The Application and Research of New Digital Technology in Marine Aquaculture

: Marine aquaculture has become an important strategy to enable the ecological and sustainable development of ﬁshery due to the decreasing natural ﬁshery resources. To solve farming pain points, improve farming efﬁciency and modernize ﬁsheries, new digital technologies, such as the Internet of Things, big data, cloud computing, artiﬁcial intelligence and blockchain, are increasingly being widely applied in aquaculture. This paper introduces the interrelationship of new digital technologies and the framework of their application in marine aquaculture. The results of the application of each new digital technology in marine aquaculture are highlighted, and the advantages or problems of each new digital technology in marine aquaculture are pointed out. Further, the application of new digital technologies in deep-sea aquaculture facilities is enumerated. Finally, the main problems faced by new digital technologies in the process of marine aquaculture production and the future development trend are sorted out and summarized to provide scientiﬁc reference for promoting the wide application of new digital technology in marine aquaculture.


Introduction
Survey data released by the Food and Agriculture Organization (FAO) shows that aquaculture is one of the fastest growing fields of food production in the world, and marine aquaculture has a substantial potential for development [1].By 2014, the production of aquaculture in the world had surpassed the fisheries production [2].The continuous growth of marine aquaculture has attracted attention from many fields, including marine aquaculture environment, aquatic product quality and safety, and supply chain traceability, among others [3].In terms of aquaculture technology, the application of new digital technology in fisheries has gradually made the global fishery farming develop in the direction of intensification and intelligence, and the aquacultural environment has gradually transitioned to a sustainable fishery farming system, which has significantly improved the efficiency of aquaculture [4,5].
China's marine aquaculture and fishery informatization began their preliminary development almost at the same time in 1978.However, the policy direction of China's marine fishery changed from "fishing" to "aquaculture" due to the gradual development of overfishing and marine pollution [6].These have been deeply developed and comprehensively integrated since 2003 and have improved the database and information network of marine fisheries.Therefore, researchers can predict and apply the information services to aquaculture information using detection techniques by combining data mining techniques, such as machine learning and neural networks [7,8].Moreover, intensive intelligent aquaculture facilities, such as water quality monitoring [9], intelligent feeding [10], automatic cleaning [11] and automatic catching [12], have been developed based on new technologies, such as sensor technology, radio frequency identification (RFID), and machine vision.However, the investment and efficiency of many factors, such as fishery labor, aquacultural range, chine vision.However, the investment and efficiency of many factors, such as fishery labor, aquacultural range, and mechanical power, are not proportional to the increase in efficiency in recent years.Ultimately, the investment in fishery science and technology is inadequate; the efficiency of fishery technology is much lower than that of other fields of agriculture [13,14].In addition, China is currently in an important transition stage from traditional aquaculture and processing modes to intelligent aquaculture and processing modes, and the imperfect fishery industry chain and the incomplete implementation of sustainable development strategy have currently become the primary bottlenecks [15].We will use the new integration of modern technology to improve the marine fishery supply chain, enhance the sustainability of marine fishery production, improve the ecological environment of marine fisheries, and finally break through the bottleneck of marine aquaculture.
New digital technologies primarily consist of the Internet of Things (IoT), big data, cloud computing, artificial intelligence (AI), and blockchain.The IoT is used for data collection.Big data technology is used to optimize data.Cloud computing technology is used to operate and maintain data.AI technology is used to process data, and blockchain technology is used to encrypt it.The five digital technologies are closely related and almost cannot be separated.This paper describes the interactions and application framework of new digital technologies, focuses on the application and impact of new digital technologies in the development of marine aquaculture, and introduces the development status of marine aquaculture equipment.In view of the advantages and shortcomings of the application of new digital technologies to marine aquaculture, reference suggestions are put forward for the construction of modernization system of marine aquaculture.

New Digital Technologies
The five new digital technologies represented by the IoT, big data, cloud computing, AI and blockchain enable data collection, data sharing, data analysis, data encryption, data prediction and optimal decision making.On the one hand, new digital technologies can transform traditional industries in an all-round and whole-chain manner and improve total-factor productivity.On the other hand, new digital technologies can also integrate with other fields to create various new industries, new business models and new modes, triggering multi-field, multi-level and systemic changes.The relationship and function of new digital technologies are inseparable, and together they constitute a digital virtual network as shown in Figure 1.In Figure 1, the IoT is the lowest information source that supports higher-level digital technologies.It collects all types of data through a series of intelligent devices to enable the identification, positioning, and monitoring of the physical environment, which provides the most accurate data to develop digital technology [16].All types of data collected by the IoT technology need to be stored, operated, and maintained within a certain period of time.Therefore, cloud computing provides configurable computing resource sharing pool services to help clean big data, structure, and integrate enormous amounts of data, so that effective data can meet the needs of users [17,18].AI replaces the human brain to deeply mine data and convert it into effective information, which can make efficient decisions in a short period of time.Moreover, it can also return decisions to the IoT devices instead of manual control [19].Each link above needs a central organization to ensure its reliable operation.For this reason, the characteristics of blockchain provide a solid foundation of trust for digital virtual networks and ensure the real operating environment of the IoT, big data, cloud computing and AI, forming a "1 + 1 > 2" enhancement effect [20].
Simultaneously, the rapid integration of new digital technologies and marine fisheries has not only promoted the development of intelligent fishing vessel research and development, marine information management and seafood supply supervision, but also provided a reference basis for in-depth exploration in the field of marine aquaculture [21].Figure 2 shows the logical framework of technical application of new digital technology in marine aquaculture.First, information on fisheries and aquaculture is collected by IoT devices, such as drones, unmanned ships, underwater wireless sensors, and underwater robots, and then uploaded and stored to various servers through different types of transmission methods.AI can mine and analyze these data and finally provide a corresponding analysis and decisions based on the needs of aquaculture to promote the accurate operation of fisheries, improve production efficiency, and enable the automation of marine aquaculture production, management informatization, and intelligent decision making.In Figure 1, the IoT is the lowest information source that supports higher-level digital technologies.It collects all types of data through a series of intelligent devices to enable the identification, positioning, and monitoring of the physical environment, which provides the most accurate data to develop digital technology [16].All types of data collected by the IoT technology need to be stored, operated, and maintained within a certain period of time.Therefore, cloud computing provides configurable computing resource sharing pool services to help clean big data, structure, and integrate enormous amounts of data, so that effective data can meet the needs of users [17,18].AI replaces the human brain to deeply mine data and convert it into effective information, which can make efficient decisions in a short period of time.Moreover, it can also return decisions to the IoT devices instead of manual control [19].Each link above needs a central organization to ensure its reliable operation.For this reason, the characteristics of blockchain provide a solid foundation of trust for digital virtual networks and ensure the real operating environment of the IoT, big data, cloud computing and AI, forming a "1 + 1 > 2" enhancement effect [20].
Simultaneously, the rapid integration of new digital technologies and marine fisheries has not only promoted the development of intelligent fishing vessel research and development, marine information management and seafood supply supervision, but also provided a reference basis for in-depth exploration in the field of marine aquaculture [21].Figure 2 shows the logical framework of technical application of new digital technology in marine aquaculture.First, information on fisheries and aquaculture is collected by IoT devices, such as drones, unmanned ships, underwater wireless sensors, and underwater robots, and then uploaded and stored to various servers through different types of transmission methods.AI can mine and analyze these data and finally provide a corresponding analysis and decisions based on the needs of aquaculture to promote the accurate operation of fisheries, improve production efficiency, and enable the automation of marine aquaculture production, management informatization, and intelligent decision making.

IoT Applications
IoT for fisheries is essentially a sub-technology of agricultural IoT, which can conduct multi-source perception and detect fishery activity, such as marine aquaculture, fishing, processing, circulation, and management.It is primarily applied to aquaculture environment monitoring, fish activity monitoring and aquaculture product data collection.
In the aspect of aquaculture environment monitoring, wireless sensor network (WSN) is widely used to monitor the aquaculture environment, and the Zigbee (802.15.4) standard is the basic infrastructure of most wireless sensor networks, which have good applications in factory aquaculture, pond farming, etc. [22].However, it is only applicable to smallscale networks and is affected by the range of sensor networks, which complicates its application in the marine environment alone [23].In a complex marine environment, the interaction of network information is affected by large transmission noise, long transmission distance and less offshore energy, which results in a high rate of data packet loss and a delay in transmission, which poses an enormous threat to data security [24,25].To solve these communication problems, technologies, such as a low-power wide-area network (LPWAN) and long-range radio (LoRa), are emerging, which not only enable efficient data transmission over long distances in harsh sea conditions, but also help the underwater Internet of Things (UIoT) to transmit underwater data, images and video, which provide a reference for the large-scale transmission of ocean data.Simultaneously, some researchers have optimized and improved the IoT technology and applied it to marine aquaculture.For example, Schmidt et al. [26] designed a low-cost (<£5000), robust and easy to extend autonomous water quality monitoring buoy system for the aquaculture environment using miniature wireless sensors with a deviation of 0.03-0.08 for temperature sensors and less than 0.05 for conductivity sensors; its accuracy meets common requirements and can withstand wind and waves well.Wada et al. [27] monitored seawater temperature, dissolved oxygen (DO) concentration, salinity, chlorophyll, and ocean currents using mobile satellite communications.The average delay time for monitoring was 15 s; with a minimum delay time of 6 s and a maximum delay time of 656 s, the monitoring effect is good.
In the aspect of fish activity monitoring, the combination of optics, acoustics and IoT enables effective counting or estimation of fish abundance.Aguzzi et al. [28] used HD cameras connected to fiber optics, which acquired the images of active fish and introduced recount errors, acoustic analysis and time series analysis to determine fish activity abundance at cable stations along the Atlantic coast.Sheehan et al. [29] used baited remote underwater video (BRUV) to automatically detect mobile marine fish, which is cost effective, easy to deploy, non-invasive and non-destructive.Hassan et al. [30] proposed the Internet of Fish (IoF, a radio technology for real-time fish telemetry applications) based on acoustic telemetry and LoRa, which achieves the real-time monitoring of acoustic telemetry data of fish in marine culture nets.This technology is not affected by wired-based receivers or communication via ultra-high frequency/very high frequency (UHF/VHF) or the Global system for Mobile Communications (GSM), and the farthest node from the gateway (2.5 km) has a communication success rate of more than 92% and a packet error rate of less than 2%.
In the aspect of aquaculture product data collection, the most common practice is to attach seafood to retroactive RFID tags when fishing vessels collect data on seafood.Users can transmit the information of aquaculture, fishing, storage, and transportation collected via sensors to the RFID reader through the wireless sensor network.They can also input the information by manually holding the RFID reader and scanning the label [31].However, there are still many problems in the supply chain application based on RFID, such as insufficient popularity of intelligent devices [32], high implantation costs [33], and strong limitations [34].In addition to the collection of information on the production environment in a complete marine fishery production supply chain, it is necessary to collect the data of production facilities.Liu et al. [35] utilized micro-service architecture to build BlueNavi, a maritime information platform which can not only provide users with data stored in remote data centers through the internet, but also receive data locally through devices connected to workstations in an offline state.This provides solutions for the positioning and communication of marine fishery facilities.To simultaneously record the complete production and transportation information of seafood, Hao et al. [36] proposed to combine RFID technology and the Beidou Navigation Satellite System (BDS) to enable the real-time monitoring of the entire transport and logistics process of the catch.
The use of IoT technology in the process of marine fishery production can save a substantial amount of time and energy, improve labor efficiency, and enable automatic identification, positioning, monitoring and management.On the one hand, the development of underwater IoT, LoRa, optical and acoustic telemetry has gradually solved the problem of difficult monitoring and transmission of marine information.On the other hand, RFID technology, GPS, BDS and other related technologies can realize the comprehensive collection of marine fishery information.In addition, improvement in the supply chain management system for seafood not only provides rich information resources for marine fishery management, but also provides data that serves as a basis for seafood traceability safety.

Big Data Applications
During the process of applying big data in marine aquaculture, the data that is acquired should be stored and managed in accordance with certain standards and analyzed using appropriate information fusion and mining technology.This reconstructed multidimensional data is then combined with effective knowledge and presented to users through scientific visualization.This would assist the understanding, application, and decision making of the processes of marine fishery production [37].Table 1 introduces the characteristics of the primary sources of marine fishery big data.It can be found that marine fishery big data can be obtained from a wide range of sources, including the IoT, the internet, professional databases, and marine fishery management systems, among others.Professional databases are fast to obtain and have high authority, but are moderately difficult to obtain.Its applications include aquaculture water quality analysis, comprehensive benefit analysis of aquaculture and visualization of aquaculture data.In the aspect of aquaculture water quality analysis, a high-quality aquaculture environment is the core factor to ensure yield and quality.In particular, marine fish have high requirements for the quality of water, and the adverse changes of water parameters, such as dissolved oxygen and pH, will directly affect the normal growth of fish [38].Real-time analysis and the prediction of water quality parameters in aquacultural environments are of great significance to identify biological abnormalities in aquaculture, prevent diseases, and reduce corresponding risks [39].In inland marine aquaculture, it is critical to analyze multiple environmental factors in a culture pond to determine its suitability for fish farming.Farmers can monitor the levels of salinity, temperature, oxygen content, pH, and oxidation reduction potential (ORP) in a pond through the IoT.The correlation between individual environmental factors and other environmental factors can be analyzed through big data, which can help farmers determine whether a pond is suitable for white shrimp farming and also help them to understand how to improve the overall water quality by adjusting individual factors when the water quality deteriorates [40].In offshore marine aquaculture, the maricultural environment is always open to the surrounding environment, and the changes in water quality parameters are usually non-linear, dynamic, variable and complex.Missing data or errors are a common phenomenon in marine aquaculture.It is very difficult to ensure the integrity of statistical marine aquaculture data and it is very easy to affect the performance of data analysis or prediction models.The use of big data analysis alone cannot manage such a variable marine environment and data integrity.Thus, it is necessary to use an artificial neural network (ANN), deep learning and other methods to analyze and predict the water quality parameters in this scenario [41], as described in Section 3.3.
In the aspect of comprehensive benefit analysis of aquaculture, it is well-known that recurrent disease outbreaks are closely related to the farming environment and that farmers spend a substantial amount of money on antibiotics and chemicals to prevent diseases.To develop marine aquaculture in a sustainable way, it is necessary to conduct big data analysis of the carrying capacity of the aquaculture environment and the cost of aquaculture.Today, a series of models, indicators and methods have been developed to study the relationship between marine aquaculture and ecosystem components to determine the marine aquaculture carrying capacity [42,43].For this reason, the ecological model EcoWin, a powerful framework for aquaculture management, has been used by Santa et al. [44] to process ecological big data and add key ecological variables to the model to analyze the carrying capacity of bivalve shellfish farming in Saldanha Bay, South Africa, to maximize the social, economic, and environmental benefits of bivalve shell-fish aquaculture.Moreover, to improve the stability of cost control of marine fishery farming, Gu [45] proposed a method to control the costs of marine fishery farming based on a big data analytical algorithm.The analysis showed that the method is highly stable at controlling the cost of cultivating marine fisheries and adapting these fisheries to control costs, which increases the profitability of enterprises.
In the aspect of visualization of aquaculture data, owing to the dynamic, heterogeneous, multidimensional, and extensive characteristics of marine fishery big data, it is necessary to organize, integrate, and visualize them to provide information to monitor the marine fishery environment and engage in decision-making applications [46].Using SpringBoot framework and JPA framework to develop collection automation and visualization of the big data platform can enable the analysis of data, which provides information on aquaculture to assist users and help experts to manage the heterogeneous water quality [47].The development of simulations of the marine environment and representation of information has resulted in the realistic results of visualizing marine data, and the real-time requirements of this process are increasingly high.Su et al. [48] developed large-scale software for measured marine hydrographic data to visualize marine hydrological environmental data using a 3D geometric rendering technique and Delaunay algorithm to enable the stereoscopic display of a spatial area that extends from a single spatial point or sectional line of the ocean.The software can efficiently and intuitively simulate and display the nature and changing processes of marine water environmental factors.Simultaneously, a standardized spatial database is needed to assist in planning future fisheries for the sustainable development of marine fisheries.Clawson et al. [49] collected and compiled big data on marine aquaculture areas of all countries (73) that produced more than 500 tons of a single marine aquaculture species in 2017, analyzed and mapped global marine aquaculture using R version 4.0.2, and visualized global marine aquaculture areas, which provided support and a reference to establish a global database of standardized marine aquaculture areas.
The ability of big data to quantify and analyze marine fishery production has a significant impact on improving the efficiency of production; therefore, farmers are able to analyze conditions such as aquaculture water quality and aquaculture sustainability with a view to making corresponding aquaculture strategies.However, the marine environment is extremely complex and needs to integrate various factors, such as the time range, geographical environment, and economic conditions.In addition, various marine aquaculture information systems can collect a large amount of data on the marine aquaculture environment; furthermore, the visualization of marine aquaculture information can help farmers to monitor the aquaculture environment in all aspects and guide the production management of aquaculture.

AI Applications
We can obtain sufficient marine fishery data through the IoT and big data.However, AI can utilize the collected data to quickly make more effective and faster judgments, analyze them and make decisions.AI almost always includes data mining and machine learning in theory, which assists computers in creating models based on experiences and accurately predicting future events instead of human operations [50].The applications of AI in marine fisheries include water quality forecast, the identification and classification of fish, and the ability to make feeding decisions and to detect fish biomass.
The influencing factors of water quality and environment in marine aquaculture are more and complex, the variety of environmental sensing sensors to be used is large, the variables are not easily controlled, and the data samples are prone to anomalies or loss.Therefore, it needs to be combined with artificial intelligence models for prediction and analysis, and Table 2 summarizes some of the work related to AI for predicting water quality in marine aquaculture.Owing to the particularity and complexity of marine organisms, much of the acquisition of unstructured data depends on the visual technology of machines.At the level of fish identification and classification, accurate fish identification facilitates accurate management, scientific aquaculture, and control of the density of aquaculture [59].A substantial amount of research has been conducted by researchers in the fields of identification of fish [60], detection of their age [61], identification of their sex [62], classification of their species [63], identification of their diseases [64] and the identification of dead fish [65], which provides technical support for the production management of intelligent aquaculture.Simultaneously, because the fish are often in motion, machine vision is prone to erroneous identification.For this reason, PelagiCam [29], a system to remotely monitor the mobile biomass on fish farms, can simplify the acquisition of biological data in the impact assessment of floating structures at sea and avoid bias in monitoring other fish on farms.In addition, observation with unmanned aerial vehicles (UVA) can effectively obtain the movement of fish in the sea, which is an important reference for monitoring other fish species in marine cages [66].
It is well-known that scientific aquaculture has higher requirements for feeding strategies.If there is insufficient feed, the fish will grow more slowly, which will prolong the aquaculture cycle.This would result in increased costs and aquacultural risks.In contrast, the use of excessive feed could lead to eutrophication of the water and damage the water quality of aquaculture areas.Thus, it does not facilitate the healthy growth of fish.The influence of many factors, such as temperature differences in different regions, aquaculture feeding periods, the nutritional requirements of fish, and market specifications, has now transformed the traditional manual feeding based on experience to knowledge-driven machine feeding [67], whose feeding strategy is based on fish behavior, feed allowance, and other relevant factors.Table 3 describes the application of some AI to feeding decisions in marine aquaculture.
Table 3.Some applications of AI to feeding decisions in marine aquaculture.

Key Technology Function Source of Decision Reference
DANet-EfficientNet-B2 To identify short-term feeding characteristics of shoals in specific feeding areas Behavior of fish Yang et al. [68] Two-stream recursive Network (DSRN) To identify the feeding and non-feeding behaviors of salmon in the absence of significant motor features Behavior of fish Måløy et al. [69] The improved YOLO-V4 To detect uneaten feed particles in underwater images

Residual feed amount
Hu et al. [70] Quadratic vector machine To determine if the fish are hungry based on data on the activity of electromyography Electromyography Cubitt et al. [71] Adaptive fuzzy reasoning To determine the feeding intensity by the dissolved oxygen saturation and temperature Aquaculture environment Zhao et al. [72] MLR, artificial neural networks and SVM Predictive model for shrimp biomass based on four main variables: water temperature, dissolved oxygen, pH and total feeding to develop feeding strategies Growth model Chen et al. [73] The detection of fish biomass is an important basis for aquaculture activities, such as classification, feeding, and the estimation of yields.Therefore, the combination of machine vision and machine learning to detect fish biomass can accurately estimate the size [74], weight [75], quantity [76] and other biological information of fish and improve the efficiency of detection.These technologies are sustainable, do not cause damage to the fish or contact them, and are inexpensive.Traditionally, workers on an assembly line cut the fish.This process involves complicated and repetitive procedures and makes it difficult to ensure the freshness of fish.Gamage et al. [77] first applied machine learning to fishcutting tasks.A CCD camera was used to capture images of fish heads and process them to obtain edge-enhanced two-dimensional Gaussian smooth images.The features were then calculated from the enhanced images and fed into a multiple regression algorithm to estimate the optimal cut points on the fish head.In addition, AI should be used to classify the fish, segment them, and assess them for freshness based on the biomass identified above.Odone et al. [78] proposed a fish grading system based on machine vision.The system takes measurements of the fish shape and uses a support vector machine (SVM) model to determine the relationship between fish weight and shape parameters.This system can grade fish at a rate of three items per second.Wu et al. [79] took the large yellow croaker as an example and designed an automatic system to control the grading line.This system can extract the body characteristics of large yellow croakers, establish a classifier, and extract the freshness based on the sensory index characteristics to enable the classification of quality.
As AI applied in marine fisheries and aquaculture is increasing, it also faces a series of challenges.For example, the visual technology that is used to estimate biomass and identify fish solves the problem of low efficiency of traditional artificial fishing measurement, avoids damage to the fish bodies, and improves the efficiency of labor.However, owing to the low visibility and complex environment in underwater environments, fish images that are acquired based on light vision require additional processing.Problems, such as the overlap of fish bodies, poor resolution, and the small fuzzy outline of target fish, still pose difficulties in current research.The intelligent analysis of information and its subsequent prediction of water quality performs well in real-time performance and is highly efficient, but it needs to solve the problems of easy corrosion and the high cost of water quality sensors in seawater.Intelligent feeding decisions can save feed and improve the efficiency of labor, but accurate automatic feeding requires the establishment of accurate models that can monitor fish growth, estimate biomass, and analyze fish-feeding behavior all while overcoming environmental interference and sensor errors.

Blockchain Applications
As an emerging technology, blockchain supports decentralized architecture.It provides secure sharing of data and resources across all nodes of the IoT network and eliminates centralized control, and it can help marine fisheries in many ways, including providing transparent resourcing for marine conservation, reducing pollution from plastics, enhancing seafood traceability, securing farming data and sustainably managing fisheries [80,81].
The globalization of seafood markets has resulted in challenges in the maintenance of seafood quality throughout the supply chain.As a result, many marine fisheries research and applications based on blockchain, such as the World Wildlife Fund (WWF) in Australia, Fiji, and New Zealand, which partners with the US-based tech innovator ConsenSys, the tech implementer TraSeable and the tuna fishing and processing company Sea Quest Fiji Ltd. (Suva, Fiji), use blockchain technology to track tuna from "bait to plate" in the Pacific Island tuna to help stop illegal tuna fishing in the Pacific islands [82].In addition, seafood of the same batch will go through multiple modes of transportation, processing, and sale after it is caught.Therefore, it is particularly important to ensure the authenticity of the information at each stage of the seafood supply chain.Figure 3 shows the seafood supply chain process based on blockchain.
The reasons for the consumer distrust of seafood largely comes from its source.Therefore, the source of the supply chain-the fishing ground-is one of the areas that most needs protection.Hang et al. [83] has developed a blockchain-based fisheries platform to ensure the integrity of aquaculture data.The design platform is designed to provide a secure storage for fishermen to hold a large amount of untampered marine aquaculture data.The various processes in the fishery are automated through the use of smart contracts to reduce the risk of error or manipulation.Zhang et al. [84] developed a new frozen aquatic product traceability system (BIOT-TS) based on blockchain and IoT technology for use in processing and production, which can alleviate the shortcomings of traceability management in the current process of cold chain logistics, such as weak safety performance, low efficiency of the management of centralized data, and the easy tampering of traceability information.
Stephane et al. [85] proposed that blockchain could be used as a seafood-tracking scheme, which could track the data collected during the process of seafood trading.Once a problem is found, it relies on the immutable characteristics in the blockchain and takes the data on the chain as evidence for reporting, which provides a guarantee of higher quality and security.In addition, ShrimpChain [86], the Ethereum Blockchain [87], FishCoin [81] and other frameworks, applications, or schemes have been proposed to efficiently manage the fishery supply chain in a decentralized, transparent, traceable, secure, private and trustworthy manner.
The globalization of seafood markets has resulted in challenges in the maintenance of seafood quality throughout the supply chain.As a result, many marine fisheries research and applications based on blockchain, such as the World Wildlife Fund (WWF) in Australia, Fiji, and New Zealand, which partners with the US-based tech innovator ConsenSys, the tech implementer TraSeable and the tuna fishing and processing company Sea Quest Fiji Ltd (Suva, Fiji), use blockchain technology to track tuna from "bait to plate" in the Pacific Island tuna to help stop illegal tuna fishing in the Pacific islands [82].In addition, seafood of the same batch will go through multiple modes of transportation, processing, and sale after it is caught.Therefore, it is particularly important to ensure the authenticity of the information at each stage of the seafood supply chain.Figure 3 shows the seafood supply chain process based on blockchain.The reasons for the consumer distrust of seafood largely comes from its source.Therefore, the source of the supply chain-the fishing ground-is one of the areas that most needs protection.Hang et al. [83] has developed a blockchain-based fisheries platform to ensure the integrity of aquaculture data.The design platform is designed to provide a secure storage for fishermen to hold a large amount of untampered marine aquaculture data.The various processes in the fishery are automated through the use of smart contracts to reduce the risk of error or manipulation.Zhang et al. [84] developed a new frozen aquatic product traceability system (BIOT-TS) based on blockchain and IoT technology for use in processing and production, which can alleviate the shortcomings of traceability management in the current process of cold chain logistics, such as weak safety performance, low efficiency of the management of centralized data, and the easy tampering of traceability information.Stephane et al. [85] proposed that blockchain could be used as a seafood-tracking scheme, which could track the data collected during the process of seafood trading.Once a problem is found, it relies on the immutable characteristics in the blockchain and takes the data on the chain as evidence for reporting, which provides a guarantee of higher quality and security.In addition, ShrimpChain [86], the Ethereum Blockchain [87], FishCoin [81] and other frameworks, applications, or schemes have been proposed to efficiently manage the fishery supply chain in a decentralized, transparent, traceable, secure, private and trustworthy manner.
The research and applications described above have many advantages for upstream and downstream enterprises and consumers in the supply chain.For example, suppliers can improve the quality and safety of aquaculture and establish an effective mechanism to trace the origin of seafood to meet the intentions of consumers when they purchase seafood.Processing enterprises can optimize the production and processing processes based on the labeling of real seafood, reduce costs and losses, and improve the accuracy The research and applications described above have many advantages for upstream and downstream enterprises and consumers in the supply chain.For example, suppliers can improve the quality and safety of aquaculture and establish an effective mechanism to trace the origin of seafood to meet the intentions of consumers when they purchase seafood.Processing enterprises can optimize the production and processing processes based on the labeling of real seafood, reduce costs and losses, and improve the accuracy of seafood quality control.Storage and transportation companies can optimize the processes of product delivery and improve the efficiency of temporary inventory management and selection.Consumers can provide feedback in time to protect their rights and interests when there is a problem with seafood.However, after a series of interviews with Australian prawn farmers, Garrard [88] proposed that blockchain could not provide substantial help to the aquaculture industry.On the one hand, members of the supply chain cannot determine whether the prawn data recorded by other members is reliable.Alternatively, there is no reliable way to identify and track seafood.As a result, the supply chain is independent of the storage of seafood data.It is difficult to ensure the reliable source of seafood in the production source or during processing.
The first challenge of marine fisheries blockchain is that it needs to rely on other scenarios and technologies to achieve maximum effect, making it highly dependent.For example, the number of marine IoT nodes is far lower than that of terrestrial IoT.Blockchain and IoT technologies will face simultaneous difficulties in application scenarios.In this case, the problem can be solved by simplifying the consensus algorithm, reducing the number of nodes required for consensus confirmation, and clarifying the weight ratio between important and common nodes.Therefore, if the limitations of IoT nodes in offshore ocean fisheries can be broken, the security of fishery data will be substantially improved.Another challenge comes from the inability to ensure the reliability of the information being fed into the blockchain.While blockchain applications can collect data on fisheries and trace them from marine aquaculture to consumer purchase, there is no guarantee that the products themselves or the hardware deployed are not fraudulent.Thus, the effective adoption of this technology still faces a number of policy challenges.

The Application of New Digital Technologies in Deep-Sea Aquaculture Facilities
We know that marine fish farming, taking advantage of the natural conditions of the deep sea, has become the best option for maintaining a continuous, stable and high-quality supply of fish products.However, for the traditional artificial culture, it is certainly a challenge to carry out the culture in the open sea exposed to wind and wave action.The rise of new digital technologies in recent years has driven the development of intelligent deepsea aquaculture facilities, especially in China, where the development of such facilities has increased dramatically (Table 4), and Figure 4 shows typical deep-sea aquaculture equipment designed and manufactured in China.The equipment is equipped with approximately 20,000 sensors that enable fully automatic monitoring, feeding, and other work.
First harvest of 150,000 salmon, quality all meet EU export standards Dehai 1 10,000-ton intelligent fish farm that consists of a mixture of plate structure floating body and truss structure farming area.

2018.9
This intelligent fishery is structurally safe, can evaluate aquaculture efficiently, and does not require manual control of the aquaculture.
After more than 8 months of growth, the average weight of each fish is 750 g Zhenbao 1 Deep-sea abalone mechanized farming platform

2018.10
The platform can monitor various parameters of abalone production and the surrounding environment in real time and enable the intensification and standardization of abalone aquaculture.
The weight of each abalone is 10 to 15 g heavier than that in the bay Changjing 1 Deep-water seated-bottom aquaculture large-net cage

2019.4
The net cage is integrated to automatically lift net clothes, feed automatically, and monitor underwater and other automatic equipment, which can enable system timing and quantitative and efficient automatic control.

Annual production of over 1000 tons of black and yellow croaker
Fubao 1 Deep-sea intelligent and environmentally friendly abalone farming platform

2019.4
The platform contains oxygenation, water quality monitoring, scenery complementary, and systems for the supply of daily power, sewage treatment, video monitoring and remote data transmission.
Over 40,000 kg production in 2022 Penghu Semi-submersible wave energy farming tourism platform

2019.6
The platform is equipped with desalination and sewage treatment equipment, and the aquaculture system is equipped with modern fishery production equipment, such as automatic feeding, the monitoring of fish and water quality, and the transmission of live fish.
The maximum capacity to raise common fish is about 300 tons Changyu 1 5G technology-enabled marine ranching platform 2020.6 The platform is equipped with a big data monitoring system for marine pastures, which can enable real-time monitoring of marine data, such as meteorology, water temperature and quality, flow speed, and direction.It is also the first platform to carry a 5G communication base station.
100,000 pounds of adult fish can be harvested in two years

Characteristic Farming Benefits
Genghai 1 Intelligent multi-functional ecological marine pasture complex platform 2020.7 The platform has automatic feeding, environmental monitoring, and ship anti-collision functions, and it is also loaded with unmanned ships, underwater inspection robots and other technical equipment, which can enable intelligent control of the entire production process.The platform is equipped with a digital farming management system, an intelligent robot operating system, and a biological monitoring and operation system.Sun et al. [91] summarized the research directions of patents on aquaculture facilities in China, including water quality monitoring, the control of water levels in aquaculture ponds, box aquaculture devices, feed storage and feeding, technology to detect fish, the treatment of aquaculture water, communication technology, signal monitoring and early warning devices, among others.The essence of applying new digital technology to intelligent marine aquaculture control and management is to use sensors, underwater robots, big data analysis and other technologies, combined with fish growth models, to achieve real-time monitoring, early warning and prediction of water quality, so as to carry out accurate feeding, and then use machine vision and other technologies to automatically screen and grade finished fish to achieve automation of marine aquaculture production, management information and intelligent decision making.Take the example of aquaculture vessels: Dong et al. [92] designed a set of centralized automatic feeding systems that could be controlled remotely, operated at quantitative and fixed speeds and diagnose faults to meet the requirements for the intensive and automatic control of multiple aqua- Sun et al. [91] summarized the research directions of patents on aquaculture facilities in China, including water quality monitoring, the control of water levels in aquaculture ponds, box aquaculture devices, feed storage and feeding, technology to detect fish, the treatment of aquaculture water, communication technology, signal monitoring and early warning devices, among others.The essence of applying new digital technology to intelligent marine aquaculture control and management is to use sensors, underwater robots, big data analysis and other technologies, combined with fish growth models, to achieve real-time monitoring, early warning and prediction of water quality, so as to carry out accurate feeding, and then use machine vision and other technologies to automatically screen and grade finished fish to achieve automation of marine aquaculture production, management information and intelligent decision making.Take the example of aquaculture vessels: Dong et al. [92] designed a set of centralized automatic feeding systems that could be controlled remotely, operated at quantitative and fixed speeds and diagnose faults to meet the requirements for the intensive and automatic control of multiple aquaculture tanks of a 100,000 ton deep marine aquaculture vessel.The automatic feeding system can reduce the influence of rolling and pitching of the vessels on the accuracy of measuring the weighing sensor and reduce the influence of pneumatic conveyance in different pipeline lengths on the outlet feed speed and crushing rate.Huang et al. [93] developed a set of centralized control systems for vessels, which enabled the monitoring and centralized display of aquaculture parameters, automatic control of water exchange, and dissolved oxygen, production management and other functions.The verification of this system on real vessels substantially reduced the workload of personnel, reduced the probability of human error, and ensured the normal development of aquaculture production.[94] designed an intelligent aquaculture vessel with functions such as aquaculture, processing, refrigeration, a fish-sucking operation, and evaluation of the aquaculture load, which improved the dynamic efficiency of the aquaculture vessel and reduced its aquaculture load.In addition, for deep-sea cages, additional considerations need to be considered, such as cleaning the net coat, catching fish, and resisting wind and wave settlement.
A review of the use of marine aquaculture by other world powers shows that they have researched and developed marine aquaculture equipment earlier; thus, they have higher aquacultural efficiency.For example, Ocean Farm 1 is the world's first offshore fish farm with a capacity of 1.5 million salmon [95].In addition, Norway's NORDLAKS has proposed an enormous ship-shaped fish farm with a multi-network system called Havfarm 1 that can hold 10,000 tons of salmon at a time [96].Moreover, the marine farm "Jostein Albert" can accommodate 10,000 tons of salmon at a time, and it has the same maximum density of salmon as other aquacultural farms (25 kg per 1000 L of water, 2.5% salmon and 97.5% water) [97].Interestingly, they were both built in China but developed in Norway.Therefore, their construction in China was recognized as a milestone to accelerate the development of offshore aquaculture in China.In addition, more powerful countries in the field of marine aquaculture have developed not only some advanced integrated large-scale aquaculture facilities, but also more small-scale stand-alone equipment, such as feeders and net washers, as shown in Table 5.
Table 5.The equipment of countries other than China in the field of marine aquaculture.

MarinaCCS (feeding system) [98] Norway
A variety of sensors and cameras are used to monitor the underwater environment for accurate feeding, and self-developed software ensures the stability of the system.
The system can handle more than 40 feed lines running in parallel and more than 1000 tanks/units FEED-MASTER (feeding system) [99] America PLC-based control technology effectively solves the problem of feeding machine damage to the feed, with high reliability and high feeding accuracy.
Feeding The use of dynamic positioning systems allows the ship to automatically change its position and direction using propellers and propellers.
Holds up to 10,000 tons of salmon JOSTEIN ALBERT Norway 2020.3 It is equipped with functions to realize automatic fry transportation, automatic feed feeding, underwater light monitoring, underwater oxygenation, dead fish recovery, and automatic adult fish search and capture, and can adapt to the extremely cold climate and harsh sea conditions outside the Norwegian fjords.
Holds up to 10,000 tons of salmon.
In addition, fully automated control of the processing line of aquaculture products has been very popular, including quality inspection, boning, slicing, dividing, and product packaging [103].However, the guarantee of hygiene is crucial during the process of processing.Machine learning has been applied to automatically clean the fish bodies during processing, in which machine vision and a convolutional neural network (CNN) [104] can detect residual fish blood on the surface cleaned by the production line robot.A genetic algorithm (GA) [105] is more suitable to optimize the robot that automatically cleans the fish body.The combination of the two can enable the detection and cleaning of fish body debris in the production line.Simultaneously, the supply chain of aquaculture products involves multiple links, and the traceability is based on electronic record keeping.Machine learning alone cannot guarantee the authenticity of the data in a lengthy supply chain.Thus, Chen [106] provided certain references for the quality and safety traceability of aquacultural products based on the IoT traceability bar code and data encryption technology.Wei et al. [107] uses HACCP to control the aquaculture link, which enables the accountability of supply chain supervision based on blockchain.Currently, the traceability of aquaculture products is primarily focused on the entire process of fishing, processing, transportation, and sales of products.The government will perform real-time tracking of information, such as information on the environment, production, processing, logistics and sales of marine products, and establish traceability and mechanisms to share information.To this end, the blockchain provides important assurance that the data is authentic and also enables consumers to determine whether the aquaculture products are reasonably priced.

Conclusions and Outlook
This paper expounds on the development status of new digital technology, analyzes the application of new digital technology in contemporary marine aquaculture by researchers in China and elsewhere, and points out the important role and key pain points of new digital technology in the application of marine aquaculture in China.

1.
In marine aquaculture, the wireless communication technology of IoT enables the real-time monitoring of water quality, and the data can be processed in real time and fed back to the terminal equipment.With the rapid development of online platforms, fishery managers can obtain a large amount of data related to marine aquaculture at a low cost.Owing to the special factors of marine aquaculture environment, such as climatic conditions, geographical location, and biological diversity, not only are marine sensors susceptible to corrosion, but they also cause obstacles to underwater communication functions.This results in the continuous collection of aquaculture data that are often missing or inaccurate.Therefore, data mining and machine learning models must overcome the complexity of incomplete datasets.

2.
Machine vision technology can reliably estimate marine biomass in real-time, in a contact-free manner, non-destructively, and safely.This feeds the visual estimation of marine biomass back to the big data platform for analysis and decision and then combines the feeder and catching equipment to enable intelligent feeding and accurate fishing.Machine learning technology can provide relatively efficient technical methods for marine aquaculture in data processing, information extraction, real-time monitoring, decision management, and other aspects.This describes the developmental trend of enabling intelligent marine aquaculture.However, underwater image processing and the collection of other types of information are still difficult owing to the multiple interference of the marine environment light changes and the complex and diverse organisms.

3.
The combination of blockchain and IoT technology can trace any violation or security problem to the relevant links in the supply chain of marine aquaculture products, which ensures the authenticity and reliability of product information and enhances the trust of consumers in products.Simultaneously, data related to the choice of seafood by consumers can also be analyzed by big data and AI to select high-quality seafood and feed back to the aquacultural source to adjust the strategy for aquaculture.However, the separation of the IoT from human intervention and methods to ensure the authenticity and reliability of the data before the IoT is an important problem that merits additional study.
Currently, China's offshore marine intelligent aquaculture equipment construction achievements are increasing yearly, and the development trend is sound.However, most of them have not gone through several complete aquaculture cycles and cannot form standardized intelligent marine aquaculture equipment.In addition, in other countries around the world, some work still requires manual participation.For example, safety inspectors need to conduct diving operations, and the complex underwater environment makes their work very dangerous.The application of underwater robot inspection is the developmental trend for intelligent and automated marine aquaculture equipment.With the continuous development and improvement of new digital information technology, the manufacturing cost of intelligent aquaculture system is bound to continuously decrease, and manual operations, such as feeding, catching and underwater detection, will eventually be replaced.Overall, marine aquaculture needs not only modern aquaculture technology, but also digital marine aquaculture facilities to form a highly informative and intelligent farming system.We should increase the scientific research investment of new digital technology in marine aquaculture, guide marine aquaculture to the direction of unattended, quality and quantity, green and environmental protection, make full use of marine resources, and help the marine aquaculture and technology field to progress continuously.

Figure 1 .
Figure 1.Diagram of a digital virtual network.

Figure 1 .
Figure 1.Diagram of a digital virtual network.

Figure 2 .
Figure 2. The logical framework of the application of new digital technology in marine aquaculture.

Figure 2 .
Figure 2. The logical framework of the application of new digital technology in marine aquaculture.

Figure 3 .
Figure 3. Application of blockchain to the seafood supply chain.

Figure 3 .
Figure 3. Application of blockchain to the seafood supply chain.

First
aquaculture vessel integrates the technologies of ship engineering, seawater aquaculture, and seed aquaculture and introduces six key technologies, such as shipboard cabin raising, water exchange, shake reduction, vibration and noise reduction, dirt cleaning and corrosion prevention, and intelligent set control. or facilities for automated control, feeding, water quality testing, fish harvesting, net-washing patrol marine robots, real-time marine monitoring and marine environmental weather monitoring.Annual production capacity of 2000 tons J. Mar.Sci.Eng.2023, 11, x FOR PEER REVIEW 13 of 20 seed aquaculture and introduces six key technologies, such as shipboard cabin raising, water exchange, shake reduction, vibration and noise reduction, dirt cleaning and corrosion prevention, and intelligent set control.Ningde 1 Deep-sea semi-submersible farming platform 2022.11Equipped with systems or facilities for automated control, feeding, water quality testing, fish harvesting, net-washing patrol marine robots, real-time marine monitoring and marine environmental weather monitoring.

Table 1 .
Comparison of big data acquisition sources for marine fisheries.

Table 2 .
Relevant research that used AI to predict the water quality in marine aquaculture.

Table 4 .
Intelligent deep-sea aquaculture facilities in China.