IOT-ENABLED ENVIRONMENTAL MONITORING AND POLLUTION CONTROL SYSTEM

Abstract

Air quality, water pollution, and radiation pollution present significant challenges to the environment, necessitating effective monitoring for the achievement of sustainable growth and the maintenance of a healthy society. In recent times, environmental monitoring has evolved into a Smart Environment Monitoring (SEM) system, driven by advancements in the Internet of Things (IoT) and modern sensor development. This manuscript seeks to conduct a thorough review of notable contributions and research studies on SEM, focusing on the monitoring of air quality, water quality, radiation pollution, and agricultural systems. The review categorizes SEM methods based on their application purposes, and each purpose is further examined in terms of the sensors utilized, machine learning techniques applied, and classification methods employed. Following the extensive review, a detailed analysis is presented, outlining major recommendations and impacts of SEM research based on discussion results and research trends. The authors critically assess how advancements in sensor technology, IoT, and machine learning methods contribute to transforming environmental monitoring into a genuinely intelligent monitoring system. Lastly, the paper proposes a framework for robust machine learning methods, denoising techniques, and the establishment of suitable standards for Wireless Sensor Networks (WSNs).

Specification

Description:FIELD OF INVENTION
Environmental monitoring plays a crucial role in safeguarding both the public and the environment against hazardous contaminants and pathogens. Its utility extends to preparing environmental impact assessments and identifying situations where human activities may pose risks of adverse effects on the natural environment.
BACKGROUND OF INVENTION
The sustainable development of the global community hinges on various factors, including the economy, quality education, agriculture, industries, and notably, the environment. Health and hygiene are integral components of human sustainability and a nation's progress, both of which are deeply rooted in a clean, pollution-free, and hazard-free environment. Therefore, monitoring the environment is imperative to ensure that the citizens of any nation can lead healthy lives. Environmental monitoring (EM) involves meticulous planning and disaster management, controlling various pollutants, and effectively addressing challenges arising from unfavorable external conditions. EM encompasses the management of water pollution, air pollution, hazardous radiation, weather changes, earthquake events, and more. Pollution sources arise from a combination of human-made and natural factors, and the role of EM is precisely to tackle these challenges, thereby safeguarding the environment for a healthy society and world. With recent advances in science and technology, particularly in artificial intelligence (AI) and machine learning, EM has evolved into a smart environment monitoring (SEM) system. This technological progress enables SEM methods to monitor environmental factors more precisely and exercise optimal control over pollution and other undesirable effects.

The patent application number 201947026165 discloses a method for comprehensive evaluation of indoor environmental quality.
The patent application number 201911005534 discloses an apparatus for determining time dependent deformation as well as fracture propagation of different rock specimens at variable temperature and environmental conditions.
The patent application number 201947036922 discloses an environmental condition verification and user authentication in a security coprocessor.
The patent application number 201917041178 discloses a peptide having cytoprotective effect against environmental pollutant and use thereof.
The patent application number 202047057417 discloses a polyethylene with polyethylene glycol with increased environmental stress crack resistance.
SUMMARY
Traditional methods of urban planning are being replaced by the design of smart cities that employ wireless networks for monitoring vehicular pollution levels. Wireless sensor networks (WSNs) with modern sensors operating on AI-based monitoring and controlling methods play a pivotal role in this transformation. The integration of Internet of Things (IoT) devices within WSNs facilitates effective waste management, vehicle tracking, temperature control, and pollution control. Consequently, contemporary methods of environmental monitoring are termed SEM systems, owing to the utilization of IoT, AI, and wireless sensors. The literature reports various SEM methods, including the assessment of burned areas using multispectral data captured through satellite imaging and remote sensing, mobile health monitoring systems, and IoT-based environmental systems, as well as smart marine environment systems employing multimodal sensing networks. In the context of WSNs, the establishment of standards and protocols is crucial for the effective implementation of SEM systems. Consequently, studies are underway to develop protocols and standards for IoT-based SEM systems.
DETAILED DESCRIPTION OF INVENTION
The global effort towards safeguarding the environment for sustainable agriculture, overall growth, and the well-being of society underscores the primary objective of Smart Environment Monitoring (SEM). SEM is designed to tackle challenges arising from adverse environmental effects through intelligent monitoring, ensuring the regulation of key growth indicators, particularly the health of society. Environmental monitoring methods are applied for diverse purposes, encompassing weather forecasting, air pollution control, water quality monitoring and control, and assessment of crop damage, among others. The overarching goal is to create favorable environmental conditions for agriculture, human beings, and all inhabitants on Earth.
Technological advancements, notably in IoT and wireless networks, have streamlined environmental monitoring, making it AI-controlled and efficient. In the literature, SEM systems utilize various types of smart sensors, wireless sensor networks (WSNs), and IoT devices. These devices communicate through networks, enabling the development of a smart monitoring system capable of addressing challenges in diverse conditions.
IoT, WSNs, and appropriate sensors form the backbone of SEM systems. WSNs establish connectivity for data captured by sensors and IoT devices, which record, monitor, and control various environmental conditions, such as water quality, temperature, and air quality. To illustrate, a cloud-based SEM system serves as a comprehensive example, as depicted in Figure 1. This example illustrates the monitoring and control of water contamination using a cloud-based system that connects IoT devices and various suitable sensors. The system, equipped with IoT devices embedded with AI and machine learning capabilities, can assess whether water is contaminated or clean. Organizations responsible for monitoring water quality from various sources access the cloud to analyze data collected from sensors, such as aqua sensors, employing IoT-based analyses to conduct quality checks.

Figure 1: Intelligent Water Quality Monitoring System (IWQMS) featuring Cloud-based Connectivity through Internet of Things (IoTs) and Sensors.
Another illustration of a SEM system with a broader scope is depicted in Figure 2. This system addresses diverse environmental monitoring issues, including humidity, temperature, radiation, dust, UV signals, and more. The WSN serves as the foundational element, establishing the interface between IoT devices and data collected from various smart sensors. This represents an exemplary instance of a "smart city", utilizing a SEM system to guarantee a healthy environment for its residents.

Figure 2: Environment Dynamics Monitoring System (EDMS) utilizing Wireless Sensor Networks (WSNs) and IoT Devices.
Turning our attention to agriculture, a pivotal aspect for the progress of any nation, it becomes evident how SEM can play a crucial role in fostering "smart" or "green agriculture". This approach addresses key challenges and factors associated with sustainable growth and increased productivity within the agricultural sector. Figure 3 illustrates one such smart agriculture scenario, showcasing a SEM system functioning as a smart agriculture monitoring system. In this instance, factors such as soil health, moisture analysis, water contamination levels, water quantity, and other pertinent variables are paramount for achieving sustainable productivity in agriculture. Figure 3 demonstrates that the smart agriculture monitoring system incorporates all these factors, managed and monitored through IoT devices and appropriate sensors capturing agricultural data, subsequently transmitted to the cloud via a WSN.

Figure 3: Intelligent Agricultural Surveillance System Employing IoT Devices and Sensors.
We endeavored to investigate existing contributions through a critical examination of SEM methods; however, it is noteworthy that comprehensive reviews discussing significant findings in SEM methods are scarce in the literature. Limited literature appears to delve into the review or survey of SEM techniques. Existing surveys focus on diverse areas such as smart agriculture systems, smart home technologies, smart health monitoring systems, environment monitoring, an IoT-based ecological system, IoT for marine environment monitoring, and a survey on pollution monitoring systems, each highlighting distinct aspects of SEM. While the environment faces contamination from various factors, water pollution, air pollution, radiation, and sound pollution emerge prominently in much of the existing research. This motivates our initiative to present an extensive review on SEM, encompassing all critical factors impacting environmental health and the predominant methods employed to address challenges stemming from these factors, including IoT and sensor technologies.
The current research indicates that environment monitoring systems are intelligently implemented as SEM for diverse purposes, utilizing various methods. The authors have extensively studied a multitude of contributions on SEM, categorizing them based on purposes and methods into three main subsections: smart agriculture monitoring systems (SAMs), smart water pollution monitoring systems (SWPMs), and smart air quality monitoring systems (SAQMs). This manuscript aims to critically report the major findings and limitations of current research on SEM, covering soil monitoring (SM), ocean environment monitoring (OEM), marine environment monitoring (MEM), air quality monitoring (AQM), water quality monitoring (WQM), and radiation monitoring (RM) [1,36], offering a comprehensive analysis of SEM's application fields.
While delving into the existing literature on SEM methods, particularly advancements in IoT and sensor technologies for SEM systems, it becomes evident that an extensive review on this topic is lacking. Although some interesting literature addresses specific challenges in environmental factors such as water pollution, air quality, radiation, and smart agriculture, a comprehensive overview is missing. The objective of this study is to highlight major advances in IoT and sensor technologies used to address SEM challenges, incorporating significant research studies and contributions from various sources that emphasize specific classic work on SEM methods. The examination of advances in IoT and sensor technologies used for SEM provides insights for scientists, policymakers, and researchers to develop a framework of appropriate methods for monitoring the environment, addressing challenges mainly related to poor air quality, water pollution, and radiation. These factors also impact agriculture, the backbone of any developed and developing economy, and thus, smart agriculture monitoring (SAM) is also studied in this section.
Table 1: Exploration of Research Studies on the Purpose and Applications of Environment Monitoring.

Table 1 presents major research studies and contributions on the aforementioned areas of interest: SM, OEM, MEM, AQM, WQM, and RM. Soil monitoring methods, for instance, have been influenced by greenhouse effects. Ocean and marine SEM systems, implemented using sensors, wireless sensor networks (WSN), and IoT, have faced challenges such as cost, coverage, and installation issues. Air pollution control and AQM have been proposed using a mobile sensor network, wireless sensors, and IoT devices operating on AI and machine learning. Table 1 illustrates that various types of SEM systems are designed and implemented for diverse purposes, highlighting the absence of a robust method that can comprehensively address environmental challenges.
Exploration of Smart Agriculture Monitoring (SAM) Research
Smart Agricultural Monitoring (SAM) systems encompassing measures for crop monitoring, pest control, and fertilizer management are highlighted in Table 2, summarizing key research studies. The gCrop system focused on plant growth monitoring, utilizing IoT, machine learning, and Wireless Sensor Networks (WSN). Although achieving a high prediction accuracy of 98%, the work faced challenges related to computational complexity due to a third-degree regression model. Another study assessed crop quality, employing Synthetic Aperture Radar (SAR) data for monitoring paddy rice quality. This assessment utilized Support Vector Machines (SVMs) with back-scattering features, showcasing effectiveness despite a limited sample size.
Leaf area and dimensions play a crucial role in evaluating various crop types, determining growth satisfaction. An example is found, which measured the leaf area index using SVM as the machine learning technique, incorporating a Gaussian process model. The accuracy of measurement was reported at 89%, albeit with a restricted sample size. Another noteworthy implementation is an expert system, leveraging AI and the Naive Bayes method for machine learning. This system operated on sensor data captured in agriculture, proving valuable in monitoring fertilizer and pesticide quality, as well as determining the appropriate irrigation amount for crops. Additional works focused on crop quality assessment, with specifically monitoring soil health for soybean crops based on phenological data and real-time images captured by unmanned aerial vehicles (UAVs).
Various SEM systems find application in diverse areas, such as smart farming, pest monitoring, and crop area monitoring, reflecting the breadth of research in this field.



Table 2: Exploration of IoT-Centric Smart Environment Monitoring (SEM) Systems Research.

Examining the Impact of IoT, Sensors, and AI Techniques on Smart Environment Monitoring (SEM) in Agriculture.
Investigation into Smart Water Pollution Monitoring (SWPM) Systems Research
A comprehensive examination of literature has been conducted on smart water pollution monitoring (SWPM) methods and systems, incorporating machine learning techniques, IoT, and wireless sensors. Table 3 outlines several noteworthy contributions in the realm of SWPM. One study employed remotely sensed images and applied machine learning for predicting pollution levels in lagoon water, specifically for agricultural purposes. Despite using ordinary neural network-based machine learning, the prediction results were deemed less than satisfactory. Another research effort [65] focused on classifying water contamination into clean or polluted categories, utilizing machine learning methods and IoT devices. While presenting a real-time contamination monitoring system, the study acknowledged limitations in data capture, confined to a specific area.


Table 3: Exploration of IoT-Enabled Smart Water Pollution Monitoring Systems (SWPM) Research.

Assessment of various pollutants in water was addressed in [66], classifying pollutants using a DSA-ELM model explored AI and neural network-based predictions of water quality parameters, concentrating on alkalinity, chloride, and sulfate content estimations. Emphasizing big data analysis and classification challenges, discussed SVM-based classification for water contamination. Quality assessment of drinking water and its real-time classification into drinkable and non-drinkable water were examined, employing an AI-SVM-based classification technique explored video-based surveillance for monitoring water quality and pollutants, using IoT tools for video surveillance and machine learning for classifying water as either polluted or clean. Additionally, proposed a drinking water prediction model, employing a feature-based model for the analysis of drinking water to predict its quality before usage. In another study, chlorophyll-A concentration in lake water was assessed using various machine learning models, recommending the approach for a real-time lake water cleaning management system.

Exploration of Smart Air Quality Monitoring (SAQM) Research
Investigations into methods and systems for Smart Air Quality Monitoring (SAQM) have been explored, with Table 4 offering a concise overview of recent literature on air quality monitoring systems. One study implemented air quality characterization using heterogeneous sensors and machine learning methods. While successful in monitoring and characterizing water quality, interoperability issues were noted due to the use of heterogeneous sensors. Air quality evaluation, employing both fixed and mobile sensor nodes, demonstrated the ability to assess air quality in stationary and mobile settings. In the latter case, compatible sensors were deployed as mobile nodes, functioning effectively in a dynamic environment. Data collected through smart sensor nodes underwent processing and analysis using machine learning techniques.
Another approach to air quality control involved IoT and machine learning techniques, focusing on the assessment of air pollution through gas sensors that capture air particles and analyze pollutants present in the air. Sensor networks were established in moving vehicles for air quality monitoring using machine learning, deploying mobile sensor nodes and Wireless Sensor Networks (WSN). Infrared sensors were utilized to evaluate air quality, specifically analyzing volatile organic compounds (VOCs), with the assistance of machine learning methods. VOC elements were detected and analyzed through spectroscopic observations. The presence of PM2.5, a component contributing to air quality assessment, was predicted in using extreme machine learning techniques applied to spatio-temporal data collected over a specified duration and distance range covered by the sensors.



Table 4: Exploration of Machine Learning and IoT-Enabled Smart Air Quality Monitoring (SAQM) Systems Research.

Different forecasting models for urban air quality evaluation, focusing on components like O3, SO2, and NO2, were proposed, with a comparison of the models used in the study. An air quality control mechanism incorporating RFID and a gas sensor was implemented in to predict pollution levels, utilizing IoT for analyzing sensory data captured through gas sensors. RFID primarily aided in pollutant detection and communication with WSNs through IoT devices in a WSN architecture. An SAQM system utilizing LoRaWAN (long-range WAN) was explored in, proving valuable for detecting temperature, dust, humidity, and carbon dioxide components in the air. In, an intelligent air quality system employed AI and machine learning techniques for developing expert systems to assess CO2, NOx, temperature, and humidity. Additionally, components such as PM10, PM2.5, SO2, oxides of nitrogen (NOx), O3, lead, CO, and benzene were detected in based on machine learning methods trained by spatio-temporal data. This approach was extended using deep learning for the detection and detailed analysis of O3 components exclusively. Another study utilizing heterogeneous sensors was explored in, where SVM was employed to analyze sensor data captured through heterogeneous sensors for air quality estimation.
The work delved into SEM systems, encompassing air quality assessment, water pollution monitoring, and agriculture monitoring, along with the sub-applications of these primary studies. While the primary focus remained on recent research contributions, a few significant studies conducted in the last two decades were also incorporated. The reported contributions spanned various SEM methods, addressing purposes such as air quality assessment, water pollution monitoring, radiation monitoring, and smart agriculture monitoring systems.
Key observations derived from the extensive study on SEM methods form the basis of the discussion:
• Research on SEM encompasses various purposes, primarily focusing on SAM, SWPM, and SAQM. The study of water pollution, air quality, soil moisture, and humidity aids in the modeling and design of healthy environment systems that contribute to smart agriculture for sustainable economic growth.
• Methods under each purpose are categorized based on sensory data used, machine learning methods applied, IoT devices employed, and types of sensors utilized. The study mainly emphasizes the impact of existing research on water quality monitoring, air quality assessment, SEM applications, and smart agriculture systems.
• In most SEM methods, particularly SAM and SWPM, researchers predominantly use CNN-based deep learning methods, with other deep learning models being less frequently employed.
• Sensory data exhibit variability across various SEM applications, with no robust data universally utilized for a maximum number of methods. Data types and regions of interest differ among various research works.
• Methods are predominantly used for either classification or prediction, such as classifying water as polluted or clean or predicting the quality degradation of water and air.
• Challenges in SEM systems vary among applications, but common challenges include interoperability issues with heterogeneous sensors, limited sample sizes, and noisy data from internal and external factors.
• Traditional machine learning methods, like SVM and neural networks, are frequently used for data training and classification, while fuzzy-based methods and deep learning approaches suffer from big data issues or significant computational complexity.
There is no reported robust machine learning approach applicable to address environmental challenges across various monitoring and control purposes, data types, and sensor variations.
Research trends were also analyzed to gauge the volume of research in SEM, as depicted in Table 5. The analysis spanned from 1995 to 2020 and revealed an increasing trend in research employing IoT and WSN, as well as research utilizing IoT and machine learning. Interestingly, research employing modern machine learning methods still lags behind studies without machine learning, highlighting the ongoing significance of traditional methods. However, when IoT devices are integrated and deployed within a WSN, the role of AI becomes indispensable.

Table 5: Quantum of Research Contributions in IoT and WSN, and IoT and Machine Learning Integration.

Figure 4 depicts the analysis of research trends, emphasizing two primary categories: SEM utilizing IoT and WSN, and SEM employing IoT and machine learning. The trends indicate that the widespread implementation and study of SEM with machine learning-based training for subsequent classification or prediction are still emerging. Although research shows a consistent increase each year, the more prominent impact of IoT and WSN is evident in Figure 4.

Figure 4: Evolution of Trends in SEM Methodologies.
Based on the preceding discussion and analysis, we propose the following recommendations for the enhancement of more effective, resilient, and intelligent environment monitoring systems:
• Development of a comprehensive framework for machine learning methods is essential.
• Design and implementation of a robust set of classification, prediction, and forecasting models capable of operating on diverse data, regardless of the SEM's intended purpose.
• Implementation of suitable denoising methods as pre-processing stages in SEM to address challenges encountered in existing research where de-noising and appropriate pre-processing have proven inadequate.
• Exploration and incorporation of data deduplication approaches and other strategies to address big data challenges identified in significant studies.
• Recognition of the critical role of SEM in achieving sustainable development goals for any nation, with a specific emphasis on smart agriculture and smart environment applications. However, challenges persist in setting up necessary infrastructure, particularly in rural areas of developing and underdeveloped nations, necessitating governmental involvement at both local and global levels.
• Addressing interoperability issues associated with the implementation of various sensor types by developing suitable standards and protocols, ensuring compatibility of data for all acquisition and analysis systems.
• Acknowledgment of the challenges posed by limited infrastructure for IoT, WSN, and other sensors in rural areas of many developing and underdeveloped nations, requiring concerted efforts at both local and global governmental levels.
• Consideration of insights from major observations in select review articles on SEM, even though extensive reviews specifically focused on SEM were challenging to locate. This review process prompted an examination of key contributions addressing environmental challenges arising from primary factors, leading to conclusions and recommendations for the design of a robust SEM system capable of addressing various challenges through an AI and sensor technology framework.

DETAILED DESCRIPTION OF DIAGRAM
Figure 1: Intelligent Water Quality Monitoring System (IWQMS) featuring Cloud-based Connectivity through Internet of Things (IoTs) and Sensors.
Figure 2: Environment Dynamics Monitoring System (EDMS) utilizing Wireless Sensor Networks (WSNs) and IoT Devices.
Figure 3: Intelligent Agricultural Surveillance System Employing IoT Devices and Sensors.
Figure 4: Evolution of Trends in SEM Methodologies. , Claims:1. IoT-enabled Environmental Monitoring and Pollution Control System claims that Recognition of the necessity for extensive research in deep learning, specifically addressing challenges related to big data and noisy data issues in SEM.
2. Realization of the imperative need for a robust classification framework, focusing on water quality and air quality monitoring as integral components of smart agriculture systems capable of addressing environmental challenges.
3. Acknowledgment of major challenges in the implementation of smart sensors, AI, and WSN, demanding strategic solutions for sustainable growth through SEM.
4. Advocacy for active participation from environmental organizations, regulatory bodies, and increased general awareness to bolster SEM initiatives.
5. Recognition of the importance of pre-processing poor-quality sensory data using appropriate filters and signal processing methods, enhancing data suitability for subsequent SEM tasks.
6. Identification of the future scope for the study to encompass other environmental factors such as sound pollution and disasters.

Documents