Introduction
According to future of food and agriculture report of Food and Agriculture Organization of the United Nations (FAO), the world population is expected to increase by about 10 billion by 2050, which means that more agricultural production is needed (FAO 2017). To solve these problems, many researchers around the world are carrying out studies to increase agricultural productivity (Dhall and Agrawal, 2018; Verdouw et al., 2019). The plant grows in the limited space artificially adjust and control the environmental conditions such as humidity, airflow, temperature, CO2, light intensity. The plant production with innovative ideas and technological advances such as sensor system, wireless sensor network has been able to increase production and allocate more efficient resources (Ray, 2016). For example, farmers can integrate wireless sensors and mobile networks to control farming conditions in real-time. (Abd El-Kader and Mohammad El-Basioni, 2013; Isik et al., 2017). In addition, farmers through the technology can collect meaningful data, which is used to produce low-cost, high-quality crops in terms of precision agriculture (Vasisht et al., 2017). Each sensor and network system has advantages and disadvantages, and farmers can implement high-efficiency, low-cost by selecting the proper sensors and networks in consideration of their farm conditions and working environment. In addition, so far, the ambient environment management is used more in an individual field rather than the entire agricultural field, thus there is a need for improvement to expand plant production technology to a wide range of agriculture (Talavera et al., 2017; Tzounis et al., 2017). In particular, sensors used in precision agriculture have a major impact on crop growth and environmental control according to accuracy. Therefore, it is necessary to review existing sensors, network systems and technology for the extended application of plant production technology in agriculture.
Sensors of cultivation environment and plant growth monitoring
Various types of sensors were used for smart agriculture to collect data from multiple aspects of agriculture, such as the plant, substrate, environment and etc. Such sensors are used for collecting real-time data about multiple agricultural parameters, such as soil data, climate data, luminosity, CO2 concentration and images as shown in Table 1. Various soil sensors were used, and most were used to monitor the soil condition in real time through data measurement on temperature and humidity, or to perform environmental control under optimal conditions for plant growth in a greenhouse (Fisher et al., 2018; Im et al., 2018; Zhang et al., 2017). On the other hand, most of the air temperature and air humidity sensors used Arduino-based sensors. The Arduino-based sensor has the advantage of easy environmental control considering the factors that influence each other in the greenhouse, and in particular, optimal control according to the relationship between temperature and humidity is possible (Codeluppi et al., 2020; Mahmud et al., 2018; Mohanraj et al., 2016). Illuminance is one of the most important factors for plants in a greenhouse. An optical sensor module is used to control the brightness, and it can use artificial light can also be used to help plants to grow. Grow lights are used to stimulate plant growth where there is no sufficient sunlight or in places where daylight hours are less (Mohanraj et al., 2016; Syafarinda et al., 2018). The Low CO2 concentrations reduce the plant photosynthetic rate and affect plant growth. An excessively high CO2 concentration causes a partial closure of plant stomata, which affects plant photosynthesis. Therefore, many studies have used a sensor that output voltage value changes according to the concentration of CO2 to maintain or control the appropriate amount of CO2 (Gao et al., 2018; Mahmud et al., 2018). For plant growth monitoring, cameras and multispectral sensors were used to collect images of plants (Han et al., 2018; Xue et al., 2019).
Embedded system platforms for environment control and plant growth monitoring
Table 2 shows the embedded system used in the greenhouse. In general, it shows that Arduino is the most commonly used embedded system platform. This is because Arduino is an open source hardware that allows to develop a variety of devices using boards that extend the basic functionality. The Arduino Uno is an open source microcontroller board for IoT implementations. The microcontroller board is directly connected to multiple calibrated sensors to measure soil and environmental parameters (Cruz et al., 2018; Zervopoulos et al., 2020), ambient control (Erazo-Rodas et al., 2018; Fernández-Ahumada et al., 2019; Karimah et al., 2019; Mungai Bryan et al., 2019), and plant growth monitoring (Bhimanpallewar and Narasingarao, 2018; Montoya et al., 2017; Sarat Chandra and Srinivas Ravi, 2016)
Raspberry acts as an edge node that filters and preprocesses IoT data with the help of the TensorFlow Lite library using the server. This server collects IoT sensor data coming from IoT-Edge gateways, filters and removes possible noise, and discards repeated frames to avoid unnecessary transmission to cloud services. Recently, research related to intelligent edge-IoT platform control using Raspberry was performed, and soil management research was conducted through an expert system that integrated sensor network and artificial intelligence system to evaluate agricultural land suitability (Cruz et al., 2018; Zervopoulos et al., 2020). In addition, research related to environmental control using a relatively inexpensive Raspberry microcontroller was conducted (Campos et al., 2020; Rivas-Sánchez et al., 2019), and research related to the development of a precision agricultural monitoring system using wireless sensors and ubiquitous sensors was conducted (Alonso et al., 2020; Ferrández-Pastor et al., 2016; Navulur et al., 2017).
ESP is the module which is needed to be programmed using the micro-C with the AT+Commands. The serial terminal was used to send commands, these commands used to connect the module with a Wi-Fi router. This Wi-Fi router was acting as a medium between PC and embedded device (Sarat Chandra and Srinivas Ravi, 2016). ESP is mostly used based on Wi-Fi, and is mainly applied to the monitoring system within the greenhouse in consideration of the communication range (Aliev et al., 2018; Bhimanpallewar and Narasingarao, 2018; Sarat Chandra and Srinivas Ravi, 2016). However, when using a network that is used in a wide range such as LoRaWAN, it is possible to perform environmental control within a large greenhouse (Ali et al., 2019; Karimah et al., 2019).
Communication network protocols in greenhouse
Table 3 shows network protocols used for the IoT solutions of greenhouse. Data obtained with sensor nodes are usually sent to the destination through a wired or wireless network. Among the mentioned network protocols, CAN and Ethernet were the most used ones for wired networks (Laktionov et al., 2019; Radharamana et al., 2019). Likewise, LoRaWAN and protocols for cellular network were the most used protocols for long-range wireless networks (Codeluppi et al., 2020; Gao and Du, 2011; Im et al., 2018; Syafarinda et al., 2018). In general, ZigBee, Wi-Fi (Boonnam et al., 2017; Gao and Du, 2011; Li et al., 2020; Radharamana et al., 2019; Rodríguez et al., 2017; Sarat Chandra and Srinivas Ravi, 2016)and Bluetooth (Ferrández-Pastor et al., 2016; Kim et al., 2020; Steen et al., 2016)were the most used protocols for short and mid-range wireless networks. The maximum distance between sensor nodes in indoor agriculture enables different types of connections. Wired connections are used in small areas, while wireless connections are used in both indoor and outdoor agriculture. Wi-Fi is the most mentioned protocol within the analyzed projects due to its ubiquitous utilization in daily life.
Data processing technologies and application for smart agriculture
Table 4 shows commonly used technologies by application. It also shows that the most commonly used technologies to support data processing are artificial intelligence and big data. Plant monitoring is the type of application that uses the most diverse technologies for data processing (Adam Ibrahim Fakherldin et al., 2019; Liu et al., 2016; Patil and Sakkaravarthi, 2017; Xia et al., 2018). This is because IoT solutions for plant monitoring collect large amounts of data and rely on big data to process this data. Soil management handles a lot of soil data similar to plant monitoring. In particular, AI and big data are mainly used for continuous data acquisition through wireless sensors (González-briones et al., 2018; Jarolímek et al., 2019; Zervopoulos et al., 2020) and optimization of the greenhouse soil environment (Adenugba et al., 2019; Figueroa and Pope, 2017; Vincent et al., 2019). Fuzzy logic was used in IoT solution applications that need to handle multiple variables such as ambient control and plant monitoring (Alpay and Erdem, 2018; Liu et al., 2015). Fuzzy logic was used to process several temperature and humidity variables in the greenhouse and determine when to start the cooling system and ambient control system (Mungai Bryan et al., 2019). Similarly, fuzzy logic used to optimize the number of sensors to monitor soil temperature and moisture (Karimah et al., 2019)
Conclusion
This study reviewed the sensors, embedded system platforms, communication network technologies and the data processing technologies used in the greenhouse. Recent work has shown an increasing use of these systems as well as data management systems. Cloud and big data computing technologies for processing large amounts of data and data are mainly used for the management of the recent greenhouse. Additionally, recent studies have confirmed that the use of artificial intelligence technology to maintain or improve the environment of smart agriculture has become commonplace. Various sensors could be used simultaneously in conjunction with various network protocols for smart agriculture. A comparison of networks for smart agriculture also showed that wired networks (e.g., CAN; Ethernet) are used indoors in small area, whereas wireless networks (e.g., Bluethooth; ZigBee; WiFi; LoRaWAN; Celluar) are mainly used in large-sized greenhouses or when wireless control and real-time management are required from outside. Bluetooth and ZigBee have a relatively low date rate compared to WiFi, but low-power design is possible. LoRaWAN has low date rate, but it can communicate over a wide range. Cellular (e.g., 5G, 4G/LTE, etc.) can be used at the highest speed over the widest range, but consumes a lot of power. Bluetooth and ZigBee-based sensors are advantageous when measuring crop temperature and indoor environment in real time, and are mainly used when the main PC is configured within a maximum distance of 100 m. LoRaWAN and WiFi is used when fusion of various sensors in a larger space. Celluar is used to control the environment from inside or outside with a mobile phone.Therefore, it is very important to adopt the communication technology suitable for the agricultural environment. It also suggests the growing relevance of IoT solutions for smart agriculture. In the future, it is judged that it is necessary to conduct research related to the optimal design of a smart greenhouse in consideration of configuration cost and efficiency through IoT applications. In addition, it is necessary to develop technologies applicable to greenhouses through the expansion of big data and AI technologies mainly used for plant monitoring.
