With a burgeoning population over the world, food creation and cultivating needs to get progressively gainful and prepared to do exceptional returns in confined time. As per the UN Food and Agriculture Organization, the world should create 70% more food in 2050. To satisfy this need, ranchers and farming organizations should push the advancement furthest reaches of their present practices. Similarly as the Industrial Revolution took cultivating to the following level during the 1800s, arising advances, and even more fundamentally, the Internet of Things (IoT) is scheduled to similarly affect the horticultural business future. This progress from farming to agronomy is currently basic, to guarantee to put food on the tables of everybody around the globe while dodging ridiculous time and work necessities. This chapter gives a short study on IoT sensors utilized in horticulture and presents a survey of different machine learning (ML), artificial intelligence (AI), and deep learning (DL) strategies used to take care of the current issues of ranchers. With the examination of the study, troubles in past methodologies and recommending a superior answer for the current issues in the field of agribusiness are discovered. The review unveils that there are more prospects in recognizing an issue explicit model in a completely computerized way.
Keywords: Smart agriculture, IoT in agriculture, sensor statistics in agriculture, artificial intelligence, machine learning, deep learning, precision farming, artificial intelligence in agriculture
Farming is considered as the significant wellspring of food grains and other crude materials. It is considered as the life of the living species. The biological limits, for instance, temperature, sun controlled radiation, relative tenacity, soil sogginess, and wind, impact the effectiveness of infrequent harvests, as they impact grain design and procure development. Drained soils, poor storerooms, and lacking information for ahead of schedule recognizable proof of plant sicknesses influence the rural efficiency. Modernizing the horticulture makes the harvest upkeep simpler. Brilliant agriculture alludes to utilizing mechanization and IoT in agribusiness. Besides, site-specific agriculture is the idea of utilizing novel advancements and gathering field data. The fundamental exercises of site-specific agriculture are information assortment, preparing along with variable rate uses of data sources. Moreover, IoT innovation is right now molding various parts of human existence. Exactness farming is solitary of the standards which can utilize the IoT focal points to enhance the creation effectiveness over the horticulture fields, upgrade the nature of the yields, and limit the negative ecological effect . Soil readiness is the initial step prior to growing a harvest. A definitive target is to create a dense and sans weed seedbed for quick germination and development of the yield. One of the main undertakings in soil readiness is plowing whirling the dirt with slackening it. In addition, soil planning is one of the mainly energy burning-through pieces of agribusiness and necessitates critical contributions of fuel and time. Contingent upon the field’s area, it might likewise expand the danger of soil disintegration. Nowadays, precision cultivating gear exists that encourages ranchers to utilize impressively a smaller amount fuel and time in soil planning by improving the precision, productivity, and also manageability of the cycle as shown in Figure 8.1.
8.2 Literature Survey
8.2.1 Internet of Things in Field of Agriculture
With the quick advancement of IoT purpose in the cloud stage, the number of connected gadgets has expanded in an exceptionally rapid way. It has been understood that the gadgets are further than the individuals on the Earth in 2011. Also, the associated gadgets are relied upon to arrive at 24 billion by 2050. These gadgets are associated by means of cloud stages for various purposes. IoT and distributed calculate working in joining makes another worldview, which have been named as Cloud of Things (CoT) . In CoT, IoT things are stretched out from sensors to each front-end thing in the Internet. Furthermore, disseminated locales are associated with the overall body, for instance, a keen house, savvy manufacturing plant, perceptive city, and brilliant planet. In view of CoT, a legitimate engineering of a keen city is given. With the convergence of cloud stage and IoT, CoT is needed to improve the capacity for gigantic gadgets intuitive with interoperability in order to help keen as well as clever applications. At the position when the quantities of gadgets are growing, diverse information and direction will be associated through CoT. For further information and assets in a single viewpoint on cloud or existing IoT application, CoT will provide more contemplation in a business indulgent. The issues acknowledged with coordination of IoT with disseminated computing that entail a savvy door to cooperate out the rich errands with preprocessing, in which conventional sensors are not equipped for satisfying the task of information accumulation and processing, are shown in Figure 8.2.
Astute water framework  is good for giving the water to the entire ground reliably, schedule, and reins the water deftly indirectly so that each plant has the adequate proportion of water it needs, neither a ton of nor exorbitantly slight, Water use capability in the ground can be directed by field water efficiency rate.
WT is water efficiency is and WA is water added to ground, where WF is water efficiency. The architecture is intended to develop a programmed water system framework that adjusts the siphon engine ON/OFF to detect the dirt dampness material. The structure is intended to detect the dirt dampness substance. The organization of remote sensors has created comprehensive robotization events. This helps to obtain subtleties in the turf from assets such as water, soil, or air and analyze them to get the option to enhance the yield and spare assets. Inconsequential MQTT distributor endorser agreement is used for correspondence because of which harvests would be gracefully tested for soil dampness, mugginess, temperature, and water to crops. Observing these limits will help to understand and supervise land assortments for visible capitulation. For additional indication, the natural boundaries are put away in DynamoDB. To inform the client/ rancher, AWS entryway with AWS Lambda is used. The advantage of using this method is that individual intercession is reduced and an adequate water system is still assured. Figure 8.3 demonstrates a genius irrigation framework using CloudIoT.
The Green IoT (G-IoT) is expected to introduce considerable changes in everyday life and would facilitate realize the apparition of “green surrounded intelligence”. Within the minority years, we will be surrounded by an immense amount of sensors, devices, as well as “things”, which will be capable to communicate through 5G, act “intelligently”, and supply green support for users in supervising their tasks . G-IoT inducts cutting-edge sensor-based technologies for real-time accomplishment of micro-climatic data from farmers’ field, timely measurement of crop situation, and personalized agro-advisory services to farmers.
The distinctive remote correspondence advancements have different reaches , which should be represented when planning the IoT arrangement, along with different factors, for example, information rate, power utilization, correspondence conventions, or expenses. In arable cultivating, because of the bigger homestead sizes and due to the work of portable sensors and gadgets on vehicles, this test turns out to be significantly more basic. Besides, depending on the inexact correspondence scope of a remote innovation can be misdirecting, for example, Wi-Fi is regularly depicted to have 100 m reach, however, a test investigating the parcel conveyance proportion regarding separation to door shows bundle misfortunes at ≥ 60 m. Notwithstanding the decision of remote technology, network geography in WSNs, for example, work geographies, can likewise build the correspondence range by utilizing hubs to speak with the focal hub. Contrasts in correspondence conventions can root specialized interoperability, which can prompt problems of compatibility and similarity between the equipment and programming used. Organization conventions are isolated into assorted layers shaping a convention stack, where errands are partitioned into more modest advances. In the foundation layer, some remote guidelines that characterize correspondence conventions are generally utilized by various remote technologies, e.g., IEEE 802.15.4, which is used by, among others, ZigBee, 6LowPAN, or 3GPP, which is used by, among others, GPRS, LTE, or 5G. Standards such as HTTP, MQTT, or XMPP are usually used in IoT applications within arable cultivation in the application layer, as shown in Figure 8.4.
As indicated by the applications that concentrate on QoS (Quality of Administrations) and push the board, the sensors for IoT purposes are relegated to unique functionalities. The administration of sensor arrangements is expected to advance regional control as well as energy usage and increase the length of the sensor center. In WSN, the LEACH shown in Figure 8.5 calculation was used to bring the organization together into groups in which each bunch contains a bunch head and a few group individuals are using the situation of WSN-based IoT applications consisting of sensor hubs, switches, and a descend. This modified the measurement of LEACH  to preserve geography to ensure the ideal degree intended to save the hub’s transmission intensity. By sending HELLO messages and receiving the REPLY message, and afterward, benevolent and ideal decision of community individuals, the system learns the evolving organizational interface qualities.
The updated LEACH calculation aims to achieve the ideal K degree hub and also aims to ensure the organization’s solid availability with the ideal geography chosen. The estimation involves two stages: the stage of setup and the stage of consistent state. The calculation is done consecutively in all sensor hubs at the arrangement level, which collect data from their neighbors. The sensor hubs will select the bunch head depending on the spread power and ideal degree of the hub at that point. At the consistent stage of conditions, the data is actually transmitted between the sensor hubs and the sink. The group head is selected by the degree of its ideal center, separation from the sink hub, and lingering energy. The optimal degree of hub depends on the full utilization of resources or the group head’s remaining energy. Expect that the head of the bunch is only affiliated with individuals of the K group, and the number of groups in the organization can be determined by
The length of broadcasting expression can be controlled by
In the setup process, each hub looks for the group leader, and the hub can choose to be a CH if there is no CH in the organization. Each hub will produce an arbitrary number selected anywhere in the range of 0 and 1 in the modified LEACH. If the irregular number is not as much as edges Pthr, the hub becomes CHs. The hub will communicate the HELLO message else, it will hang tight for the HELLO message. The limit to pick a group head is characterized by
The head of the bunch is selected by its ideal degree of hub, separation from the downward hub, and remaining energy. The optimal degree of hub depends on the use of all out energy or the remaining strength of the head of the bunch. Accepting that the bunch head only interfaces with individuals of the K group, the number of groups in the organization can be determined by the number of groups
The length of broadcasting expression can be controlled by
In Table 8.1, detailed analyses of the top level of IoT in cultivation are provided. Configuration, level, and geology of the agricultural network help to enable IoT spine and facilitate farmers to enhance the effectiveness of the reap. Similarly, this chapter offers a comprehensive overview of current and ongoing developments in IoT agricultural applications, devices/ sensors, topologies, and protocols with various inventive developments that have been presented.
8.2.2 Machine Learning With Internet of Things in Agriculture
A forecasting engine has been developed that is important for an IoTenabled frost prediction framework , which accumulates natural information to anticipate ice occasions utilizing artificial intelligence (AI) procedures. Expectation ability outflanks current proposition regarding affectability, accuracy, and F1 were appeared. Specifically, the utilization of SMOTE as shown in Figure 8.6 during the preparation stage has demonstrated an improved execution as far as review in both RF and Logistic Regression models. In explicit applicable cases, the incorporation of neighbor data assists with improving the accuracy or review of the estimated grouping model were noticed. Then again, relapse models have fewer errors, as well as neighboring data. In these instances, there is a corresponding enhancement in model execution with the spatial ties.
Table 8.1 Review on IoT in agriculture.
|1||2020||Anusha Vangala, Ashok Kumar Das, Senior Member, Neeraj Kumar, Mamoun Alazab ||Investigations of various cryptographic natives using the commonly known “Multiprecision Integer and Rational Arithmetic Cryptographic Library (MIRACL)” to estimate the usual time needed for primitives such as Texp, Tecm, Teca, Tsenc = Tsdec, Th, Tmul, Tadd, Tecsiggen, and Tecsigver to indicate the time required for a “bilinear blending”, a “measured exponentiation”, and a “elliptic bend point”||MIRACL is a C/C++ programming language-dependent cryptographic library that includes the “Elliptic Curve Cryptography Open Source SDK”.|
|2||2018||Ahmed, A.N.; de Hussain, I.D ||1. In the current WSNbased solutions for covering longer ranges, implement the WiLD network with fog computing With a less important pause. |
2. To minimize the delay, a cross-layer–based MAC with a routing response is used.
|Wi-Fi–based long distance (WiLD) network with links up to 100 km can easily provide connectivity and is used efficiently to associate the provincial areas. With the allinclusive scope, haze registration and distributed computing arrangements can be integrated into Wi-Fi gadgets for better and proficient IoT in such locations.|
|3||2019||Sudhir K. Routray, Abhishek Javali, Laxmi Sharma, Aritri D. Ghosh, Anindita Sahoo ||The ongoing structures of PA utilizing IoT were shown how it can help the agrarian areas of the agricultural nations in the long haul. The significance of IoTbased PA for better return. We mostly center the need of an IoT-based PA with regards to the agricultural nations are given||Precision agriculture (PA) is a way of working with ranch executives that employs data innovation (IT) to ensure that harvests and soil get what they need for optimum comfort and profitability in particular. PA seeks to ensure the benefit, manageability, and protection of the climate.|
|4||2018||Mohamed Abdel Basset1, Laila A. Shawky1, Khalid Eldrandaly ||For upgrading the zone inclusion level of WSN, an enhanced met heuristic calculation named multi-verse optimizer via overlapping detection stage (DMVO) is presented. The proposed calculation is attempted on various datasets with different data sets. Standards and contrasts and different calculations were discussed, as well as the first MVO, MVO, optimization of particle swarm, along with the algorithm of flower pollination.||Multiverse optimization (MVO) algorithm is an amazing metaheuristic calculation dependent on laws of material science. Notwithstanding, it is anything but difficult to fall into a nearby ideal when taking care of complex multimodal advancement issues with high measurements.|
|5||2019||Tesfa Tegegne, Hailu Beshada Balcha, and Mebiratu Beyene ||ZigBee remote sensor organizations can achieve self-sorting remote information transmission, which has been commonly used in large-scale farming. RFID (radio-frequency identification) invention. RFID technology. It is usually used in creature distinguishing evidence that can be performed on creatures to perform astute perception, position, follow, observation, discernibility, and the managers are examined.||Radio transmission technology is transmission and discovery of correspondence signals consisting of electromagnetic waves that move through the air in an orderly fashion. |
RFID is an innovation that utilizes radio waves to inactively distinguish a labeled object.
Regression models record the Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). If, as stated in the test set, Yreal denotes the real minimum and Ypred denotes the expected form value, the metrics are defined as follows:
For classification models, we can compute the following metrics:
Sensitivity: True Positive / True Positive + False Negative
Precision: True Positive / True Positive + False Positive
F1-score: F1 = 2 × precision × sensitivity / precision + sensitivity.
In order to achieve easy, reliable, and accurate agribusiness, drones equipped with suitable cameras, sensors, and coordination modules would make it easier. When combined with various  ideas for machine learning (ML) and the IoT, drones will help to extend the scope of additional enhancement, as shown in Figure 8.7.
For the grouping and examining of information that is transferred in the cloud, an uphold vector machine or SVM is utilized, which is an administered learning model, incorporated with AI calculations that chiefly centers around relapse and order issues. The fundamental target of the SVM is to prepare a model with the end goal that it allocates the new items to a particular classification. It begins by displaying the circumstance which makes a component space (vector space of limited measurement) wherein each measurement portrays a “highlight” of a specific article. SVM chooses the most ideal arrangement shown in Figure 8.8. The SVM can likewise be utilized in accuracy horticulture utilizing UAV.
IoT and ML-based ways to deal with soil dampness  levels ideal for the development of yields, regardless of the climate circumstances for the following 24 hours, were engaged. A savvy water system framework helping in appropriate water the board and giving ideal harvest recommendations dependent on noteworthy soil condition information and furthermore give the amounts of minerals should have been added to the dirt were clarified.
PART  classification procedure is proposed for crop profitability and dry spell prediction. This technique ends up being generally precise in giving dry season expectation just as the efficiency of harvests like Bajra, Soybean, Jowar, and Sugarcane. The WPART technique accomplishes the most extreme precision contrasted with the current incomparable standard calculations, and it is acquired up to 92.51%, 96.77%, 98.04%, 96.12%, and 98.15% for the five datasets for dry spell classification and crop profitability individually. Moreover, the proposed technique beats existing calculations with exactness, affectability, and F-score measurements. Given preparing vectors Xi ∈ Rn, i = 1, …, l, and a name vector Y∈Rl, a choice tree recursively parcels the space with the end goal that the examples with similar marks are gathered. Leave the information at hub m alone spoken to by Q. For every up-and-comer split θ = (j, tm) consisting of a component j and limit tm, segment the information into and subsets QL(θ) and QR(θ).
The contamination at m is registered utilizing impurity function H (), the decision of which depends on the errand being tackled (classification or regression).
Table 8.2 Review on ML with IoT in agriculture.
|1||2020||Yemeserach Mekonnen, Srikanth Namuduri, Lamar Burton, Arif Sarwat, and Shekhar Bhansali ||A far reaching audit of the utilization of various AI calculations in sensor information investigation inside the rural environment. It further talks about a contextual analysis on IoTbased information driven brilliant ranch models as an incorporated food, energy, and water (FEW) framework.||Used the model of classification and regression trees (CART) to identify possible indicators of crop yield, FEW communications, and yield efficiency. Use the model of autoregressive integrated moving average (ARIMA) forever arrangementbased sensor data.|
|2||2016||Suyash S. Patil1, Sandeep A. Thorat ||The Hidden Markov Model is accustomed to observing frameworks which will distinguish the odds of grape illnesses in its beginning phases.||Hidden Markov Model (HMM) is a factual Markov model in which it is assumed that the structure being displayed is a Markov cycle with undetectable (“covered up”) states. Well accepts that there is another cycle whose conduct “depends” on The objective is to find out about by noticing.|
|3||2020||Uferah Shafi, Rafia Mumtaz, Naveed Iqbal, Syed Mohammad Hassan Zaidi, Syed Ali Raza Zaidi, Imtiaz Hussain,  and Zahid Mahmood||A coordinated methodology for observing yield well-being utilizing IoT, AI, and robot innovation is used. NDVI gives data about the harvest in light of the chlorophyll content, which offers restricted data with respect to the yield well-being. To acquire rich and itemized information about yield wellbeing, In order to construct crop well-being charts, the variable length time arrangement information of IoT sensors and multispectral images was changed to a fixed-size representation.||The Normalized Difference Vegetation Index (NDVI) is a simple graphical marker that can be used frequently from a space stage to break down distant detecting estimates, surveying whether live green vegetation is included in the goal being noticed.|
|4||2017||Rob Dolci ||In order to verify how CO2, temperature, humidity, and PH differ, AI algorithms such as Bayesian network analysis and multi-variant analysis are used.||Bayesian networks are a sort of probabilistic graphical model that utilizes Bayesian deduction for likelihood calculations. Bayesian organizations mean to display restrictive reliance, and hence causation, by speaking to contingent reliance by edges in a coordinated chart.|
Table 8.2 summarizes different ML algorithms used to solve agriculture related problems like odds of grape illnesses, food, energy, and water (FEW) framework and other such problems with the integration of IoT devices.
8.2.3 Deep Learning With Internet of Things in Agriculture
Codling  moth assault is the most widely recognized issue for apple plantations. Neural organization calculations of IoT sensors will naturally identify the codling moth: the system snaps a picture of the snare, preprocesses it, crops every creepy crawl for grouping, and, if any codling moth is detected, finally, sends a note to the rancher. The application is built on a low-energy stage fueled by a board of a few hundred square centimeters centered on the sun, recognizing an energy self-governing system suitable for consistently operating unattended over low-force large-area organizations. A quick part of this IoT arrangement is the low force stage for an AI calculation utilized for IoT quick prototyping. The machine relies on the Raspberry Pi 3 board and the Intel Movidius Neural Compute Stick , as shown in Figure 8.9, liable for the preprocessing strategy and the neural organization usage, separately.
On a Raspberry Pi 3 that provides the preprocessing step, the framework is implanted. At that point, an Intel Movidius neural register stick (NCS) with an Intel Myriad X neural quickening agent as a dream handling unit (VPU) groups the images obtained after the deep neural organization (DNN) preparation shown in Figure 8.10 using the model.
Table 8.3 summarizes different deep learning (DL) algorithms used in different agriculture related applications like hydroponics, soil health, and other such applications with the coordination of IoT gadgets.
To create an IoT structure for crop fine-grained sickness ID, IoT with DL is used. As a result, this system will classify crop infections and give ranchers indicative results. We propose the remaining neural organization (MDFC-ResNet) model for multidimensional portion remuneration for fine-grained disease distinguishing evidence in the system. From three metrics, in particular, organisms, coarse-grained disease, and fine-grained infection, MDFC-ResNet recognizes and creates a pay layer that uses a remuneration calculation to meld multidimensional recognition outcomes. Tests show that the neural organization MDFC-ResNet  has a stronger effect on recognition and is more enlightening than other popular DL models in genuine horticultural development exercises. MDFC-ResNet is an agricultural IoT device to accurately identify crop diseases, and MDFC-ResNet shown in Figure 8.11 recognizes species, coarse-grained sickness, and fine-grained disease from three measurements, in particular, and sets up a pay layer that uses a remuneration calculation to intertwine multidimensional recognition effects.
Two standards are embraced anticipating the reasonable harvest for the following yield pivot and ad libbing the water system arrangement of the field by particular water system. The above objective is accomplished by occasionally observing the field. The observing cycle includes gathering data about the dirt boundaries of the field. To collect this information and have knowledge of the past, a remote sensor organization (WSN) is set up by uploading it to the cloud inconsistently. This information conveyed forms the justification for the inquiry. Long-Short Term Memory (LSTM) through experimentation, RNN, and GRU  networks shown in Figure 8.12 are discovered to be the appropriate calculation. The derived outcomes are contrasted and the ideal qualities and the most appropriate yield are suggested to the client through SMS administration.
Table 8.3 Review on DL with IoT in agriculture.
|1||2019||Author: Kirtan Jha; Aalap Doshi; Poojan Patel; Manan Shahd ||Deep learning algorithms like ANN to identify the soil moisture.||This system presented an idea to develop a system with IoT and ML, to automate the traditional practices in agriculture|
|2||2019||E. Alreshidi ||A holistic forum for IoT/AI to cover all areas in the SSA ecosystem (Smart Sustainable Agriculture) to perform tasks like govern data flow, integration of components, sustainable storage, etc.||To fix issues arising from the fragmentary nature of the agricultural method, the AI/IoT framework for SSA is used.|
|3||2020||Disha Garg; Samiya Khan; Mansaf Alam ||DNN is used for hydroponic system development (multiple input parameters)||A comparison between two algorithms to make predictions for agricultural applications on agricultural data. |
DNN = 88% accuracy
Deep CNN = 96.3%
A shrewd agribusiness IoT framework dependent on profound support realizing which incorporates four layers, to be specific horticultural information assortment layer, edge registering layer, farming information transmission layer, and distributed computing layer, is introduced. The introduced framework coordinates some serious data procedures, particularly computerized reasoning and distributed computing, with horticultural creation to build food creation. Uncommonly, the most exceptional computerized reasoning model, profound support learning is joined in the cloud layer to settle on prompt keen choices, for example, deciding the measure of water should have been inundated for improving harvest development climate. A few delegate profound models learning models like progressive neural network are shown in Figure 8.13; single-task policy distillation and multi-task policy distillation are shown in Figure 8.14; and with their extensive applications that we explained, various memorybased deep Q-network is shown in Figure 8.15.
Agribusiness, similar to other businesses, is going through a computerized change. The measure of information being gathered from ranches is expanding dramatically. The utilization of remote sensor organizations, IoT, advanced mechanics, robots, and AI is on the rise. AI calculations allow useful data and bits of knowledge to be extracted from the information storm. The ML techniques commonly used by analysts in the previous 2 years relevant to remote sensor organizations were audited by this chapter. An increased use of further evolved processes such as distributed (or edge) DL may be seen in the coming years. In order to extend the computerization of errands in agribusiness and boost the yield while advancing the use of routine properties, computer-based intelligence must be used. This chapter has shown various models of IoT, ML, and DL implemented within the exact biological framework of agribusiness in numerous applications. The checked on work has been centered explicitly around IoT, conventions, cloud, ML, and DL-based exactness cultivating application. The engineering, equipment, correspondence convention, and information securing foundation are nitty gritty. The usage of cell phone applications and, furthermore, the back-end information investigation structure for forecast of climate, crop yield, and harvest quality, just as illness detection, irrigation, use of robots, and so forth, are introduced.
1. Dewi, C. and Chen, R.-C., Decision Making Based on IoT Data Collection for Precision Agriculture, Department of Information Management, Chaoyang University of Technology, Taichung, Taiwan, Republic of China. Asian Conference on Intelligent Information and Database Systems Springer Nature Switzerland, ACIIDS 2019: Intelligent Information and Database Systems: Recent Developments, pp 31-42, AG, 2020.
2. Cai, H., Xu, B., Jiang, L., Vasilakos, A.V., IoT-based Big Data Storage Systems in Cloud Computing: Perspectives and Challenges. IOT-0592-2015,2327-4662 (c) 2016 IEEE IEEE Sens. J., IOT-0592-2015,2327-4662 (c) 2016.
3. Raikar, M.M., Desai, P., Kanthi, N., Bawoor, S., Blend of Cloud and Internet of Things (IoT) in agriculture sector using lightweight protocol. IEEE Sens. J., 978-1-5386-5314-2/18, 2018.
4. Villa-Henriksen, A., Edwards, G.T.C., Pesonen, L.A., Green, O., Sørensen, C.A.G., Internet of Things in arable farming:Implementation, applications, challenges and potential, Biosystems Engineering, 191, 60-84, March 2020, Elsevier, 2019.
5. Nguyen, T.N., Ho, C.V., Le, T.T.T., A Topology Control Algorithm in Wireless Sensor Networks for IoT-based Applications. IEEE Sens. J., 2019.
6. Vangala, A., Das, A.K., Kumar, N., Alazab, M., Smart Secure Sensing for IoT-Based Agriculture:Blockchain Perspective. IEEE Sens. J., 1558-1748 (c) 2020.
7. Saha, A.K., Saha, J., Ray, R., Sircar, S., Dutta, S., Chattopadhyay, S.P., Saha, H.N., IOT-Based Drone for Improvement of Crop Quality in Agricultural Field. IEEE8th Annual Computing and Communication Workshop and Conference, 2018.
8. Goapa, A., Sharmab, D., Shuklab, A.K., Rama Krishnaa, C., An IoT based smart irrigation management system using Machine learning and open source technologies, Elsevier, Computers and Electronics in Agriculture, 155, 41-49, December 2018.
9. Rezk, N.G., Hemdan, E.ED., Attia, AF. et al. An efficient IoT based smart farming system using machine learning algorithms. Multimed Tools Appl, 80, 773–797, 2021.
10. Syed, F.K., Paul, A., Kumar, A., Cherukuri, J., Low-cost IoT+ML design for smart farming with multiple applications. 10th ICCCNT IIT, IEEE, Kanpur, 2019.
11. Brunelli, D., Albanese, A., d’Acunto, D., Nardello, M., Energy Neutral Machine Learning Based IoT Device for Pest Detection in Precision Agriculture, IEEE Xplore, IEEE Internet of Things Magazine, 2019.
12. Aruul Mozhi Varman, S., Baskaran, A.R., Aravindh, S., Prabhu, E., Deep Learning and IoT for Smart Agriculture using WSN. IEEE International Conference on Computational Intelligence and Computing Research, 2017.
13. Vangala, A., Das, A.K., Kumar, N., Alazab, M., Smart Secure Sensing for IoT-Based Agriculture: Blockchain Perspective. IEEE Sens. J., 1558-1748 (c) 2020, 2020.
14. Ahmed, A.N. and de Hussain, I.D., Internet of Things (IoT) for smart precision agriculture and farming in rural areas. IEEE Internet Things J., 5, 6, 2018.
15. Routray, S.K., Javali, A., Sharma, L., Internet of Things Based Precision Agriculture for Developing Countries. Second International Conference on Smart Systems and Inventive Technology, IEEE Xplore, 2019.
16. Abdel-Basset, M., Shawky, L.A. & Eldrandaly, K. Grid quorum-based spatial coverage for IoT smart agriculture monitoring using enhanced multi-verse optimizer. Neural Comput & Applic, 32, 607–624, 2020.
17. Mekonnen, Y., Namuduri, S., Burton, L., Sarwat, A., Bhansali, S., Review— Machine Learning Techniques in Wireless Sensor Network Based Precision Agriculture. J. Electrochem. Soc., 167, 037522, 2020.
18. Patil, S.S. and Thorat, S.A., Early Detection of Grapes Diseases Using Machine Learning and IoT. Second International Conference on Cognitive Computing and Information Processing, IEEE, 2016.
19. Boursianis, A.D., Papadopoulou, M.S., Diamantoulakis, P., Liopa-Tsakalidi, A., Barouchas, P., Salahas, G., Karagiannidis, G., Wan, S., Goudos, S.K., Internet of Things (IoT) and Agricultural Unmanned Aerial Vehicles (UAVs) in smart farming: A comprehensive review, Elsevier, Internet of Things; Engineering Cyber Physical Human Systems,16,100187, 2020.
20. Dolci, R., IoT solutions for precision farming and food manufacturing Artificial Intelligence applications in Digital Food. 41st Annual Computer Software and Applications Conference, IEEE, 2017.
21. Jha, K., Doshi, A., Patel, P., Shahd, M., A comprehensive review on automation in agriculture using artificial intelligence, KeAi, Artificial Intelligence in Agriculture, 2, 1-12, 2019.
22. Alreshidi, E., Smart Sustainable Agriculture (SSA) solution underpinned by Internet of Things (IoT) and Artificial Intelligence (AI), arXiv, International Journal of Advanced Computer Science and Applications, 10, 5, 2019. 93-102, 2019.
23. Garg, D., Khan, S., Alam, M., Integrative Use of IoT and Deep Learning for Agricultural Applications, Department of Computer Science, Jamia Millia Islamia, New Delhi, India P. K. Singh et al. (Eds.): Proceedings of ICETIT 2019, LNEE 605, pp. 521–531, 2020, Springer Nature Switzerland AG, 2020.
24. Al-Turjman, F., Kamal, A., Rehmani, M.H., Radwan, A., Pathan, A.-S.K., The Green Internet of Things (G-IoT). Wireless Commun. Mobile Comput., 2019, Article ID 6059343, 2 pages, 2019, https://doi.org/10.1155/2019/6059343.