1.1 Introduction
The increasing global population demands quality food while maintaining the food quantity and the environment. It is estimated that the global population will be approximately 9 billion by 2050 and hence food production should increase by 70 per cent [1]. In developed and underdeveloped countries, the demand for the food is completed from agriculture-based products. Sustainable agriculture has a remarkable prospect in the field of agricultural production to fulfil the food requirements of the global population by maintaining the environmental ecology. In sustainable agriculture, advanced technologies are incorporated to detect plant stress, which has a direct impact on the quality and quantity of the agriculture product [2]. The growth of a plant in an unsatisfactory environment may be the cause behind the plant stresses. The effects of stresses can lead to deficiencies in growth, permanent damage, or death of plant; this reduces the quality as well as the quantity of the food production [3]. Plant stress factors are generally divided into two sets: biotic (includes living biological factors such as fungi, bacteria, virus, insects, parasites, weeds, etc.) and abiotic (includes non-living environmental factors such as light, water, temperature, drought, flood, nutrient deficiency, and other environmental factors) [4, 5]. In agriculture, environmental and biological factors play an important role in the growth, development, and productivity of the plant. Every plant gives good productivity if stress factors are within appropriate limits. Any unexpected changes in environmental and biological factors can cause deficiencies or damages in plant productivity and these changes may affect several parts of the plant, such as root, leaf, etc. [6]. It is observed that every biotic and abiotic stress consists of threshold values for the plant, and under this threshold value, good productivity can be expected [3, 7]. These unexpected or critical changes in the environmental or biological factors can exceed the threshold levels and it is then the main cause behind the deficiencies in productivity [8]. Usually, the identification and monitoring of the plant stresses are done by farmers with the naked eye, but nowadays detection of plant stresses due to biotic factors is also done using an Internet of Things (IoT)-based intelligent system [9, 10]. In this respect, IoT-based intelligent systems in sustainable agriculture is the most significant way to identify plant stresses as quickly as possible, and it is possible to decide correct and accurate amount of pesticide or another remedy [11–13].
Currently, IoT intelligent systems are used widely in the agricultural domain. Advancement and existing challenges present in IoT will make it more popular in the future [1]. An IoT-based intelligent system performs the commendable job of continuous monitoring in the concerned agriculture field and making predictions and decisions regarding the plant stresses of the field [14]. This advanced system has collected the raw data or information from numerous disparate systems. As per system requirements, these data are processed, segmented, and systematized by some smart algorithms [15]. In another process, some advanced algorithms are applied in agriculture for analyzing the data to identify and detect the early threats or warnings that are produced by the IoT-based prediction system [16]. If a system generates any threat or warning, then an alert is created automatically and an appropriate solution or remedy is suggested to the agriculture field-monitoring objects such as farmer, agriculture robot, agriculture vehicle, etc [17]. An IoT-based intelligent system is cost-effective because it works on low power and stores the data into distributed databases. Further, analyzed or processed overheads by these systems are effortlessly transferred to the cloud, where more devices are connected in a distributed manner, and, thus, computational speed is increased [1, 18].
In the agriculture field, identification and monitoring of plant stresses with the naked eye is very tedious and leads to low-quality, insufficient production. For this reason, the IoT-based intelligent system is introduced to sustainable agriculture to detect or identify the plant stress. The survey of this chapter covers the IoT-based hybrid intelligent techniques which help to identify, detect, or predict the plant stress in agriculture production. The other objective of this survey is to perform a comparative study on different machine learning (ML) methods which helps to observe and analyze different plant stress parameters. The rest of this survey is organized through the discussion of IOT and intelligent system techniques for processing an object. First, the survey concentrates on the IoT structure and its detail layer-based discussion. In Section 1.3, methodology to perform this survey is described. The sub-section consists of the machine learning process for agriculture with a detailed view of each phase of data processing. Section 1.4 depicts the importance of IoT based intelligent system for plant stress detection. Finally, the survey concludes with the future scope.
1.2 Organization of IoT
The Internet of Things is an extremely favourable hybrid technological system in the present world that provides various activities like sensing, monitoring, controlling, actuating, etc [19–21]. It is represented through physical and virtual systems, which are interconnected with the internet. Data are collected in the IoT systems from numerous objects like physical devices and transferred to the server instances through high-speed internet. The virtual system or instances of cloud performs various computational analysis on these data to create appropriate decisions. In IoT, each device either physically or virtually is connected to the system is known as “Things.” Examples of these physical devices are digital cameras, numerous sensors, etc. The virtual system or resources in the cloud service help us with its machine learning services. These services are basically associated with a user application from which the user may get the required information [22–25]. In this study, Figure 1.1 represents a layer structure of IoT for agriculture systems. The layers are perception or sensing layer, data transfer or network layer, and data storage and manipulation layer or application layer. In the perception layer, data are collected from various near-field sources like cameras and sensors or sensor-based devices. Image-based data are also collected or fed into the perception layer like weather station data, historical data, statistical reports, etc. In the network layer, these data are transferred to the database instance of the cloud system via numerous network devices and gateways. In the final layer, all system intelligence techniques are applied, with high computation and decision-making taking place [1, 15, 17, 19, 23].
Figure 1.1 Layered structure of IoT.
- Layer 1: Perception Layer—In the agriculture sector, this layer consists of numerous sensors and cameras. Different types of sensors and cameras are collecting various types of raw data from the agriculture field. Agriculture field sensors are collecting data related to humidity, ...
