Marine Internet of Things Platforms for Interoperability of Marine Robotic Agents: An Overview of Concepts and Architectures

Abstract

The creation of a Marine Internet of Things platform, including the Underwater Internet of Things, is needed to ensure the interaction and digital navigation of heterogeneous marine robotic agents. It is necessary to combine the following robotic agents: autonomous underwater vehicles, remotely operated vehicles, active and passive marine sensors, buoys, underwater sonar stations, coastal communication posts, and other elements of the platform. To ensure the interaction of all these elements, it is necessary to use a common communication system within the platform, as well as a common navigation and control system to solve complex problems of the navigation and control of the movement of robotic agents in order to implement a joint mission to collect and transmit data, including video information in real time. The architecture of the Marine Internet of Things platform must first be defined in order to use a unified approach to data exchange. This article provides an overview of approaches to determining the architectures of network underwater and marine communication systems based on the concept of the Internet of Things. This paper provides a comprehensive study of MIoT applications, challenges, and architectures. The main contributions of this paper are summarized as follows: we introduce potential MIoT applications; we point out the challenges of MIoT (i.e., the differences between MIoT and IoT); and we analyze the MIoT system architecture.

1. Introduction

The universal implementation of the information network concept of physical objects equipped with built-in technologies for interaction with each other and the external environment (the concept of Internet of Things—IoT) is a sustainable trend of modern societal development and one of the significant indicators of world economy transformation to the sixth technological mode. Due to the global spread of wireless networks, the development of machine-to-machine interaction technologies, and the growth of computing powers, the gradual replacement of humans in routine maritime operations by “smart” robotic information technology systems can be predicted. The projection of the IoT concept onto the maritime industry and the expansive and irreversible nature of its implementation, together with the fundamental nature of human maritime activity, will ensure radical growth and the long-term development of promising maritime technology sectors. It is possible to identify the following main applications of IoT technologies: scientific, industrial, military, etc.

Scientific applications are associated with marine environment observation, i.e., the monitoring of geological processes on the ocean floor, the marine environment, and marine life. For example, IoT-based marine environmental monitoring applications include [1,2,3,4,5,6,7,8]:

• Ocean sounding and monitoring—a general system for monitoring the marine environment, formerly a long-established system using oceanographic and hydrographic research vessels;
• Water quality monitoring—generally for tracking water conditions and quality, including water temperature, pH, turbidity, conductivity, and dissolved oxygen content in ocean bays, lakes, rivers, and other bodies of water;
• Coral reef monitoring—tracks coral reef habitat and the environment;
• Monitoring deep-sea fish farms—tracks water state and quality, including temperature and pH, measures fecal waste and uneaten food, as well as fish condition, including number of dead species;
• Wave and current monitoring—measures waves and currents for safe and reliable navigation;
• Ocean pollution monitoring—includes chemical and biological analysis of ocean pollution and temperature analysis;
• Analysis of pressure and temperature changes for given areas;
• Monitoring oil and gas field areas and pipelines.

IoT technologies allow the implementation of new models of underwater research in the tasks of underwater archaeology, seabed mapping, natural resources search, and many other applications.

Table 1. Summary of existing MIoT applications.

Reference	Key Features	Perspectives	Main Research Objects	Country
Salhaoui et al. [1]: MDPI (remote sensing)	AUV model system that overcomes latency challenges in the supervision and tracking process by using edge computing in an IoT gateway	Extension to hybrid cloud/edge architecture	AUVs, AI	Spain
Jahanbakht et al. [2]: IEEE Communications Surveys & Tutorials	Architectural challenges analysis	To cover new tools and techniques, as well as to make informed decisions and set regulations related to the maritime and underwater environments around the world	Big Marine Data (BMD)	Africa
Kong et al. [3,15,16]: IEEE Photonics Journal, Hindawi	The first underwater optical wireless sensor network prototype. Real-time digital video surveillance	Popularization of the future human–robot interaction applications	Sensor nodes, underwater visual monitoring	China
Domingo et al. [4]: Journal of Network and Computer Applications	The IoUT is introduced and its main differences with respect to the Internet of Things (IoT) are outlined	Detailed description of application scenarios that illustrate the interaction of IoUT components	IoUT architectures	Spain
Xu et al. [5]: MDPI (sensors)	The potential application of IoT and Big Data in marineenvironment protection	Description for potential application of IoT and Big Data in marine environment protection	BMD	China
Kao et al. [6]: MDPI (sensors)	Investigation and evaluation of the channel models	The channel models to further investigate the design of different IoUT communication protocols, such as the MAC protocols and routing protocols will be used	Underwater Wireless Sensor Networks (UWSNs)	Taiwan
Khan et al. [7]: IEEE Xplore	A completely decentralized ad hoc wireless sensor network for the ocean pollution detection	To substantiate making it feasible purchase and deploy many underwater sensor nodes	Sensor nodes	India
Alippi et al. [8]: IEEE Sensors	Solar-powered WSN framework for aquatic environmental monitoring	To provide quantitative indications related to cyclone formations in tropical areas	Wireless sensor networks (WSNs)	Australia
Manjula et al. [9]: IEEE Xplore	A scheme for sensor deployment design aimed at optimal coverage of the monitoring area with minimum number of sensor nodes	To consider other space filling structure	Sensor nodes, UWSNs	India
Khaledi et al. [10]: Systems and Information Engineering Design Symposium (SIEDS)	Design of an underwater mine detection system	Reducing the approximate system cost	AUVs, sensor nodes	USA
Cayirci et al. [11]: Ad Hoc Networks	A new wireless sensor network architecture is introduced for underwater surveillance systems where sensors lie in surface buoys when nodes are first deployed	–	UWSNs, sensor nodes	Turkey
Cardia et al. [13]: Mobihoc’ 19	System that supports real-time monitoring of divers’ positions and health conditions, at the same time allowing unprecedented enhanced visits of the sites	–	UWSNs	Italy
Coutinho et al. [14]: Q2SWinet’ 19	The challenges for the design of TC (topology control) algorithms for IoUTs	New research directions will be tackled when considering the new advancements and characteristics of IoUTs	Network topology	Canada
Marini et al. [17]: MDPI(Marine Science and Engineering)	H2020 ENDURUNS project that describes a novel scientific and technological approach for prolonged underwater autonomous operations of seabed survey activities, either in the deep ocean or in coastal areas	To develop new applications in seafloor exploration and surveying	Monitoring systems, AUVs	Italy
Qin et al. [18]: IEEE Access	An autonomous underwater vehicle (AUV)-assisted hierarchical information acquisition system composed of a marine stationary sensor layer and an AUV motion layer	To involve new AUV path planning strategies	AUVs, sensor nodes, UWSNs	China
Lin et al. [19]: Chinese Journal of Mechanical Engineering	The future trend of the ocean observation systems with docking technology and sustained ocean energy	–	Ocean energy, AUVs, monitoring systems	China
Binnerts et al. [20]: IEEE: MTS/IEEE Kobe Techno-Oceans (OTO)	Development and demonstration of live data-streaming capability using an underwater acoustic communication link	It is planned to collect more representative channels at open sea for further testing and optimization	Acoustic communications	Netherlands
Lu et al. [21]: arXiv	The cognitive ocean network (CONet) is proposed and described	Using next-generation artificial intelligence technology	IoUT architectures	China
Saha [22]: IEEE Xplore	IoT-based automated fish farm aquaculture monitoring system	To develop a better way to capture images and use better image processing techniques to provide better results	Sensor nodes, ocean monitoring	Bangladesh
Li et al. [23]: Journal of the World Aquaculture Society	The major challenges and future trends of underwater object counting in aquaculture are discussed	To implement new counting tasks in aquaculture	Ocean monitoring	China
Wang et al. [27]: IEEE Communications Surveys & Tutorials	The concept of machine-type communication (MTC) for maritime IoT and its services, requirements, and challenges	To avoid the potential pitfalls in the development and standardization of maritime MTC technology	IoUT architectures, MTC	USA
Xia et al. [28]: IEEE Wireless Communications	An intelligent energy control scheme named the residence energy control system (RECoS) is proposed	To provide the sufficient attention of MIoT that it deserves in the 5G community	AI, ocean energy	China
Yang et al. [29]: IEEE Network	Explanation on how various AI methods can facilitate the operation of the parallel-network-driven maritime network	To speed up the AI methods	AI, IoUT architectures	China

summarizes the main investigations examined in this paper.

Within the framework of maritime accident and disaster prevention, IoT technologies make it possible to implement subsystems, e.g., flood warning, earthquake and tsunami warning, and maritime navigation.

Monitoring and research tasks based on IoT technologies can be solved on both small and large scales, from the implementation of fish observations in the aquarium to the tasks of monitoring large water areas of the seas and oceans. Examples of model applications in monitoring tasks are given in [2,3,4,6] and many other works.

Military security applications include protecting waters, securing port facilities, ships in harbors, mine clearance, and communications with submarines and divers [4,9,10,11,12].

The mentioned application scenarios usually refer to Internet of Underwater Things (IoUT) or Underwater Internet of Things (UIoT) technologies [2,3,4,6,13,14,15,16].

There are three steps for implementing IoUT [15]. The first step is to implement a dynamic, continuous, comprehensive and intelligent real-time perception of the underwater environment. Over the past few decades, underwater sensor networks consisting of a variety of instruments such as conductivity, temperature and depth sensors, biological sensors, and current sensors have enabled the large-scale, long-term, continuous collection of oceanographic data [17]. Demand for real-time multimedia data increases with new applications based on human–robot interactions, such as underwater pipeline inspection, underwater volcanism and hydrothermal source studies, seafloor mapping, tactical observation, and underwater rescue [15]. The second step in the implementation of IoUT is the transmission of large amounts of underwater data in real time. Mobile platforms such as autonomous unmanned underwater vehicles (UUVs), remotely operated unmanned underwater vehicles (ROVs), and underwater gliders have opened new ways to create mobile underwater networks and have become key tools for high-quality video monitoring [18,19,20]. The third step in IoUT implementation is the intelligent processing of massive underwater data. In the past, due to a lack of facilities and low investment, the acquisition of marine data took several years or months and, therefore, the amount of data was small. In recent years, with the continuous development of appropriate information technology and equipment, the period of data collection has gradually shortened and the amount of data has increased exponentially. In the future, large data, cloud computing, artificial intelligence, and virtual reality will be the dominant technologies for the implementation of the intelligent processing of large amounts of data [21]. Therefore, the real-time dynamic and visual monitoring of underwater scenes is becoming one of the important research areas for future human–robot interaction applications.

Industrial marine IoT applications designed to track and monitor commercial activity largely focus on both subsea and surface IoT technologies. The IoUT market segment for industrial applications in fisheries and aquaculture, which deals with fish farming in confined spaces, can be singled out separately. The task of identifying each fish, for example, using RFID tags and transmitting various information about their condition to a central server for further processing, is solved here. The central server, in turn, processes the data and develops methods of control and decision making. Another group of tasks consists of searching and analyzing large groups of fish and marine animals in open waterbodies [4,5,22,23].

It should be said that, in addition to IoUT, in the literature there is the concept of Internet of Things Ocean (IoTO) [24] or Ocean of Things. The IoTO technology itself is described as a network of interconnected intelligent underwater objects. IoTO is a promising technology for the systematic management of a variety of marine data. The fields of application of IoTO marine monitoring include: monitoring of coral reefs; monitoring of marine fish farms (on the shelf or in the open ocean); monitoring of water quality; monitoring of waves and currents, etc. In this sense, we will consider OIoT and IoUT technologies as identical.

Separately, it is worth highlighting the maritime shipping segment, for which the concept of Maritime Internet of Things (Maritime IoT) was introduced [25,26,27,28,29], defining that all devices related to surface marine objects are connected by an information network for uninterrupted service around the world. Such applications and services currently include [27]: search and rescue, e-navigation, collision avoidance, marine environment monitoring, etc.

A concept similar to Maritime IoT—“e-Navigation” [30]—to support various types of navigation services was originally developed to modernize the maritime industry by the International Maritime Organization (IMO). Subsequently, other services were added, which are listed above. Therefore, Maritime IoT is a distributed hardware–software complex of information data transfer from various devices of above-water ocean engineering systems and objects, utilized through an information system (usually the Internet, or a marine information network or a network of marine services) through a unified machine-type communication.

Starting from the considered fields of application and expanding the scope of data exchange, including the transfer of information data (this includes video information, measurement results, etc.) from devices and objects located underwater, and receiving control commands by them, the scope of IoT expands and, as we see it, combining Maritime IoT and IoUT leads to the consideration of a group of technologies which will be called Marine IoT (MIoT)

The MIoT platform will be considered as a multidisciplinary hardware–software system of data and control command exchange between surface and underwater ocean engineering objects and systems, including ships, buoys, sensors, ROVs, various telemetry facilities, etc.

The purpose of this article is to review MIoT technologies and their applications in the maritime domain. The development of the MIoT platform requires defining its architecture in order to use a unified approach to data exchange. This article provides an overview of approaches to determine the architectures of underwater network communication systems based on the concept of the Internet of Things.

This paper presents some of the maritime Internet of Things platform prototype results, performed during the demonstration tests. This fact increases the importance of the conducted research, especially in the direction of the MIoT architecture choice. The general scheme of the platform is shown in Figure 1 and will be described in detail further in the text. It can be seen here that the IoT is an important component of the MIoT.

Figure 2 shows the layout of the test equipment. List of essential equipment: surface buoy with hydroacoustic modem uWav, marine communication buoy, hydroacoustic station “Manjetka”, receiving station of hydroacoustic modems and onshore station for AUV’s “Marlin-350” control. The maneuvering area of the AUV with the hydroacoustic modems installed on them is marked in yellow on the scheme.

The paper is structured as follows. Section 2 presents the features and differences between MIoT and IoT. An analysis of the MIoT system architecture is described in Section 3. At the end of the article, a discussion of the results and conclusions are presented.

2. Features and Differences of MIoT from IoT

The design, development, and deployment of MIoT-based marine research and exploration systems are necessary to achieve several critical objectives, including increased autonomy, adaptability, scalability, and ease of implementation. When designing, developing and deploying MIoT-based marine environmental exploration and development systems, the following system and component requirements specific to the marine environment and tasks to be performed should be taken into account:

• The heterogeneity and versatility of the system elements determine that the system must provide intermediary communication, providing interaction between underwater, surface, air and ground elements (underwater robots, surface ships, buoy systems and underwater stations, UAVs, coastal operator stations);
• Various possibilities of communication channels (especially bandwidth) require the development of a special model for the information interaction of system elements;
• Computational limitations for data processing operations by on-board computers;
• Low power consumption and the ability to generate and store energy leads to the need to apply energy conservation, generation and storage measures in the system elements;
• High equipment reliability imposes higher demands on equipment reliability due to aggressive marine environment. Sensors, actuators and other assemblies are required to have a very high level of watertightness. There is a need for auxiliary devices, such as buoys and mooring devices, etc.

Table 2. Features and differences of MIoT from IoT.

Features	MIoT	IoT
Communication technologies	Most communications in the IoUT are based on acoustic links.	Mostly radio waves.
Tracking technologies	In the MIoT, things (usually fish) are tracked with different technologies: acoustic tags, radio tags, passive integrated transponders.	The IoT uses basic radio frequency identification (RFID) for tracking.
Battery recharge	Battery capacities are limited and it is difficult (sometimes impossible) to recharge or replace them.	As part of the IoT, replacing batteries is not difficult.
Energy-harvesting technologies	Piezoelectric energy harvesting can also be exploited in the IoUT. The IoUT also benefits from specific underwater energy-harvesting techniques such as ocean thermal energy.	Two of the most promising energy-harvesting technologies for IoT devices are solar energy and piezoelectric harvesting.
Network density	The IoUT is deemed to be sparse due to the cost and challenges associated with underwater deployment.	In the IoT, it is expected that a very large number of devices communicate if all the ‘things’ join the network.
Localization techniques	Terrestrial localization approaches: the localization with directional beacons (LDB) scheme.	The location of mobile devices in the IoT is afforded by global positioning system (GPS) satellites

summarizes the key differences between MIoT and IoT, providing an overall picture.

The MIoT and the IoUT have some similarities with their IoT counterpart, for example, in terms of structure and function. However, there are also differences, which are primarily reflected in the communication technologies used. The main features of the various communication technologies used in MIoT systems are discussed below.

2.1. Hydroacoustic Communication Means

Radio waves are poorly propagated underwater. The most important limitation of radio frequency communication (RFC) in fresh water is large antenna size, and in seawater is high attenuation. Consequently, most communications in the IoUT use acoustic channels as a base.

Although the acoustic wave is accepted as the main underwater physical transmission medium, the long propagation delays of acoustic waves and the high error rate in the underwater acoustic channel complicate data exchange.

The heterogeneity of the acoustic channel properties (primarily due to the change in sound velocity with depth) causes refraction of the signal and results in a channel with internal variation. As a result, shadow zones are formed, which cause significant bit rates and loss in communications. The available bandwidth of acoustic channels underwater is severely limited and depends on both transmission range and frequency. Long-range systems operating at several tens of kilometers can have a bandwidth of only a few kHz, while short-range systems operating at several tens of meters can have a bandwidth of over hundreds of kHz. In addition, acoustic waves are affected by turbulence caused by tidal waves, acoustic noise, and pressure gradients. Negative effects of sound propagation especially affect hydroacoustic communication (HAC) in shallow water (at depths up to 100 m). In such a situation, temperature gradients, surface ambient noise, and multipath propagation due to reflection and refraction have a negative effect [31,32].

There are many companies involved in the development of underwater hydroacoustic communications. Some of them are presented below.

“Water linked” is one of the world’s leading companies in the production of underwater sensors and establishing an underwater Internet of Things ecosystem. The main product is an underwater modem: Modem M64 [33,34]. The modem provides a reliable acoustic link between two underwater locations where space, weight or power are limited. Data are transmitted at 64 bps over a reliable two-way, 31–250 kHz half-duplex acoustic link that uses the latest technology to provide reliable data transmission over 200 m in the most challenging circumstances. The low-power electronics, consuming 2.6 W, are packaged in a tiny miniature case. Typical applications where the Modem M64 can be used include AUV telemetry, wireless ROV control, and wireless underwater sensor monitoring/control.

“Popoto Modem”—the company’s core business modem—includes acoustic data transmission, software device communication, and digital data processing. The main product is the “Popoto Boardset” modem. [34,35]. The “Popoto” modems provide reliable data transmission at different bitrates, from 80 bit/s up to 10,240 bit/s at distances of 3–8 km. Easy to use APIs such as Python, Matlab, C++ or JSON-based sockets, and several apparatus interfaces such as RS-232, RS422, Ethernet, and GPIOs make this modem easy to integrate and use. The output power of the modem is up to 20 watts. “Popoto S-series” also provides sub-meter accuracy using a two-way range, making it easy to know the distance to your modem at any time.

“Underwater Communication and Navigation Laboratory” (“UC&NL”) is a Russian developer and manufacturer of commercially available devices and systems for underwater wireless navigation and communication, used in a wide range of underwater works both by divers and ROVs/UUVs [36]. The main products of the company are:

(1).

Hydroacoustic micromodem “uWave”. The device allows you to provide hydro-acoustic digital communication between 20 callers within a radius of up to 1000 m with a transmission speed of 70 bps, using the code mode and transparent channel mode. The extremely small size of 41 × 45 mm, low power consumption, and ease of use make “uWAVE” the ideal solution for AUV control, as well as data transmission.

(2).

Hydroacoustic modem “uWAVE Max”. The device is designed to transmit data over a hydroacoustic channel at a distance of 3 km, even in complex hydrological conditions and in shallow water at a speed of 80 bps. The small size of 64 × 62 mm, low power consumption, and ease of use make uWAVE Max the ideal solution for both small ROVs/UUVs and larger devices.

“EvoLogics” is a German manufacturer of hydroacoustic devices. EvoLogics’ underwater acoustic modems provide full-duplex digital communications using patented S2C (sweep-spread carrier) technology, providing excellent performance that is immune to the challenges of dynamic underwater environments. Self-adaptive algorithms adjust the S2C parameters to maintain the highest possible data rate under given conditions. The company’s main products are [37,38]:

(1).

“S2C M 30/60” underwater acoustic modems. The “S2C M 30/60” modems and USBL transceivers with near hemispherical transducer diagrams are high-speed devices for efficient transmission in reverberant waters, providing data rates of up to 31.2 kbps in the 1000 m range.

(2).

“18/34H” underwater acoustic modems. These are devices for communication in reverberant shallow waters, providing data rates up to 13.9 kbps in the 3500 m range. The wide-angle beam pattern is optimal for vertical, oblique and horizontal transmissions. The high operating frequency ensures high performance even in noisy environments.

(3).

“15/27” underwater acoustic modems. “15/27” modems are deep-lying devices for long-range transmissions, providing data rates up to 9.2 kbps in the 6000 m range. The beam pattern with wide coverage is optimal for vertical and oblique transmissions to/from stationary systems. Nominal depth and operating frequency provide high performance for long-range transmissions between the seafloor and the surface.

“Develogic subsea systems” is a German company focused on the development and production of customized turnkey data collection and telemetry solutions for maritime monitoring. The main product is a hydroacoustic modem, “HAM.BASE” [31,39]. This modem is designed to satisfy the research and exploration requirements for various submarine communication projects. Flexible modulation and coding tasks, a variety of available transducers in different frequency bands of 40–65 kHz, a high maximum transmission power of 10 kbit/s, energy efficiency, and flexible interfaces are the reasons for the powerful solution of underwater communication damping. HAM.BASE is optimized for transmission ranges up to 1000 m.

For a convenient comparison of the technical characteristics of hydroacoustic modems,

Table 3. Hydroacoustic modems’ technical characteristics comparison.

Manufacturer	Modem	Bandwidth, kHz	Power Consumption, W	Range, km	Transmission Rate	Data Buffer	Maximal Depth
Water Linked	Modem M64	31–250	2, 6	0, 2	64 bps	1 Mb	200 m
Popoto Modem	S100RP	20–40	20	1–2	80–9000 bps	N/A	100 m
	S1000RP	20–40	20	1–4	80–10,240 bps	N/A	2000 m
UC&NL	uWave	10–30	1–6	1	70 bps	127 bytes	300 m
	uWAVE Max	5–20	1–10	3	80 bps	127 bytes	400 m
EvoLogics	S2C M 30/60	30–60	10	1	31.2 kbps	1 Mb	2000 m
	18/34H	18–34	7	3, 5	13.9 kbps	1 Mb	2000 m
	15/27	15–27	10	6	9.2 kbps	1 Mb	6000 m
Develogic subsea systems	HAM. BASE	40–65	100	1, 2	10 kbps	N/A	6000 m

shows the main information.

An overview of other hydroacoustic modems developed by universities such as the University of Calabria, Massachusetts Institute of Technology, University of New York at Buffalo, National University of Singapore, and several others, can be found in [31,32,34].

When using multiple digital HAS devices with a random mutual overlap of “audibility” zones, they must, at a minimum, provide multiple access points to the signal propagation environment (in particular, collision resolution when receiving data packets), as well as the routing of data streams from source to destination. In general, to merge sensor information from heterogeneous stationary and/or mobile devices collected over a large area, both on the surface and underwater, a representation (coverage of the current situation and achievement of integral/situational awareness in the sensor system coverage area) of the usage of hydroacoustic modems is a necessary but insufficient condition.

Creating a stack of specialized environmental access control layer protocols and network protocols capable of efficient networking in the hydroacoustic environment is essential for fusing sensor information from heterogeneous assets, both on the surface and underwater, assembled over a large area.

The development of technical means of digital hydroacoustic sensors is hampered by the need to simultaneously consider combinations of several effects of the hydroacoustic environment on the receiver side: a very limited frequency band, long delays in signal propagation (propagation speed is several orders of magnitude lower than that of the radio signal), reverberation of high intensity and duration, the formation of large shadow zones, multipath signal fading, large Doppler distortion, rapid variability of the environment characteristics, and opposite side differences in the characteristics of communication channels.

Compared to radio, these effects are much more intense in hydroacoustics. Consequently, methods and techniques for digital communication in hydroacoustic networks must be able to counteract these effects. At present, the impact of these effects separately on the ability to communicate digitally under water has been satisfactorily studied. However, the influence of the combinations of these effects is poorly understood. The presence of many such combinations in the hydroacoustic environment makes it difficult to create methods of digital hydroacoustics that would work equally well in a wide variety of signal propagation conditions. This is a promising and relevant direction of research and development in the field of underwater hydroacoustics.

2.2. Optical Communications Means

Underwater wireless optical communication (UWOC) may be considered as an alternative to hydroacoustic channels [3,15,16,40,41,42,43,44,45,46,47,48,49]. Optical communication channels provide greater bandwidth than acoustic channels, but their range is limited due to the strong attenuation of light in water. They can be considered a good alternative for transmitting large amounts of information (e.g., streaming video) and organizing the communication of robot groups, but only under good visibility conditions and over short distances (tens of meters).

Depending on the configuration of the communication channel between the nodes, submarine networks based on the UWOC can be divided into four categories (Figure 3) [40,42,43,44,50]:

• Point-to-point line-of-sight configuration;
• Diffused line-of-sight configuration;
• Direct visibility configuration based on a reflector;
• Non-line-of-sight configuration.

The point-to-point line-of-sight configuration is the easiest to implement. In this configuration, the receiver detects the light beam in the direction of the transmitter. Since point-to-point systems typically use light sources with a narrow divergence angle (e.g., a laser), precise pointing between the transmitter and receiver is required. This requirement limits the application mainly to static transmitters and receivers (e.g., two sensor nodes on the ocean floor). For mobile platforms, such as AUVs, this configuration requires a sophisticated guidance and tracking mechanism to keep both the transmitter and receiver in the line of sight. When organizing such communication for stationary devices, it is necessary to take into account that direct visibility can be blocked by sea creatures, because underwater lights can attract them. It is known that marine fish are attracted to blue–green wavelengths, while freshwater fish are attracted to yellow–green wavelengths. Therefore, in order to avoid fish entering the line-of-sight area, it is preferable to use flashing or random lights [40,51].

Diffused line-of-sight configuration uses scattered light sources with a large divergence angle (e.g., high-power light-emitting diodes (LEDs)). This configuration allows for broadcast data transmission from one node to multiple nodes. In this configuration, the point-to-point requirement is weaker compared to a point-to-point configuration, but in this case the scattered light communication line is subject to more attenuation due to the provision of a larger interaction area.

The reflector-based line-of-sight configuration can be considered a special implementation of a point-to-point configuration. This configuration is used in limited duplex communications where receivers have low power to support full transceiver operation. In this case, the source has more power and payload than the receiver, so it serves as a requester that sends modulated light signals to the remote receiver. The receiver is equipped with a small optical reflector that reflects the incoming interrogating beam back to the source when it detects it. In the modulating reflector, incoming light is reflected back from the modulated reflector. During this process, the information that the reflector transmits to the transceiver will be encoded in the reflected light. One limitation of this configuration is that the backscattering of the transmitted optical signal can interfere with the reflected signal, thereby degrading the system signal-to-noise ratio and bit error rate. Since the optical signals will pass through the underwater channel twice, the received signal will suffer additional attenuation.

The non-line-of-sight configuration has no limitations due to the requirement for the precise aiming of the transceivers. In this configuration, the optical line is realized by means of back reflection of the propagating optical signal at the water–air interface. The receiver must be pointed toward the water surface in a direction approximately parallel to the reflected light to ensure proper signal reception. The main problem with communication lines in this configuration is the random gradients of the water surface caused by turbulence sources [42,43,52]. These effects will reflect the light back to the transmitter and cause signal dispersion. In [40], configuration using scattered light also refers to the configuration outside the line of sight, although it is not quite correct.

The significant growth of UWOC and video fusion technologies in the last few years has allowed the entrance of research that requires dynamic real-time visual monitoring. For example, in [15], the authors developed a prototype of 2K real-time digital video surveillance based on UWOC. The system was named “AquaF-seer”. Using an LED and a diffused line-of-sight UWOC system based on an avalanche photodetector, this prototype achieved real-time video transmission with 1920 × 1080 pixels resolution over various channels (1.5 m free space channel; clean water channel; clean water channel with turbulence; channel with pure sea water simulation and channel with coastal ocean water). The specified result was achieved using a single LED, FPGA and an on–off keying modulation circuit. The prototype demonstrated excellent switching characteristics and reliability in a 1.5 m free space.

In [16], the authors consider automatic underwater operations using real-time UWOC as a new IoUT paradigm. For this purpose, the authors developed a prototype underwater optical wireless sensor network called AquaE-net, which consists of an optical concentrator and two sensor nodes with the ability to measure temperature, electrical conductivity, and pH. Using the AquaE-net prototype, a real-time bi-directional communication network was realized with 100% packet transmission success in free space and muddy water in the harbor of the Red Sea at King Abdullah University of Science and Technology.

Compared to the acoustic means, UWOC has high data transfer rates, low channel delay and low implementation costs. The UWOC can achieve data transfer rates of the Gbit/s order over distances of tens of meters. This high-speed advantage guarantees the realization of many real-time applications, such as underwater video transmission. Because the transmission speed of light in water is much higher than acoustic waves, UWOC channels are immune to latency. UWOC also has better communication safety than acoustic and radio-frequency (RF) methods. Most UWOC systems are implemented in a point-to-point line-of-sight configuration rather than in a scattered broadcast scenario such as acoustic and RF waves, making it more difficult to “eavesdrop”. In addition, the UWOC is much more energy-efficient and cost-effective than its acoustic and RF counterparts. Instead of using large and expensive acoustic and RF transceivers, which consume a lot of power, UWOC systems can implement relatively small and inexpensive optical underwater transceivers such as laser diodes and photodiodes. This advantage could improve the large-scale commercialization of UWOC and accelerate the deployment of underwater sensor wireless networks.

Although optical transmission has many advantages over acoustic and radio-frequency methods, its implementation remains a challenge. The main problems of optical signal transmission are:

• The optical signal is strongly attenuated and diffused;
• Underwater optical channels can be temporarily interrupted due to optical transceiver mismatch and obstacles.

Typically, blue–green lasers or LEDs are used as light sources because of their narrow divergence characteristics, but this requires the precise pointing of transceivers to realize an optical link. Since the underwater environment is turbulent at relatively shallow depths, line-of-sight drift can often occur, especially in applications with vertical communication channels (random movements of the sea surface can lead to loss of communication).

To summarize, it can be said that the use of underwater optical communication due to the wide frequency band makes it possible to ensure a high data transmission speed, which, however, is accompanied by a number of significant physical limitations. The main one is a limited data transmission distance which depends both on physical factors (mainly on water transparency) and features of used equipment. Another limitation is the narrow sector of angles in which the optical signal is transmitted and received, which causes the necessity of the precise mutual positioning of underwater correspondents. As hydro-optical communication is rapidly developing nowadays, it can be predicted that in the near future, AUVs will be equipped with underwater combined (acousto-optical) communication equipment [37,38].

2.3. Radiofrequency Communication Means

Another alternative to acoustic communication channels is electromagnetic waves. Electromagnetic waves are almost unaffected by multipath, turbidity and light in the water. However, the high conductivity of seawater limits the communication range to a few meters.

In fact, extremely low-frequency (ELF) communication systems on submarines are believed to be the only successfully implemented underwater applications of electromagnetic waves. This communication system operated at 76 Hz for U.S. developments and 82 Hz for Russian systems and allowed several characters per minute to be transmitted over distances around the world. In these systems, one-way communication was implemented to transmit a single submarine surfacing signal to organize a higher bandwidth communication session using land-based radio communications. Since broadband analog voice communications over long distances in the water proved unrealizable, it was believed that electromagnetic signals had no other use in the aquatic environment. However, in recent years, there have been many studies that have determined the conditions for a possible effective organization of underwater wireless communications (UWCs) based on electromagnetic waves [53,54,55,56,57,58,59,60,61,62,63,64,65].

Research on the application of RFS for the organization of underwater vehicle group interaction was carried out in [53]. In [54], investigations into the electromagnetic signal behavior in the 2.4 GHz frequency band in underwater conditions using devices compatible with IEEE 802.11 and IEEE 802.15.4 standards were conducted. From the application point of view, it seems that underwater communication in the 2.4 GHz band is useless and has no practical application, since water has a high attenuation of these frequencies. However, as shown in [54], at 2.4 GHz frequency, it is possible to organize stable communication at short distances (up to 17 cm) which can be used for the accurate monitoring of ecosystems.

Studies of electromagnetic signal penetration at megahertz frequencies were considered in [55], where it was shown that a radio frequency signal at frequencies up to 5 MHz can be transmitted up to 90 m in seawater. It has also been shown that audio data modulated to this signal can be transmitted over these distances. These tests were performed in relative shallow water (<5 m), which is known to be a problematic condition for hydroacoustic communication.

In [56], the wireless underwater transmission of electromagnetic signals at frequencies from 100 kHz to 6.35 MHz using orthogonal frequency division multiplexing was considered at a depth of 5 m. The channel throughput was about 10 Mbit/s in the 100 kHz–6.35 MHz range within half a meter at a transmit power of 1 W. Such characteristics are suitable for the organization of contactless data collection by means of submersibles from one or several communication nodes.

It is important to note that RFC, compared with HAC, has a number of advantages. Thus, at frequencies above 10 kHz, electromagnetic propagation is more than a hundred times faster than acoustic propagation. This has important advantages for command delays and network protocols, where many signals must be exchanged. The Doppler shift is inversely proportional to the speed of propagation, so for electromagnetic signals it is much smaller.

Another important advantage and possibility for RFC applications is the effect of the air–water interface. The large refractive angle created by the high dielectric permittivity triggers the signal almost parallel to the water surface. This effect can be used to establish communication between shallow submarine stations and land without the need for repeater buoys on the surface. From this point of view, the aerial path can be a key advantage of RFC in the realization of inter-medium communications.

Increasing the range and bandwidth of RF in UWC applications can be achieved by:

• Allowing electromagnetic signals to cross the water-to-air boundary and provide long-range horizontal communications using the air path, followed by air-to-water signal transmission if necessary;
• Studying the phenomenon of guided waves on the water side of the water–air interface.

Research was carried out in [57] to compare the characteristics of electromagnetic wave propagation in sea water on both sides of the water–air interface in a thin layer of water and in air after crossing the interface in order to determine the optimal operating frequencies, communication ranges, antenna orientation, and vertical displacement of the submerged antenna. It was shown that with a signal power loss of about 70 dB, controlled electric fields across the air–water interface can be transmitted over a distance of 1 km at 10 kHz when the transmitter is at a depth of 5 m (5 m from the air–water interface).

Ref. [58] provides an overview of the most widely used RF-based UWC techniques, and identifies future research directions and recommendations for the usage of 5G communication technologies in this area. It is assumed that 5G will have the greatest impact on applications such as communication between devices (D2D) in the IoUT [64]. The possibility of obtaining high data rates and monitoring underwater activities based on 5G technologies depends on the development of new techniques based on the filter bank multi-carrier methods and generalized frequency division multiplexing. Thus, in [65], an overview of 5G multi-carrier UWC methods is presented which compares different UWC methods and related issues.

As a result, it can be concluded that electromagnetic waves under water have a range of practical applications in navigation, sensing and communication tasks. The bandwidth of the RFC exceeds that of the HAC only when used over a very short distance. Still, it is a propagation mechanism that offers distinct advantages in addition to the use of existing underwater communication systems. The advantages of RFC for underwater navigation, sensing, and communication systems are: electromagnetic waves cross the water–air boundary, which allows the realization of horizontal communication by air over long distances without additional repeaters; less reflection from the surface and seabed; immunity to acoustic noise allows for the effective use of RFC to organize distributed communication and navigation systems for shallow harbors and ports; the possibility of using distributed transducers (radiation cables may perform navigational and communication functions); RFs are immune to marine fouling, not affected by poor visibility (suspended sediment and seafloor rise from operational effects), making them more effective in long-term communication deployments than optical systems; high propagation speed, low Doppler shift, and low propagation latency are especially important for network protocols that require multiple communications for acknowledgement and error checking; high joule-to-bit ratio efficiency; wireless transmission of power and data through the underwater vehicles’ hulls; and RFC provides shorter and more constant time delays, which is very convenient for organizing group control.

2.4. Combined Use of Different Communication Means

The advantages and disadvantages of all three types of communications are summarized in

Table 4. Comparative characteristics of different wireless underwater communication technologies.

Types of Information Transfer	Pros	Cons
Acoustics	- Most commonly used technology; - Transmission of information over long distances (up to 20 km); - Transmission distance up to several kilometers; - Transmission power within tens of watts; - Antenna size ≈10 cm.	- Low transmission speed (up to a few kbps); - Long latency (seconds); - Bulky and power-consuming transmitters; - Damage to marine life; - Signal attenuation depends on transmission range and frequency (0.1–4 dB/km); - Wave propagation velocity: 1500 m/s; - The bandwidth depends on the transmission range: 1000 km < 1 kHz; 1–10 km ≈ 10 kHz; <100 m ≈ 100 kHz. - Transmission frequencies 10–15 kHz.
Radio	- Simple technology for air-to-water communication; - Flows and swirls of water practically do not produce disturbances; - There are no fixed location and directional rules; - Average close-range transmission speed of up to 100 mbps; - Wave propagation velocity ≈ 2.255 × 108 m/s; - The bandwidth ≈ MHz; - Transmission frequencies 30-300 Hz (for direct connection) or MHz (for buoy-based systems); - Transmission power within a few megawatts or hundreds of watts (depending on distance).	- Small range; - Bulky and power-consuming transmitters; - Signal attenuation depends on frequency and water conductivity (3.5–5 dB/m); - Transmission distance up to 10 m; - Antenna size ≈ 50 cm.
Optics	- Fast data transfer speeds (up to Gbps); - Does not depend on the propagation speed in the liquid (the propagation speed is approximately equal to the speed of light); - Cheap and small transmitters; - Wave propagation velocity ≈ 2.255 × 108 m/s; - The bandwidth ≈ 10–150 MHz; - Transmission frequencies 1012–1015 Hz; - Antenna size ≈ 10 cm;	- Problems with the air-to-water transition (due to refraction); - Visible light is quite strongly absorbed and scattered; - Short range; - Signal attenuation: 0.31 dB/m (in clear ocean) −11 dB/m (in muddy water); - Transmission distance ≈ 10–100 m; - Transmission power within a few megawatts.

below.

Implementing dynamic sensor transceiver networks as the foundation of real-time MIoT requires both powerful inter-media communications and the use of robotic assets as mobile nodes in such a network. In practice, marine robots and sensor networks will always require a certain degree of operational consistency, in particular to obtain related measurements and coordinate the operation of actuators. An important task is to ensure the connectivity of sensor measurements and sensor coordination by developing methods and technologies for network intermediary communication; this provides a rapid exchange of information between nodes of a distributed sensor system and/or multiple interconnected network segments among themselves. This development will be in demand, for example, in tasks that require real-time access to network measurement data by the operator. For example, AUVs operating in a group require an intensive exchange of navigational data during cooperative maneuvering, the performance of coordinated measurements, and the possibility of mission modification by the operator on the accompanying watercraft.

In general, it is currently expected that AUVs will become a kind of bridge between the sea surface and the seabed in ocean exploration tasks [18]. In particular, using AUVs as mobile communication nodes, it is possible to create hierarchical communication architectures that use different technologies to transfer the collected information to the surface for subsequent intelligent data analysis. However, with the current state of methods and technology, the technical means of digital HAC are capable of solving only a very limited range of tasks. In this regard, the organization of combined (hybrid) wireless communication systems, taking into account the possibility of using the most effective means of communication (hydroacoustic, radio frequency, optical) to solve the problem of data exchange and interaction between heterogeneous and diverse system nodes in specific operating conditions (shallow water, deep water or air; small or close distances; small or large volumes of data, etc.) becomes obvious.

The performance of UWOC systems may be seriously reduced by the effects of seawater absorption and scattering, turbulence, transceiver misalignment errors, and other influencing factors. All of these undesirable factors may cause frequent communication failures. One possible way to improve the reliability of UWOC systems is to use the acoustic channel as a backup scenario [43,66]. The hybrid system is presented in [43] as two configurations of UWC based on a combination of acoustic and optical means. The first configuration uses acoustic and optical waves as a duplex transmission medium. As an implementation variation on this configuration, the two ends of the link could be mobile submersibles equipped with both acoustic and optical transceivers. When the two communication nodes are a short distance away and the state of the water is clear, the system will use the optical wave as the carrier to achieve high data rates. If the distance between the two nodes is long or the water is muddy, the system will instead use acoustic methods to provide communication. The advantage of this implementation is high flexibility and reliability, but with high power consumption and bulky acoustic transceivers at both ends.

In the second configuration, the system consists of one stationary control platform and several mobile sensor nodes. An acoustic wave is used to transmit control information from the control platform to each sensor node. The optical wave is used in the communication channels between each sensor node as well as in the uplinks from the sensor nodes to the main control platform.

For long-term underwater monitoring tasks, including the real-time video streaming of seafloor imagery, a hybrid underwater wireless sensor network (UWSN) is being considered [67]. A hybrid UWSN consists of several sensor nodes. Each node contains a Bluetooth module, an acoustic module, and an optical module that operates with a green light. Each node can be communicated with, for example, in clear shallow water using an optical signal or in a water/air environment using Bluetooth. This hybrid UWSN can provide an acoustic link of 400 m with a data rate of 300 bps and set the UWOC up to eight meters with a data rate of 330 kbit/s.

A hybrid optical–acoustic UWSN was proposed in [68] to solve the problem of high-speed real-time video and image transmission. In this work, the authors proposed a new energy-efficient contention-based media access control (OA-CMAC) protocol for the optical–acoustic underwater wireless sensor network. The proposed OA-CMAC protocol, based on optical–acoustic fusion technology, combines a delayed access mechanism with multiple access and a carrier control with collision avoidance and multiple access technology based on multiplexing with spatial separation to achieve high-speed and real-time data transmission. The protocol first performs an acoustic “handshake” to obtain information about the location of the transceiver node, ensuring that the channel is free. If not, it performs a delayed access and waits for the next time slot to compete for the channel again. An optical handshake is then performed to determine if the channel state satisfies the optical transmission, and at the same time beam alignment is performed. Nodes transmit data using optical communication. If the channel conditions do not meet the requirements for optical communication, a small amount of data with high priority is transmitted by acoustic communication.

Ref. [69] presented a system that connects the surface and underwater areas, forming an end-to-end joint connection in which the RF and free space optical (FSO) channels transmit data in parallel with the cascade channel UWOC, forming a joint RF/FSO-UWOC system. The proposed hybrid RF/FSO return channel makes the system more reliable, because the optical channel is very sensitive to weather conditions.

In [70], a wireless communication system based on optical and electromagnetic tracking was proposed for monitoring and research into the aquatic environment for various industrial applications. This paper concludes that optical, electromagnetic or hybrid wireless communication modes with high data rates can be an acceptable solution for ROV-ROV and ROV-type communication applications.

Considering all the above features of various communication technologies, it is possible to formulate a hypothesis about the structure and composition of a distributed set of facilities to provide the intermediate communication of various network nodes, which can be heterogeneous and diverse robotic agents (underwater and surface vehicles, drones, buoys and bottom diving stations, etc.) depending on the functions assigned to them and the expected conditions of functioning.

Creating a stack of specialized environmental access control layer protocols and network protocols capable of ensuring the effective functioning of the network in the marine environment is a prerequisite for the fusion of sensor information from heterogeneous means, both on the surface and underwater, collected over a large space.

3. Analysis of Approaches to the Formation of the Architecture of MIoT Systems

3.1. MIoT Network Architecture

TCP/IP is one of the most common architectures used for MIoT systems. The TCP/IP architecture consists of five separate layers: application, transport, network, data link, and a physical layer [2,24].

The data link and the physical layer are sometimes combined to form a four-layer TCP/IP model, in which the combined data link and physical layer are referred to as the network access or network interface layer.

Figure 4 shows a comparison of the TCP/IP layers and the MIoT network architecture. Terrestrial computers and submarine sensors are seen as network endpoints located at the TCP/IP application layer. This architecture covers the entire network, from the submarine application layer to the surface application layer.

Each TCP/IP layer, shown in Figure 2, relies on some protocols to control the data exchange between MIoT devices, regardless of their underlying structure. Based on these protocols, each individual object on the network will know about the data formats used, communication syntax, synchronization methods, security issues, and error control schemes. The layered set of these protocols is called the protocol stack. Similar to IoT, the MIoT ecosystem does not require a universal protocol stack and relies on multiple protocols that coexist at the model levels of a single project. These protocols are required to support different aspects of the MIoT infrastructure for different tasks, including: connectivity to existing IoT and WEB protocols; providing acceptable delay; matching packet length to channel coherence time; overcoming significant signal attenuation; matching bit error rate to application; accommodating low underwater bandwidth; and countering security threats.

The following is a brief description for each of the five levels of this architecture [2,24]:

• The application layer in MIoT is responsible for identifying each individual object (e.g., sensor ID, sensor type, sensor location, etc.) and then collecting data, processing information, and delivering commands. Data collection at the application level includes data detection, recording, and streaming. Some actuators may also belong to this layer to respond to the environment according to given instructions, or based on a machine learning model. These actuators can be recorded and monitored through the MIoT platform. Existing IoT systems have a sufficient number of application layer protocols to meet all technological requirements in various IoT applications. However, not all of these protocols are suitable for MIoT (IoUT) applications. Some of the application-layer protocols rely on broadband operation, while others have protocol headers that are redundant for underwater communication, without any significant impact on communication reliability.
• Transport Layer: Given the nature of the data created by a particular service, the transport layer is responsible for separating the data into packets before passing it through the network gateway. This layer also takes into account the order of the data and its potential loss.
• Network Layer: The main tasks of the network layer include the bi-directional sending of data packets between individual endpoints as well as protocol translation between neighboring layers. These tasks are performed at the network layer by implementing both Internet protocol (IP) and data routing. Along with Internet protocols, there are network layer routing protocols that determine how routers communicate with each other. An efficient data routing scheme is an integral part of any UWC system.
• The channel layer supports underwater sensor access to the surface stations by appropriately sending/receiving data from the physical layer, as well as coordinating the sending of data. This data exchange with the physical environment requires a conversion between data frames depending on the underlying communication technology (acoustic, electromagnetic, and optical).

Nodes and receivers using any of these technologies usually share a physical telecommunications channel and, therefore, scheduled access to the shared channel becomes a mandatory action in each UWC. This necessity gives rise to the concept of channel layer media access control (MAC). The scheduled access provided by MAC helps to avoid or manage data conflicts and ensure reliable real-time data transmission. MAC access control becomes even more important when connecting existing UWCs to the IoUT infrastructure. In this respect, the MAC scheduling protocol can be either collision-free or competition-based. Because of the unique characteristics of underwater acoustic channels, using MAC protocols based on competition in IoUTs is inefficient and expensive. In contrast, collision-free MAC protocols provide higher performance through lower power consumption and increased network bandwidth. Time-division multiple access (TDMA), also known as link-division MAC, is the basic scheduling protocol without conflict in most UWC standards. According to the network topology, in time-division multiple access MAC, the cluster head takes responsibility for coordinating the data frames. On the other hand, the use of ALOHA as a competition-based MAC protocol is also common in UWC applications that do not require any node at all.

5.

The physical layer demonstrates different behavior in response to various propagation modes and different types of channels. Signal attenuation is calculated differently for electromagnetic, acoustic and optical media. Regardless of the types of propagation technologies, an appropriate channel model is needed. This model can be used to predict communication system performance, design optimal node locations, and reduce overall system power consumption prior to actual deployment.

As mentioned in [71], TCP/IP-based solutions cannot initially meet some of the requirements of MIoT, such as presenting the device identifier with a unique name, data origin management, the authentication of trusted elements during communication, etc. Therefore, it is suggested in [72] to use named data networking (NDN) architecture to implement MIoT. Data security-oriented NDN offers authentication for requested data to bind to the requested name. NDN builds a network based on named data, instead of connections between hosts [73]. In NDN, data exchange is managed by data receivers by sending packets containing the name or a name prefix of the desired data. Data senders generate named data packets that can be received by the NDN network through packets with the same name. Once the packet reaches the node that has the requested data, the node returns the data packet, which contains both name and content, as well as the signature of the producer key that binds them. The data packet follows the return path chosen by the packet to return to the requesting recipient. This communication model of securely retrieving named data naturally matches the semantics of MIoT applications. It also allows applications to define their own data names outside of domain knowledge. To deal with various basic communication technologies compatible with the TCP/IP protocol stack, in practice, an NDN overlay can be created.

Additionally known is the delay-tolerant networking (DTN) approach, which solves technical problems in heterogeneous networks without constant network connectivity [74,75,76].

Underwater wireless networks can be considered delay-tolerant networks (DTNs) because of their sparse deployment, node mobility, and severely degraded underwater acoustic link with limited bandwidth.

DTN architecture has been designed to accommodate network connection discontinuity and provide a framework for heterogeneous sensor network gateways [74,75,76]. In this architecture, different IP-based and non-IP-based protocols (TCP/IP, raw MAC protocol, serial line, etc.) can be used for data delivery. The routers that handle packets in retention and forwarding mode are called packet senders or DTN gateways. In this way, the packet architecture operates as an above-ground network that provides a message delivery service to connect heterogeneous parts of the larger network.

In addition to the presented studies on DTN, relatively new research [77] associated with the practical implementation of near-vertical incidence skywave (NVIS) communications can be highlighted. An important judgment is worth noting here. Acoustic signals propagating in water tend to experience greater losses in the channel than radio signals propagating in air or in a vacuum. For the same link length and data rate, a relatively higher transmission power is usually required for an acoustic signal than would be required for a radio signal. As a result, energy preservation is more critical in under water acoustic sensor networks than in traditional wireless sensor networks [76,77].

3.2. MIoT Architecture Based on a Data Transfer Scenario

One approach to the architecture definition is based on the description of a cooperative data transmission scenario between deployed sensor nodes (Figure 5), where one sensor node sends its data packets to its neighboring nodes [9,30]. In this case, only some nodes decide to process the received packets according to predefined criteria, such as the quality of the transmission channel or depending on the available resources of the retransmission node. In addition, any cooperating retransmission node may also simply amplify and then retransmit or decode the received signal according to a specified metric.

Figure 3 shows a typical MIoT network architecture in which multiple sensor nodes are randomly located, exist at depth, and are separated from each other by Euclidean distances in a certain area to perform their tasks. Nodes may be stationary, semi-autonomous, or self-contained mobile nodes, and connect to the MIoT cloud using a surface gateway. The surface gateway is often equipped with both acoustic modems and radio modems, where the acoustic channel is used for submarine communications between submarine nodes and the RF channel is used for surface and/or satellite communications. For communication between the source node (S) and the destination node (D), repeater nodes (R) are scattered over -buses which contains repeaters, and each bus is considered as one node in the message-forwarding path, assuming that transmission is performed on communication phases on each route.

3.3. Functionality-Based MIoT Architecture

In terms of functionality, MIoT architecture can be divided into three layers [4], as presented in Figure 6.

The main features of these layers can be described as follows. The perception layer identifies objects and collects information. It consists mainly of underwater sensors, submersibles, ground stations, monitoring stations, data storage tags, acoustic/radio/PIT tags and hydrophones/receivers/tag readers. The network layer consists of a converged network consisting of wired/wireless private networks, the Internet, network administration systems, cloud computing platforms, etc. It processes and transmits information received at the perception level. The application layer is a set of intelligent solutions that apply MIoT technology to satisfy user needs.

The perception layer considers ROV and underwater micro- and nanosensors, acoustic tags, radio tags, and PIT tags. ROVs must be able to estimate their position and avoid collisions. To facilitate navigation, the ROVs are equipped with hydroacoustic sensors. Underwater sensors and ROVs (Figure 4) communicate with each other for specific applications, such as oceanographic data collection, cooperative monitoring, or observation. The resulting data are sent to a station (receiver) that floats on the surface of the water and uses, for example, radio communications to send this data to shore over long distances. The shore station is placed in the monitoring center and performs further data analysis. An acoustic tag is a small device that emits sound. It periodically transmits data about the specific speed of a particular device to an array of hydrophones (acoustic sensors) or ROV. RFIDs transmit a radio frequency signal that can be detected over water. Radio tags, unlike acoustic tags, become less effective in highly conductive water. A PIT tag system consists of a tag, an antenna, and a transceiver. Each PIT tag contains an individual alphanumeric code for identification. Because PIT tags are passive, they remain active for long periods of time.

The network layer provides the surface station access to the radio channel to process and transmit the necessary information received from the perception layer. This information is relayed to the remote command center using various network access technologies such as satellite communications, general packet radio services (GPRSs) or wideband code-division multiple access (WDMA).

Note that not all underwater sensor nodes support IP. Therefore, new protocols are needed to enable the interaction between heterogeneous sensors, as well as efficient gateways to enable communication between different transmission media (underwater acoustics and radios/satellites) [78].

3.4. Common MIoT-Based System Architectures

Similar to IoT, it is assumed that a typical IoT-based marine environmental monitoring system may have five levels [40]: the perception and execution level, the transmission level, the preprocessing level, the application level, and the business level. This assumption is based on the fact that the IoT is able to perceive, control, and analyze the surrounding world by collecting, processing, and analyzing data [5].

The perception and execution layer represents the architecture’s bottom layer. It includes sensors and actuators designed to collect data and execute commands. In IoT-based marine environmental monitoring and protection systems, this layer may also include GPS sensors and energy-harvesting devices.

The main purpose of the data transfer layer is to transfer various collected information to the data processing layer via the UWC.

Control measures carried out by users or intelligent applications are moved from the application level to the perception and execution level, so that the relevant devices or actuators can perform the required actions.

The pre-processing layer is in the middle of the system architecture, where the raw data received can be stored and pre-processed using intelligent data analysis. It also performs data validation and sometimes initiates alerts or warnings based on predefined rules.

The application layer provides services according to the different applications requested by users.

The business layer is the top layer and manages all system activities and services, including the creation of business models according to the data received from the application layer. It also monitors and validates outputs from the other four layers according to business models to improve service quality and ensure user privacy [79,80].

In [28], the machine-type communication (MTC) architecture [71] is proposed, which consists of three functional objects: a network controller, a marine application server, and a control station, as shown in Figure 7, where the mobile station is a mobile transceiver capable of mobile communication with the network through the control station. In this structure, the mobile station provides the interaction between the MTS marine network and the marine device or local network. It is also capable of interfacing directly with other mobile stations to form a specific local network for maritime proximity services. In fact, this architecture ideologically coincides with the above [4,81].

The MIoT architecture, based on the environment of finding different devices and interaction objects [82], is shown in Figure 8.

The proposed architecture is a three-tiered network system including: a large number of underwater sensors, acoustic positioning systems, autonomous underwater vehicles (AUVs); unmanned surface vehicles, surface buoys and drones; and local radio channels and cellular or satellite links.

The architecture of underwater wireless sensor networks based on 1D–4D (UWSN) [83] is shown in Figure 1.

In the one-dimensional (1D) UWSN architecture, sensor nodes are deployed autonomously. Individual sensor nodes represent an autonomous network responsible for detecting, processing, and transmitting information to remote stations. A typical example of a node in this architecture would be a floating buoy detecting underwater features, or one that could be deployed underwater for a period of time to pick up information and then float to the surface to transmit information to its remote station. It may also be an autonomous underwater vehicle (AUV) located underwater to detect or collect necessary underwater data and transmit the information to a remote station. In a 1D architecture, the communicating nodes may use UWSN acoustic, RF, or optical communications. Moreover, the topological nature of the 1D UWSN is star-shaped, where transmissions through the sensor node and the remote station are made through a single transition.

Two-dimensional (2D) FPSSs use deep-sea anchors to hold (attach) sensor nodes. The anchored underwater nodes use acoustic links to communicate with each other or with underwater nodes. Underwater nodes are responsible for collecting data from deep-water sensors and transmitting these data to remote command stations via ground stations. Underwater nodes are connected to horizontal and vertical acoustic transceivers. Horizontal transceivers are designed to communicate with a sensor node to collect data or send commands to nodes, as is performed by marine control stations. The vertical transceiver is designed to transmit data into the command post. A vertical transceiver must have a long range because the ocean depth can be as deep as 10 km. In addition, surface nodes equipped with acoustic transceivers can manage parallel communications using multiple organized submarine nodes. Surface nodes are also equipped with long-range radio frequency transmitters to communicate with marine nodes.

In a three-dimensional (3D) architecture, sensor nodes float at different depths, tracking specific activity in three dimensions. Traditionally, underwater three-dimensional sensor networks use surface buoys to simplify the deployment of these networks. However, there is vulnerability to weathering and disturbance. The three-dimensional architecture uses the ocean floor to anchor the sensor nodes. Wires attached to anchors control the depth of the sensor nodes.

Four-dimensional (4D) UWSN includes fixed networks, such as the mobile UWSN in 3D networks, consisting of remote operational underwater vehicles (ROVs) to collect data from anchor nodes and transmit it to remote stations. ROVs may be autonomous submersible robots, vehicles, ships, and even submarines. Individual sensor nodes may autonomously retransmit collected data directly to the ROV depending on proximity. Hydroacoustic communication, optical communication, or radio-frequency communication are used, depending on the distance and data transmitted between the nodes. Since the transmission is directly to the ROV, sensors that have a large amount of data and are near the ROV can use radio communications, while sensors that have small data to transmit or are far away from the ROV use acoustic communications [83].

To summarize the described MIoT architecture solutions, it is possible to consolidate the analysis results into

Table 5. Comparison table of three MIoT system platforms.

Architecture	Layers	Principle	Topology
MIoT network architecture	5 separate layers: application, transport, network, data link and physical layer	TCP/IP	Mesh
MIoT architecture based on a data transfer scenario	3 layers: surface gateway, ROVs, source/destination/repeater nodes	Data transfer	Mesh
Functionality-based MIoT architecture	3 layers: perception layer, network layer, application layer	Functionality	Mesh
Common MIoT-based system architectures	5 levels [40]: perception and execution level, transmission level, preprocessing level, application level and business level	Machine-type communication	Mesh
The MIoT architecture, based on the environment of finding different devices and interaction objects	3 layers: underwater sensor network, surface AD HOC network, Internet	The environment of finding different devices and interaction objects	Mesh
1D–4D	4 layers: 1D, 2D, 3D, 4D	Functionality, TCP/IP	Mesh, 1D UWSN is star-shaped

. As a general characteristic of each architecture, it is worth noting the fact that an important decision in the development of MIoT platforms is the choice of information exchange protocols. New reliable algorithms for transmitting information are currently being developed. However, this issue is more specific to each structure of the chosen system [78,83]. Moreover, each platform has its own set of tasks and most often they are associated with the data arrays’ accumulation, which may have different dimensions, resulting in a number of other problems associated with Big Marine Data (BMD) applications.

4. Discussion

Considering a variety of signal transmission means underwater and under different conditions of marine robot application and their cooperation, it seems reasonable to equip marine robots with a complex system of underwater–air communication. In other words, wireless communication system should be implemented, using several receivers and several transmitters. The system should be able to switch the appropriate interface. Such a system is a system with multiple inputs and multiple outputs (MIMO). A MIMO system of wireless communication between underwater robotic agents, underwater sensors and the surface platform for the transmission of real-time data, including video information, to the ground servers for its further processing, including the development of intelligent control, can be used for transmission with one or more carriers [84].

The 1D–4D (UWSN) architecture, as mentioned above, is defined as a group of sensor nodes that are installed underwater for the wireless exchange of information with the base station and underwater robotic agents (AUVs, nodes, etc.). Given the proposed MIMO approach, it is assumed that AUVs and communication (retransmission) buoys have transmitting and receiving hardware typical of acoustic, optical, and radio frequency for data transmission in various applications with the maximum possible speed for each specific situation. In particular, the data include video information with the necessary resolution.

Therefore, when certain information needs to be transmitted, the underwater robot transmits a radio-frequency broadcast request, including information about the type of data to be transmitted. Communication buoys that have received the request transmit information to the underwater robot about their readiness to receive information and their coordinates. The underwater robot performs a certain maneuver to approach the nearest communication buoy. By turning on the appropriate data interface, for example, optical for video picture transmission, the underwater robot performs the transfer. If the transfer for some reason becomes impossible, for example, due to an optical signal in turbid water, then the decision is made either to replace the interface, if it is acceptable for the transmitted information, or to search for a new communication buoy. A similar algorithm for data exchange can also be considered using the architecture based on functionality.

Note that, in addition to data transmission, acoustic and optical transponders are used for marine robots’ navigation.

The next step, as mentioned above, is the transfer of information from the communication buoy to the cloud, either directly or through a translational surface platform. Additionally, in some cases, there may be a need to perform an intermediate data transfer between communication buoys. Since the transmission takes place in airspace, RFC is already used here.

In this case, the Zigbee protocol used in the IoT technology can be considered as an alternative. Zigbee is a specification of network protocols of the upper layer (the application support sublayer and network layer (NWK)) using the services of the lower layers (MAC environment access control layer and physical layer (PHY)), regulated by the standard IEEE 802.15.4. The Zigbee specification is focused on applications that require guaranteed secure data transmission and the long-term operation of network devices on autonomous power sources (batteries). The Zigbee specification includes the ability to select a routing algorithm depending on application requirements and network status, application standardization mechanism application profiles, the library of standard clusters, endpoints, bindings, and flexible security mechanism.

For the future of underwater communication, we note 5G technology because it has a high data transfer rate, extremely low latency, and improved quality of service. The use of 5G supporting acoustic, optical, and radio-frequency signals will significantly increase the probability of communication [85]. Note, the implementation of 5G in the area of optical communication will lead to a better bandwidth and higher data transfer rate.

Due to the expansion of marine environmental monitoring and protection, international cooperation projects are emerging one after another. According to the projects’ interests, different application platforms and methods have their own characteristics and cannot be compatible with each other. Although a number of network protocols and standards for IoT have been proposed and developed, they are not sufficient for marine-monitoring applications. In addition to IoT network standards, it is also important to provide the industry with standards for MIoT devices, equipment and platforms for marine-monitoring and protection applications, and to provide marine management agencies with standards for marine data management, analysis, and reporting. As a result, the standardization of platform design for marine-monitoring and protection systems causes problems, including:

• The standardization of MIoT devices specifically designed to monitor and protect the marine environment, including sensors and actuators, routers, and gateways;
• The standardization of MIoT platforms and system technologies for monitoring and protecting the marine environment, including communication network structures, protocols, and algorithms;
• The standardization of computing and data storage technologies for monitoring and protecting the marine environment, including cloud, fog and edge computing mechanisms, data archiving, and storing methods;
• The standardization of data analysis results and reporting formats for sharing between different organizations and governments.

In the experiments announced in the Section 1, the communication-line operator-receiving station-communicating buoy AUV exchanged data packets containing information about the location of the search object, its distance, and the required speed of the vehicle.

Figure 9 shows pictures of the underwater equipment, including the buoy, the hydroacoustic modem, the AUV, and its trajectory (red track).

The analysis of the results obtained during the demonstration tests showed the consistency of the developed MIoT architecture platform to ensure the interaction and digital navigation of marine robotic agents and the technical possibility of intermediate information transmission over the line to provide remote control of the ROV under water by a remote operator.

In addition to the above, there are a number of other experimental projects [86,87,88] related to the MIoT. The common goals are object detection and providing cooperative control of the AUV in information exchange via various underwater communication means.

5. Conclusions

The main tasks of the MIoT platform can be summarized as the collection, sharing and interim analysis of data, as well as the possible navigation and control of marine robots and their actuator triggering. Despite numerous research and development efforts, the data exchange and message dissemination underlying the MIoT remains a complex process. For example, there are different network architectures, there are no universal and simple user interfaces, and data handlers use a variety of data types. Such problems complicate MIoT communication solutions.

Analyzing the reviewed MIoT architectures, it can be concluded that the most successful in terms of the generality of representation is the combined architecture based on the fusion of the 1D–4D functionality and network model, which generalizes both the functionality of the MIoT and the ability to impose a network representation on the specified architecture (because the network architecture can be defined depending on each specific implementation).

Considering a variety of underwater signal transmission means and different conditions of marine robot application and their cooperation, it seems reasonable to equip robots with a complex system of underwater communication. In other words, an underwater wireless communication system with multiple receivers and multiple transmitters must be implemented. The system must be able to switch the respective interface. Such a system is a system with few inputs and several outputs which can be used for both single-carrier and multiple-carrier transmissions.