Application of Deep Learning for Smart Farming

6.1 Introduction

Machine learning is the ability provided to the machine to enhance with experience. It provides the computer the ability to adjust itself to a changing environment so that the programmer does not need to think of all possible inputs. Since the last decade, AI and automated tools had been an area that has seen tremendous growth and popularity among many people. With the advancement of such technologies, life is becoming easier and machine learning provides the backbone to them. It helps in finding many solutions in vision, speech, database recognition, etc. It has been used to prevent fraud, the advancement of AI, medical sites, agriculture, etc. Machine learning has always been an area of keen interest for scientists as it provides them a hope that this tech can help them and creates a synthetic human brain or a super humanoid robot in the future. The practical application of machine learning can be seen in agriculture. India is known as the farmer’s nation and has benefited from the improvement of technologies for the agriculture sector. The farmers of our country are getting well adapted to these techs which are safe and reliable, and enhance the quality of life. Agriculture does not only restrict the growing of crops but also means the growing of flowers, the recreation of a forest, animal husbandry, etc. There has been the development of many machineries with the rise of machine learning and AI that covers various aspects of agriculture ranging from the sowing of seeds to harvesting of grains. This machinery not only makes the agricultural practices easier but also prevents occupational hazards such as hazards occurring due to grain bins, contact of chemicals, excessive noise, musculoskeletal injuries, overheat, and others, and maybe exposure to animals that can transmit diseases. A large number of animal-related diseases have been discovered and cured with advanced tools, but the challenge still exists. All the farmers cannot afford technical practices over the traditional methods of farming to reduce the chances of crop failure. Hence, it is a great matter of concern for the inventor and needs to think about the ideas that make these gadgets affordable for large- and small-scale farmers.

The Internet of Things (IoT) is commonly known as “IoT”, a new model that is transforming computing. It can be explained as the linkage between the individually distinguishable embedded computing instrument in the approachable internet framework. The “IoT” connects multiple different devices with the help of the internet as well as electronic sensors. Nowadays, many devices are connected to the web, whether it is mobile devices or appliances. There are many applications of IoT in different sectors:

• • Elder care
• • Smart city
• • Smart home
• • Medical and healthcare
• • Home automation
• • Agriculture
• • Product digitalization, etc.

The IoT’s biggest notable vogue in recent years is the increased number of gadgets connected to and supervised by the Internet. The large range of programs for IoT technology refers to that the details can vary from one device to other device but there are foundational features that are shared by most. The “IoT” generates chances for more straight combination of actual world into computerized systems which results in better improvements and reduced human efforts. The number of IoT devices has increased 31% every year to 8.45 billion in year 2017, and it has been calculated that there will be more than 30 billion devices by 2021. As we already know what is IoT and it is benefits now let us learn a bit about Green IoT. Green IoT represents the problem of bringing down energy consumption of IoT devices that attain a feasible environment for IoT systems [20]. Green IoT is identified as an essential skilled course in IoT to decrease the nursery impact brought about by existing applications. It is all about using the IoT to reduce the greenhouse effect and CO₂ emission.

When we are talking about the use of Green IoT in the environment, we mean using solar panels, IoT-based drones, soil moisture sensor, etc.

Benefits of Green IoT are as follows:

• • Better health
• • Healthier environment
• • Smart energy-efficient system
• • Cleaner water system
• • Smart farming

6.2 Literature Review

Table 6.1 closely contrasts the various techniques for managing farm and different farming activities to enhance the productivity level with quality.

6.3 Deep Learning in Agriculture

Deep learning refers to a subdivision of “Machine Learning” which is further a subset of Artificial Intelligence (AI). AI basically refers to any program that emulates the functions and working of the human brain. It works on various techniques and algorithms for creating different patterns and processing data which can be later used in process of decision making. It gradually selects useful data and information from the raw facts and figures. Deep learning AI is capable of learning and performing different tasks on various structured or unstructured data sets. The methods are based on artificial neural networks also known as neural networks. Artificial neural networks are computing structures that are roughly inspired by the human neural network. It is a collection of interconnected nodes also known as artificial neurons. Deep learning works on multiple layers in a network. Earlier the work was done on a linear perceptron but as the time passed, it was observed that new algorithms with more layers should be used. A modern alternative which was introduced for working on multiple layers was deep learning which grants practical application and advanced implementation. The data on which it is implemented is very vast and it is known as big data. This data can be found anywhere in different forms and structures from different zones of this world. Evaluating and comprehending this data and extracting the useful information manually can take decades, deep learning comes to rescue in these types of situations. Deep learning is a very powerful and effective tool whenever we are dealing with unstructured data. Deep learning algorithms easily become overfitted if the data we are working upon is very simple or incomplete, thus it fails to generalize the new data in a good manner.

“Deep learning” methods are based on “artificial neural networks” which can be grouped in the following types [18].

6.3.1 Feedforward Neural Network

“Feedforward neural network” is the easiest and first category devised of artificial neural network. In this type of neural network, the connections among the nodes do not form any cycle. It is totally different from the successor: recurrent neural networks (RNNs). The facts and figures progress in only a single direction forward from input nodes, across the hidden node and output node, in this network. No loops or cycles are formed in this network.

6.3.1.1 Single-Layer Perceptron

“Single-layer perceptron network” is the extremely basic type of neural network, which consists of a single (one) layer of output nodes; the inputs are sustained right to the outputs via a sequence of weights. In every node, the addition of the products of weights and the inputs is computed, and if the value is above doorway (typically 0), the neuron launches and takes the current activated value (typically, 1); else the disbanded value is taken (typically, −1).

Neurons with this type of simulation function are also known as false neurons or linear threshold units. A perceptron commonly known as algorithm that can be generated using any values for the currently activate states and deactivated states as long as the doorway value lies between the two.

Delta rule calculates the mistakes between calculated output and sample output data. A “single-layer neural network” computes a constant output instead of a single step function. A usual choice is the so-called logistic function:

The single-layer network is similar to the organizing regression model, commonly used in statistical modeling The derivative of this function is easily calculated:

If its activation function is modulo 1, then it can solve XOR problem with exactly one neuron.

6.3.1.2 Multi-Layer Perceptron

In this type of networks, there are many computational units’ layers that are connected to each other in a feedforward way. Every neuron in a layer has straight connections to neurons of the next layer. The units of these networks have a sigmoid function as a starting function. It uses a variety of learning approaches, and the most popular is back-propagation. In this, the output values are then contrasted with the actual right answer to compute the value of some already defined error-function, and after that, the error is then fed back again through the network. After repetition of this process for multiple times, the network normally converges to a state where the chances of mistake are less. To alter weights correctly, one applies a general method for non-linear development that is called “gradient descent.” The issue of the back-propagation algorithm is the speed of merging and the chances of ending up in a local minimum of the error function. Today back-propagation in multi-layer perceptron’s the tool of choice for various machine learning tasks.

6.3.2 Recurrent Neural Network

RNNs are a strong and robust type of neural network. It belongs to the most favorable algorithms in use as it is the only unique algorithm with an inner memory. RNN is the state of the artwork for consecutive.

In this, the relation among nodes forms a directed graph along a secular series.

This allows it to show secular dynamic behavior. Obtained from feedforward neural networks, RNNs use their inner memory to process length of a variable series of inputs. This makes them relevant to tasks such as non-segmented attached handwriting recognition or speech recognition. It is used by Apple’s Siri and Google’s voice search feature. It is the first algorithm that recalls its input, due to an internal memory. Over the past years, impressive achievements were seen in the field of deep learning.

The filtering units in RNN form a cycle. The output from layer 1 becomes the input for the upcoming layer, which is typically the sole layer; hence, the output of the layer becomes an input to it, setting up a feedback loop. It permits the network to have memory about the earlier states and use that to affect the present output. One notable result of this contrast is far from feedforward neural networks, and RNN can have a series of inputs and create a series of output values as well, giving it very useful in applications which require filtering sequence.

Here,

xt is the input at time t.

U, V, and W are learned parameters (shared by every step).

Ot is the output at time t.

St is the state at time t.

f is the activation function.

6.3.4 Kohenen Self Organizing Neural Network

Kohonen self-organizing neural network arranges the network model into the input data on its own using non manageable learning. In other words, it is also known as Kohenen self-organizing maps. It comprises two fully interconnected layers, known as input layer and output layer. The output layer is arranged as a 2D grid. There is no starting function, and the weights represent the characteristics (position) of the output layer’s node. The Euclidean interval among the input data and each output layer node are computed. The weights of the nearest node and its adjacent from the input are updated to bring them near to the input with the formula below:

Here, x(t) is the input data at time t;

wi(t) is the ith weight at time t, and ηj∗i is the adjacent function between the ith and jth nodes.

6.3.5 Modular Neural Network

A modular neural network is another artificial neural network that is distinguished by a continuous series of self-dependent neural networks decreased by some mediator. Each network obeys as a module and works on multiple different inputs to complete some subtask of the task in the network [1]. The mediator takes the results of each module and operates on it to generate the output of the network. The mediator accepts the outputs of the modules only—it does not respond nor otherwise signals. The modules do not interact with each other. Unlike one big network which can allocate to random tasks, each module in a modular network must be allocated a particular task and attached to other modules in particular ways by a designer. Example, the brain developed (rather than learned) to generate the lateral geniculate nucleus (LGN). In some cases, the generator can opt to go after biological models. In some cases, other models may be better. The standard of the output will be a function of the standard of the blueprint.

Food production is used to get processed through agriculture. History provides us the traces that our ancestor discovered the concept of civilization due to agriculture. In modern world, blending the trend, i.e., automated tools with traditional agriculture. Being from the land of farmers and farmlands, we come across a wide variety of crops that are grown in respective seasons. Agriculture sector shares 16% of total GDP of India and provides employment to around 60% of the population. But sometimes, the blessing turns out to be curse for the farmers. There are a wide variety of challenges farmers face such as climatic disasters, pests, quality of seeds, and land quality, and here, we are working on using the modern machinery methods to solve these issues. Agriculture might be seemed to be an easy task for an individual living in a high-tech city, but the reality just opposes this whole mirage of thoughts. The cycle mentioned below showcases the various steps taken to ensure the healthy crop production and quality. Figure 6.1 shows the distinctive crop cycle applicable across the world [18].

Each of these steps participates equally in the crop production and must be performed within the stipulated time and in almost accurate number of resources in use. Taking example of fertilizing, fertilizer is something which is a major resource and factor for a good quality of crop, but it should be applied during a phase of the crop growth that too in the adequate amount. Otherwise, it may lead to acidification of soil ultimately turning a fertile land into a barren one. Each of the crops required a certain type of climatic conditions and soil type to grow depending upon the nutrients and nourishment required by them for growth. Even, distance between the seeds while sowing plays a vital role so that there should be no conquest between plants for nutrients. Moreover, in last couple of years, some part of our nation especially regions of Maharashtra had been adversely affected by the shortage of rainfall that too leading to crop failure. As per the Jal Shakti Ministry, 42% of fertile land turned into barren due to shortage of rainfall and inappropriate irrigation methods since independence. On the other hand, farmers in Kerala adopted the modern farming methods that includes the machines that provides crops enough water that too in a proper way without any wastage. There are certain regions where sowing of seeds is done through drones developed using machine learning and there are many examples that prove that the agricultural sector needs to get even more help from researchers and investment to meet the requirements of the global population.

6.4 Smart Farming

The management using techs to increase the quantity and quality of agricultural products is known as smart farming. In the 21st century, many of the farmers have access to GPS, soil and weed scanning, and other technologies. With the change in climatic conditions, it is now important to precisely measure the variations for the farmers to improve the effectiveness of pesticides, fertilizers, and use them more selectively. In the field of animal husbandry, smart farming will help farmers to know about the needs of animals and adjust their nutrients. In the past years, there are certain techs developed which can measure the fodder quality which is provided to animals. All these developments are leading to the third green revolution using the advancement of genetics and application of ICT solutions such as exactness tools, the IoT, sensors and actuators, geo positioning systems, big data, unmanned aerial vehicles (UAVs, drones), and robotics. The experience proved that smart farming has huge possibility to carry a more productive and sustainable agricultural production. Smart farming provides key to save the resources from getting over exploited and ultimately leading to complete degradation. For example, if an excessive number of fertilizers are being added to soil, then it enhances the acidic content in it degrading its quality. Here comes the role of the new techs which calculates the number of fertilizers required over an area of land. Advancement in drone or aerial techs helps in sowing seeds too in a proper way by maintaining adequate distance among the plants which prevents clash of nutrients among them. It also provides a better geographical and territorial study of the land and helps farmers to decide the kind of crop and cropping pattern they should follow to maximize the yield. The farming industry will become more important than before in the upcoming years. The world needs to produce 70% more food in 2050 than we used to in 2006, to feed the growing population of Earth, according to the UN Food and Agriculture Organization. To meet this demand, farmers and agricultural companies are moving to the IoT for analytics and greater production capabilities. Figures 6.2 and 6.3 graphically represent applications of IoT technologies in agriculture.

6.5 Image Analysis of Agricultural Products

The following are the summarized points of previous research work done by different authors with publication references in Green IoT.

• • Author Rushen Arshed found an investigation on energy saving practice. Its advantage is that it was about recycling and reusability of hardware components in order to reduce the impact of greenhouse effect [21].
• • Faisal Karim found an Enabling technology for green IoT. It was a survey of the green perspective of IoT. It also gives a clear picture regarding the Green IoT [22].
• • Kamalesh Sharma’s smart system for universities. It gives an idea of controlling electrical devices from anywhere. In this, the sensor will sense if no one is present in the classroom so that the accountable person can switch the lights on and off from anywhere using a device [23].
• • Prabhu Ramaswamy introduced an idea for smart parking IoT-based systems for reducing greenhouse gas emission. For that, the idea was to use a Raspberry Pi which will sense the free slot and send the details to the driver who has that application [24].
• • Sheetal Valuri’s idea was for greenhouse using IoT and cloud computing. The technology will sense the atmosphere and farmers can cultivate crops according to it with the information [25].
• • Afgen Syeda proposed IoT by energy harvesting and trends and techniques for Green IoT. It accumulates energy from the surroundings and uses it for charging rechargeable batteries [26].

Image analysis has provided computers the ability to look in this world even more effectively and learn to adapt to the changes with respect to the environment. Emerging from the facial recognition system on mobile phones to the agricultural data collection, identification of weeds, soil type, and image analysis proved to be very vital in the agricultural sector. Through the imaging methods with various spectra such as infrared, hyper spectral imaging, and X-ray helps in deciding the vegetation indices, canopy measurement, irrigated land mapping, etc., with more accurate figures. Weeds which are the unwanted plants are said to be the biggest threat for the crops, can be correctly classified with the image processing algorithms. The approach of image analysis helps to save the cost and environment, but its accuracy of classification varies from 85% to 96% depending on the algorithms and limitations of image acquisition. The image processing techniques and better communication methods are being adopted by the agriculture sector and the data generated by them has increased with passage of time. The availability of multidimensional imaging combined with modern algorithms and increased possibility to fuse multiple sources of information from satellite imagery and sensors planted in fields. The major points to be taken care in agriculture are water stress, quality of yields and the use of pesticides. The satellite imaging provides the data regarding the amount of irrigation being done in an area over time providing additional means to analyze and monitor irrigation which provides cost benefits to the farmers. Another point is to be noted that large amount of water is as deadly for the plant as a shortage of water. Weeds in farms can also be detected by combining image processing and machine learning techniques. With development of herbicide applications and green color recognition algorithms, data regarding the texture of plants are integrated to enhance identification accuracy and helps in setting up farm management strategy.

6.6 Land-Quality Check

Traditionally, farmers have difficulties in calculating the soil’s nitrogen level and dampness level of agricultural fields. These calculations will help them to evaluate and take necessary steps to get maximum productivities for their crops. Deep learning can be usable for measuring various parameters such as nitrogen and moisture status to convert the traditional farming techniques in smart farming methods with more accuracy and reliability.

6.7 Arduino-Based Soil Moisture Reading Kit

Many people are fond of gardening and planting small plants and even have nurseries, but as we all know that nowadays, the weather keeps on changing rapidly, in the morning, it seems hot, in the afternoon, it is raining, and at night, we feel cold so we are not able to identify that what type of clothes should we wear similar is in the case of plants, people are not able to decide that whether they should water the plants or not or how much amount of water is required by a particular type of plant. So, to overcome this problem, we can use soil moisture sensor which will take the reading from the soil and whenever the water level will low it will send a message to the person that “The water level is getting low, please water the plant” or whenever the water level increases due to rain or overwatering it will give a message “water overflow please store our plant in a dry place or save your plant.” Whenever the user will water the plants when humidity is high and moisture in soil is already up to the required mark it will give the message that “Do not water the plants as the water level is already up to the required amount.” Here, we are proposing the Arduino-based soil moisture reading kit. The major idea is to create a product that can overcome the problems listed below in an easy and cheaper way. Figure 6.4 represents prototype of the proposed soil moisture reading kit.

6.8 Conclusion

The days of hardship and non-profitable farming will be remembered as past if the modern machineries are used with efficiency and made affordable even to small-scale farmers. All these innovations have been supported by machine learning which enhances their skills and accuracy every day. The basic aim is to enhance the number of products without compromising the quality. A facial recognition system was first introduced as a security technology, but with the period, its applications increased. In today’s digitalized world, it can be used to provide support to AI and machine learning to identify animals, to analyze their health and behavior analysis. It enhances the capabilities of the computer to create a proper database for a group of animals. Expanding such techs will completely change the scenario of agricultural and animal farms. An example related to such kind of technology can be seen in china. Companies like Alibaba manage pig farms using AI. Integrating all the requirements gathered by the various sensors, habits, and health of each pig is collected. They also use a voice-recognition system to trace diseases such as pig cough. Traditional methods of gathering data about the various crop healths are vivid and time-consuming. This led to the emergence of techs which are important to be intelligent enough to monitor, detect, and analyze the field of various data to study the status and feasibility of growing certain types of crops. The backbone of certain types of automated systems will be based on machine learning, hyperspectral images, and 3D laser scanning which will enhance the data collection and accurately provide the details about the crop disease if any found. This all will significantly minimize the use of pesticides and other insecticides which are harmful to the land as well as for consumers. Since the idea of robotics emerged several tech giants have invested in it. The evolution of robotics in each field has shown a positive sign that they have a vast capability to replace humans in each activity, including agriculture. Equipment such as drones, smart tractors loaded with sensors, radars, and GPS systems does not need human supervision. All this enhances the productivity and quality of living of a farmer.

6.9 Future Work

After having some research done, it is found that there are still many progresses needed in the machineries, and level of AI and machine learning still needs upgrades to work with even more precision and accuracy to eliminate the margin of errors while performing various functions such as, in image analysis, the embedded program or the AI sometimes gets confused among the crops and the weeds due to lack of proper skills that provides them the ability to differentiate between them. The main issue that stands in the center of all these inappropriate functions is the unavailability of a universal database that can be fed and updated in these devices with variance in the environment around them to make them work even more efficiently. A global database of farming can be created by taking the help of farmers from all around the globe and that will certainly make computers even smarter for farming related activities. This can also help gaining proper numbers of crop failures and other important data and providing a proper community for farmers to interact and exchange their ideas. Even in different regions, water cycle varies from place to place, and if the machine is connected to a proper global database, it knows how much it needs to dig up to achieve the desired water level preventing wastage of water and time and would be even more cost efficient for the farmers. A universal flu detector can be very useful for the farmers involved in animal farming as it can help them to exchange the ideas about the flu among them and help them to cure and take proper actions regarding them. As per the data, around 31.08% of the farmers directly or indirectly related with animal husbandry must suffer loss due to delay in detection of disease in their farm animals. As a result, it sometimes turns out to be a drastic pandemic not only among the animals but even to the humans consuming them and the products made of them. Having accurate data of various regions will help the computer to detect sudden rainfall or other calamities that can affect the crop quality adversely. It will help farmers to set up the action plans in case of emergency to reduce the crop damage. In India, various crops are destroyed due to Vermin, nuisance animals which get in the field and destroy crops, rather than pests. The farms in our country firstly require a good alarm system that makes the farmer aware if any vermin gets inside the farm. It must also contain a night vision camera to know about the animal as most of these animals are nocturnal. A proper and specialized team of people should be made to remove these animals to a safer place. Many of the small-scale farmers still have no idea about such advanced techs of agriculture and on the same side many of this machinery are electric or fuel driven which makes it cost effective for the farmers to afford them. To solve this issue, solar energy should be promoted which is cheap and even more reliable environment friendly.

References

1. Herbut, E., Modern animal production and animal welfare. Agric. Eng., 22, 3, 5–10, 2018.

2. Singh, R.K.P., Singh, K.M., Kumar, A., A study on adoption of modern agricultural technologies at farm level in Bihar. Economic Affairs, 60, 1, 2015, 49–56, 2014.

3. Rehman, A., Jingdong, L., Khatoon, R., Hussain, I., Iqbal, M.S., Modern agricultural technology adoption its importance, role and usage for the improvement of agriculture. Life Sci. J., 14, 2, 70–74, 2016.

4. Masrie, M., Rosman, M.S.A., Sam, R., Janin, Z., Detection of nitrogen, phosphorus, and potassium (NPK) nutrients of soil using optical transducer, in: 2017 IEEE 4th International Conference on Smart Instrumentation, Measurement and Application (ICSIMA), 2017, November, IEEE, pp. 1–4.

5. Singh, N. and Shaligram, A.D., NPK Measurement in Soil and Automatic Soil Fertilizer Dispensing Robot. Int. J. Eng. Res. Technol., 3, 7, 635–637, 2014.

6. Kulkarni, Y., Warhade, K.K., Bahekar, S., Primary nutrients determination in the soil using UV spectroscopy. Int. J. Emerg. Eng. Res. Technol., 2, 2, 198– 204, 2014.

7. Garvade, L.C., Detection of N, P, K using fiber optic sensor and PIC controller. Int. J. Eng. Sci., 13787, 2017.

8. Ramane, D.V., Patil, S.S., Shaligram, A.D., Detection of NPK nutrients of soil using Fiber Optic Sensor, in: International Journal of Research in Advent Technology Special Issue National Conference ACGT, 2015, February, pp. 13–14, 2015.

9. Yusof, K.M., Isaak, S., Rashid, N.C.A., Ngajikin, N.L.E.D., Based Soil Spectroscopy. Bul. Optik, 3, 1–7, 2016.

10. Dan-Qun, H., Zhen, L., Hou, C.J., Jun, Y., Xiao-Gang, L., Huan-Bao, F., Jun-Jie, L., Recent advances on optical detection methods and techniques for cell-based microfluidic systems. Chin. J. Anal. Chem., 38, 9, 1357–1365, 2010.

11. Kulkarni, Y., Warhade, K.K., Bahekar, S., Primary nutrients determination in the soil using UV spectroscopy. Int. J. Emerg. Eng. Res. Technol., 2, 2, 198– 204, 2014.

12. Masrie, M., Rosman, M.S.A., Sam, R., Janin, Z., Detection of nitrogen, phosphorus, and potassium (NPK) nutrients of soil using optical transducer, in: 2017 IEEE 4th International Conference on Smart Instrumentation, Measurement and Application (ICSIMA), 2017, November, IEEE, pp. 1–4.

13. Cull, P.O., Irrigation Scheduling–Techniques and Profitability, in: National Irrigation Convention Proceedings, Melbourne, Australia, pp. 131–170, 1992.

14. Patil, G.L., Gawande, P.S., Bag, R.V., Smart agriculture system based on IoT and its social impact. Int. J. Comput. Appl., 176, 1, 0975–8887, 2017.

15. Kumar, A., Payal, M., Dixit, P., Chatterjee, J.M., Framework for Realization of Green Smart Cities Through the Internet of Things (IoT), in: Trends in Cloudbased IoT. EAI/Springer Innovations in Communication and Computing, F. Al-Turjman (Ed.), Springer, Cham, 2020.

16. Mathur, S. and Arora, A., Internet of Things (IoT) and PKI-Based Security Architecture, in: Industrial Internet of Things and Cyber-Physical Systems: Transforming the Conventional to Digital, Hershey, Pennsylvania, pp. 25–46, IGI Global, 2020.

17. Sehgal, A. and Mathur, S., Plant Disease Classification Using SOFT COMPUTING Supervised Machine Learning, in: 2019 3rd International conference on Electronics, Communication and Aerospace Technology (ICECA), 2019, June, IEEE, pp. 75–80.

18. Shrestha, A. and Mahmood, A., Review of deep learning algorithms and architectures. IEEE Access, 7, 53040–53065, 2019.

19. Buhmann, M.D., Radial basis functions: theory and implementations, vol. 12, Cambridge university press, Cambridge, England, 2003.

20. Tahiliani, V. and Dizalwar, M., Green IoT systems: An energy efficient perspective, in: 2018 Eleventh International Conference on Contemporary Computing (IC3), 2018, August, IEEE, pp. 1–6.

21. Jeyasheeli, P.G. and Selva, J.J., An IOT design for smart lighting in green building based on environmental factors, in: 2017 4th International Conference on Advanced Computing and Communication Systems (ICACCS), 2017, January, IEEE, pp. 1–5.

22. Shaikh, F.K., Zeadally, S., Exposito, E., Enabling technologies for green internet of things. IEEE Syst. J., 11, 2, 983–994, 2015.

23. Sharma, K. and Suryakanthi, T., Smart system: IoT for university, in: 2015 International Conference on Green Computing and Interne of Things (ICGCIoT), 2015, October, IEEE, p. 15861593.

24. Ramaswamy, P., IoT smart parking system for reducing greenhouse gas emission, in: 2016 International Conference on Recent Trends in Information Technology (ICRTIT), 2016, April, IEEE, pp. 1–6.

25. Vatari, S., Bakshi, A., Thakur, T., Green house by using IOT and cloud computing, in: 2016 IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT), 2016, May, IEEE, pp. 246–250.

26. Adila, A.S., Husam, A., Husi, G., Towards the self-powered Internet of Things (IoT) by energy harvesting: Trends and technologies for green IoT, in: 2018 2nd International Symposium on Small-scale Intelligent Manufacturing, 2018, April.

1. *Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.; This email address is being protected from spambots. You need JavaScript enabled to view it.