A Review of IoT and Machine Learning for Environmental Optimization in Aeroponics

Abstract

Traditional farming practices are becoming increasingly inadequate to meet global food demand due to water scarcity, prolonged production cycles, climate variability, and declining arable land. In contrast, aeroponic, smart, soil-free farming technologies offer a more sustainable alternative by reducing land use and providing efficient water use, given that aeroponics intermittently delivers water in mist form rather than maintaining continuous root zone moisture. However, aeroponics faces critical challenges in irrigation management due to non-standardized structures and limited real-time control. A key limitation is the inability to dynamically respond to temperature (T), relative humidity (RH), light intensity (L_i), electrical conductivity (EC), pH, and photosynthesis rate (P_n), resulting in suboptimal crop yields and resource wastage. Despite growing interest, there remains a research gap in integrating internet of things (IoT) and machine learning technologies into aeroponic systems for adaptive control. IoT-enabled sensors provide real-time data on ambient conditions and plant health, while ML models can adaptively optimize misting intervals based on the fluctuations in P_n and environmental inputs. These technologies are particularly well suited to address the dynamic, data-intensive nature of aeroponic environments. This review purposes a novel, standardized IoT–ML framework to control irrigation by emphasizing IoT sensing and ML-based decision making in aeroponics. This integrated approach is essential for minimizing water loss, enhancing resource efficiency, and advancing the sustainability of controlled-environment agriculture.

1. Introduction

In recent decades, the global population has grown rapidly, reaching approximately 8.2 billion people and projected to rise to approximately 9.2 billion by 2050 [1]. According to the Food and Agriculture Organization, this growth will drive a 70% increase in global food production demand [2,3,4]. As food demand escalates, optimizing water use becomes increasingly crucial, particularly in agriculture, accounting for approximately 70% of global freshwater usage [5]. Simultaneously, expanding urban development is converting arable land into residential areas, further reducing available land for cultivation [6]. Conventional irrigation practices, often used by farmers, typically disregard crop-specific water requirements and field variability [7]. Consequently, these methods are insufficient at conserving available water resources [8]. Water scarcity poses a significant threat to agricultural productivity, contributing to issues such as insufficient irrigation and declining soil fertility [9,10]. This scarcity is exacerbated by a number of factors, including climate change, drought, environmental disturbances, and conflicting crop water demands [11,12]. Addressing these challenges requires a contemporary irrigation management approach that ensures precise timing and volume control [13,14]. The modernization of irrigation techniques can significantly improve efficiency and crop yields by minimizing water waste. In recent years, researchers have explored innovative solutions, such as aeroponics, which offer promising alternatives for conserving water and enhancing crop yield [15].

Aeroponics is a soil-free smart farming technique that uses significantly less water than conventional farming practices [16,17]. According to studies by the National Aeronautics and Space Administration, aeroponics can reduce pesticide usage by 100%, water consumption by 98%, and fertilizer usage by 60% [18,19,20]. Additionally, this technique enhances crop yields by 45–75% compared with traditional farming, while requiring less manual labor [21,22]. In addition, experiments based on aeroponics allow researchers to study plant behavior under different conditions effectively, as each stage of plant growth can be precisely monitored [23]. Previous studies have shown that aeroponics requires minimal land use due to the portability of the equipment [24]. In aeroponics, plant roots are suspended in the air inside a growth chamber and intermittently misted with nutrient-rich solutions [25,26]. The roots remain in complete darkness under controlled conditions to promote optimal development [27]. Aeroponics employs either a low- or high-pressure pump to deliver water and nutrients directly to the roots [28]. With artificial lighting and precise environmental control, aeroponics enables year-round cultivation regardless of external weather conditions. This adaptability also supports various agricultural research applications [29].

Despite the several advantages of aeroponics, a significant challenge lies in the lack of a standardized irrigation strategy, which can lead to either overwatering or underwatering, both harmful to plant health and longevity [30,31]. Consequently, several studies have explored the optimization of irrigation in aeroponics by incorporating various parameters. For instance, one study developed an automated irrigation controller using the Wemos D1 Mini, which adjusts watering duration and integrates remote monitoring to prevent overwatering or underwatering. It utilizes temperature (T), relative humidity (RH), and light intensity (L_i) sensors to monitor the aeroponic chamber and activate the pump every five minutes [32]. Although the system operates solely on electricity, it lacks irrigation adjustment based on specific crop water requirements [33]. Similarly, another study proposed an Arduino Mega-based automatic irrigation controller for optimal water usage. The system activated the pump for 15 s (s) in every 400 s interval and monitored the aeroponic chamber using T, nutrient solution level (N_sl), flow rate (Q), pH, and electrical conductivity (EC) sensors [34]. However, key parameters such as CO₂ concentration, photosynthesis rate (P_n), transpiration rate (E), and L_i, which are critical for efficient irrigation and healthy crop growth, were not considered [35]. To overcome these limitations, integrating artificial intelligence (AI) and advanced sensor technologies could enable real-time monitoring of these key factors and dynamic irrigation control based on plant developmental stages [36]. AI allows computers to replicate human-like understanding, creativity, problem-solving, autonomy, and decision-making. Moreover, machine learning (ML), a subset of AI, can further enhance system performance by processing system data over time [37].

Within the last several years, AI and IoT technologies have been widely adopted in agriculture to achieve precision through real-time environmental monitoring, automated irrigations plans, and predictive analysis to improve water management and crop production [38,39,40]. Numerous review papers have been published on aeroponics; however, each focuses on different aspects of the technology. For instance, previous reviews highlighted the challenges of monitoring parameters, including the nutrient solution’s T, RH, Li, EC, and pH. Several studies have also proposed the use of wireless sensor networks to monitor environmental parameters in aeroponics [41,42]. However, most of these reviews primarily examine the use of AI technologies to monitor environmental variables. Additionally, other reviews have explored real-time control of the T, RH, Li, EC, and pH of nutrient solutions by applying various ML algorithms, including artificial neural networks (ANNs) and genetic algorithms [43]. Although these papers highlight the potential of combining internet of things (IoT) and AI technologies, their focus remains largely limited to environmental parameters and does not specifically address plant growth monitoring. Moreover, many of these reviews overlooked the critical need for adapting irrigation controllers based on crop-specific water requirements [44,45]. The present study recognizes that most reviews focus predominantly on the incorporation of IoT technology into aeroponics, whereas comparatively few have comprehensively evaluated the potential of AI technology. Furthermore, previous reviews have inadequately addressed key plant growth parameters, including photosynthesis and transpiration. These studies have also predominantly overlooked the management of irrigation based on the aforementioned parameters, which are crucial for maximizing water efficiency. These gaps underscore the need for a focused investigation to bridge the knowledge gap between plant growth dynamics and irrigation methods in aeroponics. Therefore, this review aimed to fill this gap, with a particular emphasis on irrigation control through real-time monitoring of photosynthesis and transpiration using AI technologies. Moreover, this review explores the potential of integrating IoT technologies to monitor both environmental and plant growth parameters in aeroponics. It also discusses the capabilities of AI-based techniques for irrigation strategy formulation. Finally, this review identifies opportunities and future directions for optimizing plant growth and improving irrigation efficiency through novel approaches and technologies advancing solutions that address the current limitations of modern aeroponics. This review is mostly based on references published during the period 2012 to 2025, covering recent developments in IoT-based aeroponics and photosynthesis monitoring. Most publications come from reputable academic sources such as IEEE, Springer, and Frontiers, and the literature includes both globally distributed and region-specific studies, ensuring a balanced and comprehensive foundation (Figure 1).

2. IoT-Based Irrigation in Aeroponics

Aeroponics is a technological leap forward from traditional hydroponics. In aeroponics, roots are exposed to the air; if the misting cycles are interrupted, the roots will dry out. In contrast, in hydroponics, plants are grown without soil by suspending the roots directly in a nutrient-rich water solution and without the use of any solid substrates. Hydroponics can be affected by cultivars, circulation intervals, nutrient solution depth, fertigation levels, and nutrient concentration [46,47]. In hydroponics, the IoT typically controls the solution level, EC, and pH using direct liquid measuring tools, whereas in aeroponics, the IoT systems have to coordinate the spraying interval time, nozzle pressure, EC, and pH using much faster feedback loops and much more precise control architectures. However, aeroponics is still awaiting significant advancements in automation technologies like the IoT, especially in controlling the influencing factors inside the growth chamber. These challenges can be solved through continuous monitoring of vital growth parameters, with momentary measurement of all these parameters. Intensive and extensive research has been carried out to explore the development of IoT technologies in aeroponics. IoT deployment strategies are still valid for aeroponic systems, as both systems share common characteristics such as performing irrigation by controlling pump operations and managing the crop environment [48,49,50].

The IoT is a contemporary technology that facilitates smart farming by enabling devices to communicate and operate remotely. Its integration into agriculture can resolve several challenges in conventional agriculture, including land suitability, drought monitoring, irrigation control, pest control, and yield enhancement [51,52,53]. In recent years, the global adoption of IoT technologies in agriculture has expanded significantly. According to a business research company, the market size of the IoT in agriculture is projected to grow to USD 22.65 billion by 2028, with a compound annual growth rate of 10.5% [54]. Additionally, IoT-based farming has reduced water usage by 33% to 40% and increased yield by 10% to 12% compared with conventional farming [55]. Agricultural practices have also transitioned from being input-intensive to knowledge-intensive, focusing more on data-driven decision-making and networking through IoT integration [56,57]. Integrating IoT technology into aeroponics can significantly enhance resource efficiency through real-time data collection, analysis, and automated decision-making [58]. Moreover, the integration of the IoT into aeroponic value chains offers both large- and small-scale farmers access to expanded services, improved productivity, and reduced operational costs compared with conventional aeroponics (Figure 2).

IoT-based aeroponics can execute comprehensive perception, reliable data transmission, and intelligent processing of key environmental parameters such as T, RH, and L_i. Additionally, IoT-based aeroponics monitors crop health for early diagnosis and regulates irrigation accordingly [59,60]. The primary components of these systems include sensors, microcontrollers, actuators, and centralized computers. These components intelligently automate, store, and visualize data through IoT-enabled dashboards [61]. Sensors and actuators are typically connected to microcontrollers such as Arduino and Raspberry Pi, which control the environmental conditions in the growth chamber and facilitate real-time measurements [62]. Furthermore, an IoT gateway serves as a critical bridge, enabling communication between the cloud-based service platform and the interconnected microcontrollers [63]. All sensors are essential to collect data on key environmental parameters to optimize irrigation strategies [64]. When the sensors capture context data, it is transmitted to the microcontroller, which integrates and processes the data from all connected devices within a single platform [65]. Actuators, including nutrient tanks, water pumps, lighting, and air conditioners (AC), are then triggered to execute user-defined commands that regulate both irrigation and the internal environment of the growth chamber [66]. Several studies have investigated the use of IoT technology in monitoring environmental parameters and automating irrigation control in aeroponics.

For instance, a previous study reported the IoT-based design and development of automated aeroponics for controlling irrigation and growth chamber environments using various sensors. The study methodology involved integrating sensors, such as the DHT22, BH1750FVI, HC-SRO4, and Thermo Fisher Scientific Orion meter to monitor the T, RH, L_i, N_sl, EC, and pH of the nutrient solution. This system used the Arduino Uno microcontroller to manage the operations. When any parameter, such as EC, fell outside the predefined range, the system triggered an alarm or took corrective actions. For irrigation control, the pump was programmed to run every six minutes and to shut off every four minutes [67]. Another study used the Raspberry Pi 3B microcontroller to integrate sensors such as the DHT22 and DS18B20 to monitor the T and RH in aeroponics. In this setup, the pump was activated for 15 s after every 10 min interval, ensuring a consistent and automated irrigation cycle [68]. These studies demonstrated the potential of automated irrigation to enhance resource efficiency, increase revenue, and reduce environmental impact. However, a major limitation remains: timer-based irrigation control does not adapt to real-time variations in plant growth or environmental conditions.

To address this challenge, several studies have developed automated aeroponics that regulate irrigation based on T, RH, EC, and pH levels. For example, one study implemented automated aeroponics using an Arduino Uno microcontroller with DHT 11 and DHT 22 sensors to monitor T and RH. In this study, the irrigation pump was activated when RH dropped below 60% or when T exceeded 24 °C. Conversely, the pump deactivated when RH reached 85% or when T dropped below 15 °C [69]. In another study, IoT-based aeroponics dynamically measured the pH and EC of the nutrient reservoir to regulate irrigation. When parameters deviated from predefined ranges such as pH (<5.5 or >6.5) or EC (<1.0 mS/cm for strawberries), the system triggered misting of the roots. Irrigation duration and frequency were determined using an adaptive fuzzy logic algorithm that adjusted misting based on the plants’ needs [70]. However, these studies overlooked key growth factors such as CO₂ concentration and photosynthesis. CO₂ enrichment enhances photosynthetic efficiency, optimizes water usage by reducing transpiration losses, and significantly boosts crop potential.

However, when evaluating innovative technologies for irrigation monitoring and sensing, recent studies highlight state-of-the-art advancements in systems based on IoT concepts (

Table 1. IoT frameworks for smart farming cultivation.

S. No	Cultivation Method	Collected Variables	Irrigation Strategy	Communication Protocol	Reference
1	Hydroponics	T, RH, EC, pH, Tds	_	LoRa and MQTT	[76]
2		T, RH, CO2, WL	Sensor-based	Zigbee	[77]
3		T, RH, EC, pH, Li, Tds	_	LoRaWAN and WiFi	[78]
4		Li and WL	Sensor-based	Wi-Fi (ESP8266, HTTP or MQTT to Firebase)	[79]
5		T, RH, Li	Evapotranspiration-based	WSN	[80]
6	Aeroponics	T, RH, Li, pH	Time-based	Wi-Fi	[81]
7		T, RH, Li, pH, WL	6 min ON, 4 min OFF (Cycle)		[67]
8		T, RH, EC, pH	EC and pH-based		[70]
9		T and RH	T, RH-based		[69]
10		T, Nsl, Fr, pH, and EC	15 s ON, 400 s OFF (Cycle)		[34]
11		T, RH, and Li,	Every 5 min on/off		[32]
12		T, RH, and Li,	Every 5 min on/off		[82]
13		T, RH, and Tds	T- and RH-based		[83]
14		T, RH, pH, and Nsl	20 s ON 160 s OFF (Cycle)		[84]
15		T, RH, Tds, and pH, pH probe, and TDS	T- and RH-based		[85]
16		T and RH	15 s ON, 10 min OFF (Cycle)	UART + Wi-Fi	[68]
17		T, RH, EC, pH	T, RH, and pH-based	GPRS	[86]
18		T, RH, Li, and Nsl	Time-based	Wi-Fi + Bluetooth	[87]
19		T, RH, Li, pH	8 h turn on the pump daily	_	[88]
20		T, RH, and Li	RH-based	_	[89]

) [71]. These systems range from basic automatic irrigation setups to more complex solutions involving remote monitoring and control integrating sustainability and technology for improved water management. Moreover, the application of IoT technology has positively influenced aeroponic cultivation practices [72]. Despite these developments, the adoption of IoT technology in aeroponics faces several challenges. Key issues include data offloading, system heterogeneity, and the large volume of sensor-generated data. These systems continuously collect data from climate and nutrient sensors, often using the HTTP protocol to compile and transmit data packages to a server. However, a 30 s interval between requests prolongs the latency [60]. Addressing these challenges by integrating lightweight communication protocols, including MQTT and CoAP, could replace HTTP for prompt communication, ultimately improving data transmission speed and efficiency. These findings are existing studies in precision agriculture, which show that MQTT and CoAP work better than HTTP in terms of overall latency, energy consumption, and low-power IoT devices used in real-time monitoring [73,74]. Additionally, implementing edge computing can enable local data processing and filtering, thereby reducing server load and latency and improving overall performance [75]. Following this, we highlight the importance of crops grown in aeroponics being more sensitive to irrigation control, and due to the more complex hardware configurations, improvements in precise monitoring and communication protocol stability are required.

Along with the selection of communication protocols, the IoT deployment architecture is vitally important in terms of monitoring system effectiveness. Various networking topologies are generally used in IoT deployments, such as machine-to-machine (M2M) implementations allowing direct machine communication between devices, or star topologies where every device communicates with a central device or network hub or gateway, such as in LoRaWAN IoT and most WiFi networks [90,91,92]. Mesh networking is another widely used topology which allows device-to-device relaying of data over several hops, increasing coverage and reliability. This strategy is common in Zigbee-based IoT networks [93]. The IoT systems depend on multiple wireless technologies like Zigbee, LoRa, NB-IoT, SigFox, and WiFi with their complex trade-offs of range, power consumption, and bandwidth. Zigbee and WiFi have traditionally been used in short-range, high-throughput applications, but they may suffer from interference and limited scalability, whereas LoRa and SigFox offer long-range and low-power consumption to support massive networks of sensors, and NB-IoT supports wide-area and cellular-based connectivity, as well as strong mobility and scalability features. But they have strict payload limits and can face latency issues, making them less ideal for time-sensitive applications [94,95]. WiFi, while offering high bandwidth and ease of setup, consumes more power and is less scalable for large sensor networks [96]. It is important to distinguish WSNs from the IoT as a whole: WSNs are usually represented by localized sensing and data gathering at a small scale, but in the IoT, sensor networks work closely with cloud infrastructure, data analytics, and application services capable of providing end-to-end connectivity, intelligence, and automation in various fields [97]. Therefore, a thorough understanding of both deployment architectures and communication protocols is essential for designing effective and scalable IoT systems.

3. Technological Approaches to Monitoring Photosynthesis Rates

Photosynthesis is a pivotal process in crop growth and yield, serving as one of the primary mechanisms through which crops derive the carbon substrates necessary for development and maintenance [98,99,100]. This process occurs in plants, algae, and certain bacteria that utilize CO₂ and water, converting light energy into chemical energy to obtain sugar as the final product [101,102]. Additionally, oxygen is produced as a byproduct of oxygenic photosynthesis and is released into the atmosphere, playing a vital role in biological respiration. While plant growth depends heavily on photosynthesis, it is simplistic to think that the growth rate directly reflects the photosynthesis rate [103]. Optimizing photosynthesis requires the efficient acquisition of water and nutrients [104]. Furthermore, plants respond to varying environmental factors, such as T, L_i, and CO₂ concentration, making the monitoring of such factors essential [105]. To enhance the accuracy and efficiency of photosynthesis rate estimation, various techniques have been developed. These include predictive models based on empirical data, gas exchange measurements, and machine learning algorithms, each leveraging different methodologies and data inputs.

Empirical techniques based on observational data and mathematical modeling offer predictive tools for estimating photosynthesis rates [106,107]. For instance, one study predicted photosynthesis rates by assaying plant responses to varying parameters, such as L_i and CO₂ concentrations, using a rectangular hyperbolic model. Data analysis was performed using SPSS 20 and OriginPro 8 software. The simulation results demonstrated a strong correlation with the actual measurements (R² = 0.99), indicating high model accuracy. Similarly, another study used an open-type flux chamber technique to estimate the photosynthesis rate at the canopy level [108]. Parameters such as T, CO₂, photosynthetic photon flux density (PPFD), and Q were measured. The canopy photosynthesis model was applied using ilastick software to predict the photosynthesis rate. While this model was more adaptable to dynamic environmental conditions, its predictive accuracy remained moderate (R² = 0.81), primarily due to the complexity of measuring photosynthetic activity at the canopy scale [109]. However, these studies face limitations in generalizability and real-time accuracy because of the instinctive changes in environmental conditions. Additionally, these models often struggle with the complexity of accurately capturing dynamic changes in environmental factors, making it difficult to estimate real-time photosynthetic rates [107,110]. These limitations can be addressed by integrating gas exchange techniques, which provide more precise, leaf- and canopy-scale measurements of photosynthetic parameters. Such integration improves model accuracy and facilitates its application across different environmental conditions [111,112].

For example, a previous study estimated the photosynthesis rate by measuring g_s, PPFD, and CO₂ concentration under controlled experimental conditions. The C₃ photosynthesis model was used to analyze these parameters, and R software was used for statistical analysis. Although the study reported a coefficient of determination R²= 0.79, indicating a moderate correlation between predicted and observed photosynthesis rates, it demonstrated how these factors influence the photosynthesis rate of strawberry plants, thereby assisting more effective control [113]. Similarly, another study adopted a cost-effective and accessible approach to predict the photosynthesis rate. The researchers measured CO₂ assimilation under controlled conditions using an infrared gas analyzer and aimed to simplify the application of the Farquhar, von Caemmerer, and Berry (FvCB) model for predicting the photosynthesis rate by estimating parameters such as rubisco carboxylation capacity (A_c), electron transport (A_r), and maximum carboxylation capacity (V_cmax). Raw data were initially organized in Excel, while most calculations and topographical statistical analyses were performed in R software. The FvCB model demonstrated a robust correlation (R² = 0.85) between the predicted and observed photosynthesis rates [114]. While these studies provide accurate, real-time predictions and offer valuable insights into plant physiological responses, they are sensitive to environmental fluctuations, complex in data processing, and often resource intensive. Machine learning techniques can address these limitations by analyzing multidimensional data, identifying nonlinear patterns, and enhancing prediction accuracy under dynamic environmental conditions [115,116]. In aeroponics, IoT-enabled sensors allow real-time monitoring of photosynthesis-related parameters such as L_i, CO₂ levels, and leaf temperature. These data support automated environmental adjustments to optimize plant physiological performance.

Several studies have developed machine learning algorithms to predict photosynthesis rates [117,118,119]. For instance, one study developed two machine learning algorithms—support vector regression (SVR) and a back propagation neural network (BPNN)—to predict the photosynthesis rate in cucumbers using key parameters such as T, CO₂, and L_i. MATLAB R2016a was used for data processing and model implementation. Both models demonstrated exceptional predictive accuracy, with R² = 0.998 for SVR and R² = 0.996 for BPNN. The findings indicate that SVR effectively processes small datasets, whereas BPNN can model nonlinear relationships. However, the study highlighted challenges such as the computer-intensive BPNN training and the need for hyperparameter optimization to maintain accuracy under dynamic conditions [120]. In another study, an ANN model was developed to predict the photosynthesis rate in spinach based on T, RH, CO₂ concentration, and PPFD. Python 3.7 was used to implement the ANN and establish relationships between the input parameters and photosynthetic performance. The model achieved a predictive accuracy of R² > 90, indicating strong capability in predicting photosynthesis rates [121]. Numerical models refer to equation-based or regression-based approaches derived from physiological or empirical relationships, such as the 3D plant and FvCB models. In contrast, ML and AI approaches encompass conventional algorithms like SVR and RF, as well as advanced AI techniques including deep learning models and neural networks (Table 2).

While ML models generally achieve high accuracy, they can perform poorly when applied to complex canopies under fixed growing conditions. These models often fail to consider soil conditions and require more training data and measurement points for broader applicability [122,123]. Furthermore, insufficient data collection and processing hinder the integration of various datasets, thereby reducing model efficiency. Despite these challenges, ML models remain promising for aeroponic crop production due to their generalizability across varying environmental parameters [124].

Table 2. Previous prediction techniques for photosynthetic rate.

S. No	Pn Prediction Techs	Model	Parameters	Analytical Software	Accuracy (R2)	Crop	Reference
1	Numerical models	Rectangular Hyperbolic	PPFD, CO2, Fm, and Fv	SPSS 20, OriginPro 8	R2 = 0.999	Lettuce	[108]
2		Canopy Pn	T, CO2, PPFD, and Q	Ilastik 1.3.2	R2 = 0.81	Spinach	[109]
3		3D Plant	T, CO2, and Li	3D CAD	R2 = 0.79	Mango	[125]
4		FvCB, FvCBePAR, and FvCBWd	Vcmax, Jmax, CO2, and Q	Python	R2 = 0.886 R2 = 0.928 R2 = 0.924	Bittersweet lettuce and common bean	[126]
5		FvCB	Ac, Ar, and Vcmax	R 4.0.3	R2 = 0.85	Rice and wheat	[114]
6		FvCB	Vcmax, Jmax, CO2, and T	Ms Excel, R	N/A	Soybean sunflower	[127]
7		C3 Pn	gs, CO2, and PPFD	R	R2 = 0.79	Strawberry	[113]
8		Non-Rectangular Hyperbolic	gs, E, PPFD, and WUE	GraphPad 5.0 and Sigma plot 14.0	N/A	Maize	[128]
9	ML and AI models	Sugeno	T, RH, and SM	MATLAB	R2 = 0.95	Jalapeno pepper	[129]
10		SVR and BPNN	T, CO2, Li, and T	MATLAB R2016a	R2 = 0.998 R2 = 0.996	Cucumber	[120]
11		SVR and RF	T, CO2, Li, LQ	N/A	R2 = 0.990 R2 = 0.998		[130]
12		PB and PSO-BP	T, CO2, Li, ETR, NPQ, qP, PhiPS2, and Fv/Fm	MATLAB 2015b	R2 = 0.96 R2 = 0.98		[131]
13		ANN	T, RH, CO2, and PPFD	Python 3.7	R2 > 90	Spinach	[121]
14		WDNN	T, RH, CO2, and PAR	TensorFlow 2.4	R2 = 0.97	Tomato	[132]
15		BP, SVM, and PSO-LSSVM	T, CO2, and PPFD	N/A	MRE = 0.04 MRE = 0.03 MRE = 0.02		[133]
16		SVR	T, RH, CO2, Li, and La		R = 0.94		[134]
17		BPNN	T, RH, CO2, SM, and Chl	MATLAB	R = 0.99		[135]
18		PSO-SVM	T, RH, CO2, and Li		R2 = 0.96		[136]
19		SVR and MLSTM	T, RH, and CWSI	N/A	R2 = 0.81 R2 = 0.81	Chinese Brassica	[137]

Table 2. Previous prediction techniques for photosynthetic rate.

S. No	Pn Prediction Techs	Model	Parameters	Analytical Software	Accuracy (R2)	Crop	Reference
1	Numerical models	Rectangular Hyperbolic	PPFD, CO2, Fm, and Fv	SPSS 20, OriginPro 8	R2 = 0.999	Lettuce	[108]
2		Canopy Pn	T, CO2, PPFD, and Q	Ilastik 1.3.2	R2 = 0.81	Spinach	[109]
3		3D Plant	T, CO2, and Li	3D CAD	R2 = 0.79	Mango	[125]
4		FvCB, FvCBePAR, and FvCBWd	Vcmax, Jmax, CO2, and Q	Python	R2 = 0.886 R2 = 0.928 R2 = 0.924	Bittersweet lettuce and common bean	[126]
5		FvCB	Ac, Ar, and Vcmax	R 4.0.3	R2 = 0.85	Rice and wheat	[114]
6		FvCB	Vcmax, Jmax, CO2, and T	Ms Excel, R	N/A	Soybean sunflower	[127]
7		C3 Pn	gs, CO2, and PPFD	R	R2 = 0.79	Strawberry	[113]
8		Non-Rectangular Hyperbolic	gs, E, PPFD, and WUE	GraphPad 5.0 and Sigma plot 14.0	N/A	Maize	[128]
9	ML and AI models	Sugeno	T, RH, and SM	MATLAB	R2 = 0.95	Jalapeno pepper	[129]
10		SVR and BPNN	T, CO2, Li, and T	MATLAB R2016a	R2 = 0.998 R2 = 0.996	Cucumber	[120]
11		SVR and RF	T, CO2, Li, LQ	N/A	R2 = 0.990 R2 = 0.998		[130]
12		PB and PSO-BP	T, CO2, Li, ETR, NPQ, qP, PhiPS2, and Fv/Fm	MATLAB 2015b	R2 = 0.96 R2 = 0.98		[131]
13		ANN	T, RH, CO2, and PPFD	Python 3.7	R2 > 90	Spinach	[121]
14		WDNN	T, RH, CO2, and PAR	TensorFlow 2.4	R2 = 0.97	Tomato	[132]
15		BP, SVM, and PSO-LSSVM	T, CO2, and PPFD	N/A	MRE = 0.04 MRE = 0.03 MRE = 0.02		[133]
16		SVR	T, RH, CO2, Li, and La		R = 0.94		[134]
17		BPNN	T, RH, CO2, SM, and Chl	MATLAB	R = 0.99		[135]
18		PSO-SVM	T, RH, CO2, and Li		R2 = 0.96		[136]
19		SVR and MLSTM	T, RH, and CWSI	N/A	R2 = 0.81 R2 = 0.81	Chinese Brassica	[137]

4. Proposed IoT- and ML-Based Aeroponics

To effectively monitor plant photosynthesis and transpiration rates and to ensure optimal spraying intervals in aeroponics, various key components must be integrated. In the proposed model, the system is divided into three major sections: the aeroponic chamber, the actuators, and the control and monitoring system. The aeroponic chamber is the core section where crops are cultivated under precisely regulated conditions [138,139]. The actuators comprise the AC and the nutrient supply mechanism. In this section, LED lighting provides artificial illumination, whereas AC and fans regulate air circulation, T, and RH. Pumps and nozzles are used to deliver the nutrient solution directly to plant roots [140]. Thus, the control and monitoring section serves as the aeroponics backbone. It maintains the chamber condition and manages the actuators according to the given conditions (Figure 3). Furthermore, microcontrollers collect data from sensors monitoring T, RH, L_i, and CO₂ concentrations and then analyze the data using ML algorithms to predict crop growth parameters such as photosynthesis and transpiration rates [141].

Accurate implementation of photosynthetic models and efficient spraying nozzles are essential for achieving precise droplet sizes and optimal irrigation in aeroponics [24]. ML models support the process by predicting the photosynthesis rate based on parameters such as T, RH, L_i, and CO₂. Furthermore, misting mechanisms should be designed to provide a uniform distribution of nutrient solutions, as performance issues such as clogged nozzles result in uneven nutrient uptake, negatively affecting the photosynthesis rate [22]. Therefore, the spraying nozzles should be integrated into a control and monitoring framework that uses sensor data to regulate spraying intervals based on the real-time photosynthesis rate. Several studies have applied ML to analyze T, RH, L_i, and CO₂ to assess plant health and support irrigation decision-making [142,143,144].

For instance, one study conducted in a controlled greenhouse environment developed an irrigation decision model optimized for photosynthesis rate. A multi-gradient nested experimental design was used to examine the effects of T, RH, L_i, CO₂, and soil water content on the photosynthetic performance of tomato seedlings. The resulting model based on SVR, GA, and BPNN dynamically determined optimal soil moisture levels under varying environmental conditions to maximize photosynthesis rates, achieving a high precision of R² = 0.9738 [145]. Another study developed the XGBR-ET model to predict evapotranspiration and improve irrigation management in controlled environments. The model achieved the highest prediction accuracy among other regression models, with the lowest RMSE of 0.032 and R² = 0.981. Additionally, the analysis identified five key meteorological factors—namely, net solar radiation, relative humidity, minimum relative humidity, maximum temperature, and minimum temperature—as sufficient to ensure high prediction accuracy [146].

ML models such as ensemble regression and neural networks are especially advantageous to aeroponics since they are adaptive in predicting photosynthesis rates as well as transpiration rates. In a previous study, XGBoost regression predicted a high yield accuracy (R² = 89.48) of lettuce under controlled aeroponic conditions. Based on these findings, ML algorithms will be implemented in the proposed aeroponic system to collect data from various sensors and predict irrigation intervals based on photosynthesis rates (Figure 4). Sensors for monitoring T, RH, L_i, and CO₂ concentration will be installed within the aeroponic chamber to collect real-time data. Once collected, the data will be transmitted to the prediction model, which will analyze the inputs and predict the photosynthesis rate. Subsequently, the decision-making module determines the irrigation schedule, activating the pump if the predicted photosynthesis rate is low and keeping it off if the rate is optimal. This approach maximizes plant hydration, minimizes water waste, and supplies water only when needed. Additionally, regulating the irrigation schedule based on the photosynthesis rate enhances nutrient uptake and promotes plant growth. Therefore, ML-driven aeroponics offers an ideal solution for precision agriculture, particularly in drought-prone regions, as it optimizes crop productivity while conserving water resources.

5. Challenges and Solutions in Adopting IoT and ML in Aeroponics

IoT and ML technologies are quickly transforming the future of aeroponic farming, with the potential to facilitate intelligent automation, real-time decision-making, and precision of all resources in aeroponics. Unlike conventional aeroponics relying on manual control, the concepts of the IoT and ML bring a more actuated and data-driven interface to the management of environmental and nutritional parameters. The main strength of integrating the IoT and ML into aeroponics to analyze complicated data of parameters such as T, L_i, EC, pH, and P_n is that it ultimately enhances resource efficiency and ensures more consistent crop performance. This section examines these core challenges and explores practical solutions that can support the broader and more inclusive adoption of the IoT and ML in aeroponics, with a focus on adaptability, scalability, and long-term sustainability. In addition to this feasibility, there are a number of challenges to the adoption of advanced IoT and ML in aeroponics, some of which include high start-up costs, requirements of stable power and connectivity, and the quality of data used, which are particularly acute in low-resource settings or in physically remote locations. IoT- and ML-based aeroponics thus relies on the availability of the technology in terms of feasibility and scale, which is highly dependent upon regional conditions and requires customized forms of solutions based on local limitations related to cost, infrastructure, and expertise [147,148,149].

5.1. Technological and Operational Challenges

High initial cost: The initial setup of aeroponics is expensive due to the need for specific equipment, sensors, and infrastructure to maintain a controlled environment. Additionally, training ML models for various greenhouse and geographical conditions increases implementation costs [24].

High energy consumption: Artificial lighting is essential for photosynthesis in aeroponics. While the adoption of LED technologies has improved energy efficiency, the overall power demand of IoT- and ML-integrated aeroponic systems remains high due to the continuous operation of sensors, actuators, climate control units, and computing resources. In many regions, particularly in rural or developing areas, power networks are often unreliable, with frequent outages and voltage fluctuations. These interruptions can lead to failures in irrigation scheduling, loss of sensor data, or malfunction of control systems. Moreover, high electricity costs can reduce the economic feasibility of aeroponics, especially for small-scale or resource-limited operators [140].

Connectivity issues: Poor coverage of 2G, 3G, and 4G networks poses a major obstacle to IoT functionality. Although low-power wide-area network technologies such as LoRa and Sigfox offer partial solutions, they are limited in handling large volumes of data [150].

Security and safety issues: Security remains a major barrier in IoT- and ML-enabled aeroponics. Even minor flaws can lead to data loss and failure in capturing environmental parameters [151].

High maintenance: Due to the complexity and continuous operation of aeroponics, components tend to degrade over time, necessitating frequent maintenance, repair, or replacement [152].

Data accuracy and reliability: IoT sensors are susceptible to malfunction, which may result in inaccurate data collection. Consequently, ML models trained on erroneous data can generate incorrect predictions and flawed judgments [153].

Hardware limitations: Aeroponics requires advanced hardware integration with IoT and ML technologies to handle the variability in plant growth. Specifically, as plant roots increase and are exposed to the nutrient spray, nutrient absorption may fluctuate.

5.2. Current Solutions and Future Directions

Integrating targeted solutions is essential for overcoming these challenges in IoT- and ML-based aeroponics. Several current solutions and future directions are outlined below to enhance aeroponics’ reliability, efficiency, and scalability.

Energy optimization and grid independence: In order to reduce the problem of high energy consumption and grid dependency, several energy-based ideas have been suggested. The incorporation of renewable sources of energy, including solar panels or small wind turbines, may help to avoid reliance on the conventional grid and enhance system resilience. When integrated with battery storage, these energy sources can ensure a continuous power supply during periods of grid interruption or night-time. Furthermore, cost-effective equipment includes energy-efficient hardware like low-power microcontrollers and sensors, which further reduce electricity consumption without compromising performance. Artificial intelligence can be used to build smart energy management systems that dynamically optimize energy consumption (through lights and climate control) by using real-time environmental and crop data. These approaches collectively enhance the sustainability, scalability, and affordability of IoT- and ML-based aeroponic farming [154].

Hybrid network integration: To address connectivity issues, hybrid networks such as 5G, LoRaWAN, and NB IoT can be integrated with edge computing to enable efficient data processing, even in areas with unstable network coverage [155].

Blockchain-based security: Blockchain technology may strengthen system security and safety by enabling secure data logging, encrypted connection between transmitters and receivers, and multi-layered protection of sensitive environmental and operational data [156].

Data reliability: Combining multiple sensor arrays with real-time anomaly recognition algorithms and federated learning for decentralized model training can significantly improve data accuracy by minimizing errors caused by malfunctioning sensors [157].

Hardware optimization: To enable real-time data processing and hardware adjustments, future research should focus on developing adjustable spraying nozzles and flexible sensors.

The integration of the IoT and ML in vertical farming potentially opens new opportunities to expand aeroponic technology, especially in cities and towns where space is limited. By enabling precise, multilayered control of environmental conditions, IoT-based vertical farms can optimize resource use and enhance productivity per unit area. ML algorithms further improve efficiency through predictive modeling and adaptive irrigation or lighting schedules. By enabling precise, multilayered control of environmental conditions, IoT-based vertical farms can optimize resource use and enhance productivity per unit area. ML algorithms further improve efficiency through predictive modeling and adaptive irrigation or lighting schedules.

6. Conclusions

Aeroponics delivers nutrients to plants via mist, replacing soil and differing significantly from conventional farming practices. This requires specialized plant growth models to support the system’s unique physiological conditions. This review explored the integration of IoT technologies and ML algorithms to enable the real-time monitoring of key parameters such as T, RH, L_i, CO₂ concentration, and photosynthesis rate for dynamic irrigation control. This integration enables irrigation scheduling based on actual plant physiological needs rather than the conventional time-based schedule (fixed time intervals). Furthermore, ML models have demonstrated the ability to accurately predict photosynthesis rates, thereby enhancing water-use efficiency and crop productivity. However, despite its potential, aeroponics faces several challenges including optimal spraying intervals, sensor calibration in high-humidity environments, data analysis, and transmission reliability. Addressing these issues is fundamental to developing technological interoperability between the IoT and ML, which is essential for optimized irrigation management in aeroponics. AI and IoT applications in aeroponics can be visualized as a multi-stage growth process starting with sensor-based real-time monitoring, proceeding to the incorporation of machine-learning-based predictive and autonomous control, and ultimately leading interoperable, standardized, and scalable intelligent cultivation platforms. This pathway highlights the transition from basic data collection to sophisticated, self-optimizing systems capable of adapting to dynamic environmental and crop needs. By evaluating current technological strengths, limitations, and future opportunities, this review provides valuable insights for researchers, policymakers, and technology developers aiming to implement intelligent aeroponics.