Forward Network [20] Consistent monitoring using structural similarity index. No explanations are provided for parameters monitored. Hardware field. Soil moisture, temperature, humidity, gas, solar radiation. Wind speed, wind direction, gas, soil pH.
DTC [21] Precise monitoring done to predict fertilizer quantity and water requirement. Security and integrity of sensor data has to be ensured for accurate analysis. Hardware field. Soil moisture, Temperature, Humidity. Wind speed, wind direction, gas, soil pH.
GA [22] Consistent monitoring to optimize power and water supply. Exact location of water supply has to be determined regardless of the wind. Hardware field. Soil moisture. Temperature, Humidity, Wind speed, wind direction, gas, soil pH.

Table 5.8 Details of communication protocols used with security vulnerabilities.

Method Protocol and Network Frequency Data rate Range IoT layer with security attack
SVR and RF [13] WiFi-LAN, GSM-LAN, Bluetooth-PAN 2.4–5 GHz, 850–1,900 MHz, 2.4 GHz. 54–600 Mb/s, 6.75 Gb/s, 80–384 Kb/s, 25 Mb/s. 100 m, 5–30 km, Less than 10 m Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
SVR and K-MEANS clustering [14] ZigBee-LAN 868–915 MHz2.4 GHz 250 kb/s. 10–50 m MAC layer-DoS, jamming, man in the middle, spoofing, routing attacks.
FL [15] Wi-Fi Module ESP 8266–01-LAN 2.4–5 GHz 54–600 Mb/s,6.75 Gb/s. 100 m Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
FL [16] WiFi-LAN 2.4–5 GHz 54–600 Mb/s,6.75 Gb/s. 100 m Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
RBF and FL [17] Zigbee-LAN, WiFi-LAN 868–915 MHz, 2.4 GHz, 2.4–5 GHz 250 kb/s, 54–600 Mb/s, 6.75 Gb/s. 10–50 m, 100 m MAC layer-DoS, jamming, man in the middle, spoofing, routing attacks, network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
LDA and SVM [18] Zigbee-LAN 868–915 MHz, 2.4 GHz, 250 kb/s 10–50 m Low MAC layer-DoS, jamming, man in the middle, spoofing, routing attacks.
Regression [19] WiFi-LAN 2.4–5 GHz, 54–600 Mb/s, 6.75 Gb/s 100 m Low–high Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
Feed Forward [20] Zigbee-LAN WiFi-LAN 868–915 MHz, 2.4–5 GHz, 250 kb/s, 54–600 Mb/s,6.75 Gb/s 10–50 m,100 m Low, Low–high MAC Layer-DoS, jamming, man in the middle, spoofing, routing attacks. Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
DTC [21] Wi-Fi-LAN 2.4–5 GHz, 54–600 Mb/s, 6.75 Gb/s 100 m Low–high Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.
GA [22] WiFi-LAN, GSM-LAN 2.4–5 GHz, 850–1,900 MHz, 54–600 Mb/s, 6.75 Gb/s, 80–384 Kb/s 100 m,5–30 km Low–high, High Network access and physical layer-DoS, jamming, man in the middle, spoofing, routing attacks.

Table 5.9 details about the sensors with measuring range, micro-controllers with communication interfaces and data storage used by different methods of machine learning-based irrigation monitoring.

Table 5.10 gives an overview of weather parameter monitoring methods with claimed advantages, shortcomings. Pivotal environment parameters and overlooked parameters. Mode of the work is also shown.

Table 5.11 gives an overview of communication protocols used to communicate sensor data to the cloud with security attack perception in IoT layered architechture used by different methods of machine learning-based weather parameter monitoring.

Table 5.12 details about the sensors with measuring range, micro-controllers with communication interfaces and data storage used by different methods of machine learning-based weather parameter monitoring.

Figure 5.2 below shows the mode of work in IoT-based weather and irrigation monitoring. Figure 5.3 below shows the mode of work in ML-based weather and irrigation monitoring. Figure 5.4 below shows data storage used in IoT-based weather and irrigation monitoring. Figure 5.5 below shows data storage used in ML-based weather and irrigation monitoring.

From Figure 5.2 it is quite evident that most of the work in IoT solution based weather and irrigation monitoring mainly relies on hardware field implementation in real time. Whereas, very least work is done in software simulation.

From Figure 5.3 it is quite evident that most of the work in ML-based IoT solution based weather and irrigation monitoring mainly relies on Hardware field implementation in real time. Whereas, very least work is done in Software Simulation.

From Figure 5.4 we can see that in most of the methods, database names are not defined. In rest of the methods, data has been stored in Cloud platforms like ThingSpeak, AWS, Sensor Cloud etc. Very few methods relies on traditional database like SQLite, MySQL etc.

From Figure 5.5 we can see that data has been widely stored in Cloud platforms like ThingSpeak, AWS, Sensor Cloud etc rather than traditional database like SQLite, MySQL etc. In few methods, database name are not defined.

Cloud Storage is very economic compared to traditional Databases. Cloud is capable of storing data in a single data repository and chances of losing data are also minimized. Cloud storage though offers high flexibility in data storage, it is very much vulnerable to security attack when sensor generated data is communicated by means of IoT protocols in IoT layered architechture to the Cloud.

Table 5.9 Details of sensors, micro-controllers and data storage.

Method Sensor & measuring range Micro-controller & communication interface Data storage
SVR and RF [13] DHT11—20–90%, 55–150 °C, MQ2—200–10,000 ppm. Arduino Uno Mega 2560 Rev3 with I2C, SPI, UART, Raspberry Pi 3 B+ with I2C, SPI, SERIAL. Server/Database Cloud
SVR, K-MEANS clustering [14] VH-400-0-3v, DHT22—10–40 °C, 0–99.9% RH Raspberry Pi with I2C, SPI, SERIAL, Arduino Uno with I2C, SPI, UART. SQLite database, MySQL database
FL [15] DHT-11—20–90%, 55–150 °C, YL-69-0-1023. Arduino UNO R3 with I2C, SPI, UART. Plant database.
FL [16] Unnamed Soil moisture & Humidity/Temperature sensor, flow sensor with 0.5 power. ARM Processor with UART. SQL database
RBF and FL [17] Unnamed Soil moisture Humidity/Temperature, CO2 sensor. Raspberry Pi with I2C, SPI, SERIAL Wireless Gateway with embedded web server.
LDA and SVM [18] DHT11-20–90%, 55–150 °C, VH400-0-3v. Raspberry Pi with I2C, SPI, SERIAL. ThingSpeak
Regression [19] Unnamed moisture, temperature, rain, current sensor. Raspberry Pi with I2C, SPI, SERIAL. Sensor Cloud
Feed Forward [20] Unnamed Soil moisture, temperature/humidity Sensor, CO2 sensor. ATMEGA328 with I2C, SPI, SERIAL, Raspberry Pi 2 with I2C, SPI, SERIAL. MySQL database
DTC [21] DHT-11-20–90%, 55–150 °C, YL-69-0-1023. Arduino with I2C, SPI, UART. Thingspeak
GA [22] Unnamed Soil moisture sensor. Arduino UNO with I2C, SPI, UART. Sensor cloud

Table 5.10 Review of different machine learning-based smart weather monitoring.

Method Claimed advantages Shortcoming Mode of work Monitored parameters Unmonitored parameters
ANN and RBF [23] Energy and power independent monitoring. Faster computation of data is needed. Laboratory Prototype. Humidity, temperature and solar radiation. Wind speed, direction, pressure, gas, soil moisture, temperature, UV radiation.
NN [24] Inexpensive real-time monitoring. Data storage in a dedicated place is needed. Laboratory Prototype. Temperature, wind speed, wind direction. Air humidity, temperature, soil moisture and solar radiation.
Multiple linear regression [25] Inexpensive higher precision monitoring. Security, integrity of sensor data has to be ensured. Simulation based. Temperature, pollutants, humidity, pressure, rainfall, dust particles and light etc. Wind speed, direction, pressure, gas, soil moisture, temperature, UV radiation.

Table 5.11 Details of communication protocols used with security vulnerabilities.

Method Protocol and Network Frequency Data rate Range IoT layer with security attack
[23] Xbee-LAN 2.4 GHz 250 KB/s 10–100 m Physical layer-DoS, jamming, man in the middle, spoofing, eaves dropping, routing attacks.
[25] Wi-Fi-LAN 2.4–5 GHz 54–600 Mb/s, 6.75 Gb/s 100 m Network access and physical layer-DoS, jamming, man in the middle, eaves dropping, spoofing, routing attacks.

Table 5.12 Details of sensors, micro-controllers and data storage.

Method Sensor & measuring range Micro-controller and communication interface Data storage
[23] SP-110-0–400 mV, SHT75-0–100%. Raspberry PI version B with I2C, SPI, SERIAL interface. SQLite database
[24] CNY70-0.8–4.8 V, MCP9700-(−40 °C to + 125 °C). MC9S 12DG256. –
[25] MQ135-10–1,000 ppm, MQ7-20–2,000 ppm, DHT11-20–90%, 0–50 °C BMP180-300 to 1,100 hPa. Arduino Mega ATmega2560 with I2C, SPI, UART interface. ThingSpeak
Bar chart depicts the mode of work in IoT-based weather and irrigation monitoring.

Figure 5.2 Mode of work in IoT-based weather and irrigation monitoring.

Bar chart depicts the mode of work in ML-based weather and irrigation monitoring.

Figure 5.3 Mode of work in ML-based weather and irrigation monitoring.

Pie chart depicts the data storage in IoT-based weather and irrigation monitoring.

Figure 5.4 Data storage in IoT-based weather and irrigation monitoring.

Pie chart depicts the data storage in ML-based weather and irrigation monitoring.

Figure 5.5 Data storage in ML-based weather and irrigation monitoring.

Table 5.13 ML algorithms to handle security threats in IoT layered architecture.

ML methods IoT layers Security attack solution
ANN, RL [26] Physical/Perception Layer DoS, jamming, man in the middle, spoofing, routing attacks.
SVM, K means [26] Network Layer DoS, jamming, man in the middle, spoofing, routing attacks.
DT,RF,K means [26] Application Layer DoS, repudiation, eaves dropping, Blue snarfing and Bluejacking.

Machine Learning (ML)-based IoT solution can handle unwanted tampering and illegal access of these sensor data during transmission to cloud. Machine Learning (ML) has the capability to provide security handling attacks like DoS, jamming, man in the middle, spoofing, routing attacks, repudiation, eaves dropping, Blue snarfing and Bluejacking [26].

Table 5.13 below shows different Machine Learning (ML) algorithms capable of handling different types of security threats.

5.6 Challenges

  • A keen analysis of the literature reveals that weather- and irrigation automation requires particular attention towards unmonitored parameters. More number of parameters monitored more is the accuracy in optimizing irrigation schedules for improvising crop yield.
  • Cost analysis of different IoT-based solutions relying on sensors and embedded micro controllers has to be done to device an economic IoT-based smart farming approach as farmers have a very low income.
  • IoT layers in which various IoT communication protocols function are subjected to serious threats like altering data processing, illegal access by intruders, etc. Security mechanisms must therefore be incorporated to prevent infringement of sensor generated data during processing and storage.
  • In ML-based IoT security measures, correct identification of any attack is a burning issue since if any impostor identifies the type of attack than the training data set used for implementing ML solutions can easily be modified. Hence, the attackers can change their mode of attack and its consequences on the cloud network.

5.7 Conclusion and Future Work

This paper presents a thorough analysis of weather and irrigation parameters pivotal in the IoT-based solutions for smart farming. Frequently used communication protocols working in IoT layered architecture along with security attack perception is discussed. The susceptibility of IoT security threats is also shown. Most monitored weather parameters that are critical for optimizing irrigation schedules are also established. Cheaper sensors used to predict irrigation schedules accurately help farmers to minimize manual work unlike traditional methods while obtaining high-precision data. Costly sensors would have to be replaced with economical sensors in future to derive maximum benefit by the farmers using automated IoT solutions in commercial markets. Security mechanisms like machine learning algorithms shown above and keeping data at the edge of the network may be used to trigger an intervention using actuating devices upon threat identification using Big data for training purposes for connecting things using a secure network. Cost analysis of things (sensors) has to be done to implement smart inexpensive IoT-based solution for precision in agriculture monitoring.

References

1. Dhineesh, T., Rajvi Mohammed, K., Arunprasath, S., Haribaskar, M., Madhusudanan, G., Analysis of IoT based Wireless Sensors for Environmental Monitoring in Agriculture. Int. Res. J. Eng. Technol., 6, 610–614, 2019.

2. Fourati, M.A., Chebbi, W., Kamoun, A., Development of a web-based weather station for irrigation scheduling. Third IEEE International Colloquium in Information Science and Technology (CIST), pp. 37–42, 2014.

3. Choudhury, S. and Chattopadhyay, S.P., Smart irrigation: IoT-based irrigation monitoring system. Proceedings of the International Ethical Hacking Conference, Springer, Singapore, 2018.

4. Karunanithy, K. and Velusamy, B., Energy efficient Cluster and Travelling Salesman Problem based Data Collection using WSNs for Intelligent Water Irrigation and Fertigation. Measurement, 161, 107835, 2020.

5. Zhang, S., Wang, M., Shi, W., Zheng, W., Construction of intelligent water saving irrigation control system based on water balance. Proceedings of the International Federation of Automatic Control, pp. 466–471, 2018.

6. Tenzin, S., Siyang, S., Pobkrut, T., Kerdcharoen, T., Low cost weather station for climate-smart agriculture. Proceedings of the 9th international conference on knowledge and smart technology, pp. 172–177, 2017.

7. Solano, G., Lama, F., Terrazos, J., Tarrillo, J., Weather station for educational purposes based on Atmega8L. Proceedings of the 24th International Conference on Electronics, Electrical Engineering and Computing, pp. 1–4, 2017.

8. Brito, R.C. and Favarim, F., Development of low cost weather station using free hardware and software. Proceedings of the Latin American Robotics Symposium (LARS) and 2017 Brazilian Symposium on Robotics, pp. 1–10, 2017.

9. Kodali, R.K. and Sahu, A., An IOT based weather information prototype using WeMos. Proceedings of the 2nd International conference on Contemporary Computing and Informatics, pp. 612–616, 2016.

10. Saini, H. and Thakur, A., Arduino based automatic wireless weather station with remote graphical application and alerts. Proceedings of the International Conference on Signal Processing and Integrated Networks (SPIN), pp. 605–609, 2016.

11. Savic, T. and Radonjic, M., One approach to weather station design based on Raspberry Pi platform. Proceedings of the 23rd Telecommunication Forum, pp. 623–626, 2015.

12. Carlos, A.D.J., Rojas Estrada, L., Cardenas-Ruiz, C.A., Ariza-Colpas, P.P., Piñeres-Melo, M.A., Ramayo González, R.E., Morales-Ortega, R.C., Ovallos-Gazabon, D.A., Collazos-Morales, C.A., Monitoring system of environmental variables for a strawberry crop using IoT tools. Procedia Comput. Sci., 170, 1083–1089, 2020.

13. Vij, A., Vijendra, S., Jain, A., Bajaj, S., Bassi, A., Sharma, A., IoT and Machine Learning Approaches for Automation of Farm Irrigation System. Procedia Comput. Sci., 167, 1250–1257, 2020.

14. Goap, A., Sharma, D., Shukla, A.K., Krishna, C.R., An IoT based smart irrigation management system using Machine learning and open source technologies. Comput. Electron. Agric., 155, 41–49, 2018.

15. Munir, M.S., Bajwa, I.S., Cheema, S.M., An intelligent and secure smart watering system using fuzzy logic and blockchain. J. Comput. Electr. Eng., 77, 109–119, 2019.

16. Al-Ali, A.R., Al Nabulsi, A., Mukhopadhyay, S., Awal, M.S., Fernandes, S., Ailabouni, K., IoT-solar energy powered smart farm irrigation system. J. Electron. Sci. Technol., 17, 100017, 2020.

17. Mohapatra, A.G., Lenka, S.K., Keswani, B., Neural network and fuzzy logic based smart DSS model for irrigation notification and control in precision agriculture. Proc. Natl. Acad. Sci. India Sect. A: Phys. Sci., 89, 67–76, 2019.

18. Kabilan, N. and Selvi, M.S., Surveillance and steering of irrigation system in cloud using Wireless Sensor Network and Wi-Fi module. Proceedings of the International Conference on Recent Trends in Information Technology (ICRTIT), pp. 1–5, 2016.

19. Kumar, A., Surendra, A., Mohan, H., Valliappan, K.M., Kirthika, N., Internet of things based smart irrigation using regression algorithm. International Conference on Intelligent Computing Instrumentation and Control Technologies (ICICICT), pp. 1652–1657, 2017.

20. Keswani, B., Mohapatra, A.G., Keswani, P., Khanna, A., Gupta, D., Rodrigues, J.J., Improving weather dependent zone specific irrigation control scheme in IoT and big data enabled self driven precision agriculture mechanism. Enterp. Inf. Syst., 1–22, 2020.

21. Goldstein, A., Fink, L., Meitin, A., Bohadana, S., Lutenberg, O., Ravid, G., Applying machine learning on sensor data for irrigation recommendations: Revealing the agronomist’s tacit knowledge. Precis. Agric., 19, 421–444, 2018.

22. Roy, S.K. and De, D., Genetic Algorithm based Internet of Precision Agricultural Things (IopaT) for Agriculture 4.0. Internet Things, 100201, 2020.

23. Ruano, A.E. and Mestre, G., A neural-network based intelligent weather station. Proceedings of the IEEE 9th international symposium on intelligent signal processing (WISP), pp. 1–6, 2015.

24. Shaout, A., Li, Y., Zhou, M., Awad, S., Low cost embedded weather station with intelligent system. Proceedings of the 10th International Computer Engineering Conference, pp. 100–106, 2014.

25. Parashar, A., IoT Based Automated Weather Report Generation and Prediction Using Machine Learning. Proceedings of the 2nd International Conference on Intelligent Communication and Computational Techniques (ICCT), pp. 339–344, 2019.

26. Tahsien, S.M., Karimipour, H., Spachos, P., Machine learning based solutions for security of Internet of Things (IoT): A survey. J. Netw. Comput. Appl., 161, 102630, 2020.

*Corresponding author: [email protected]

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
3.149.254.35