Source-linked AI summary
CONet: A Cognitive Ocean Network
Huimin Lu, Dong Wang, Yujie Li, Jianru Li, Xin Li, Hyoungseop Kim, Seiichi Serikawa, Iztok Humar
TL;DR
Ocean networks face limited research coverage and difficult communication conditions, including low bandwidth and transmission losses. The paper defines a Cognitive Ocean Network, proposes its layered architecture and applications, and identifies future research challenges. It positions CONet as an AI-integrated framework for intelligent ocean observation while acknowledging that many open problems remain.
Problem
Few studies address the Ocean of Things, while underwater communication has low bandwidth, high transmission losses, time-varying multipath propagation, and high latency.
Method
The paper defines CONet, proposes a four-layer cognitive ocean architecture, and describes applications including sea-bottom discovery and disaster prevention.
Results
The paper presents CONet as an AI-integrated framework for intelligent ocean-observing systems and reports offshore sensors detecting a 5-meter-high tsunami approximately 10 minutes before arrival.
Takeaways & Limitations
CONet provides a framework for applying artificial intelligence to ocean observation and for organizing demonstration applications in ocean science and engineering.
Takeaways & Limitations
Many CONet challenges remain open problems for future investigation.
Abstract
from arXiv · showhide
The scientific and technological revolution of the Internet of Things has begun in the area of oceanography. Historically, humans have observed the ocean from an external viewpoint in order to study it. In recent years, however, changes have occurred in the ocean, and laboratories have been built on the seafloor. Approximately 70.8% of the Earth's surface is covered by oceans and rivers. The Ocean of Things is expected to be important for disaster prevention, ocean-resource exploration, and underwater environmental monitoring. Unlike traditional wireless sensor networks, the Ocean Network has its own unique features, such as low reliability and narrow bandwidth. These features will be great challenges for the Ocean Network. Furthermore, the integration of the Ocean Network with artificial intelligence has become a topic of increasing interest for oceanology researchers. The Cognitive Ocean Network (CONet) will become the mainstream of future ocean science and engineering developments. In this article, we define the CONet. The contributions of the paper are as follows: (1) a CONet architecture is proposed and described in detail; (2) important and useful demonstration applications of the CONet are proposed; and (3) future trends in CONet research are presented.
1. Introduction
Ocean-observation networks have expanded from national and regional installations to large-scale subsea infrastructures, supporting long-term measurements and monitoring. These systems provide data for earthquakes, tsunamis, severe weather, environmental change, and other oceanographic research.
- Few studies have examined the Ocean of Things compared with traditional territorial wireless sensor networks.
- NEPTUNE uses an 800-km submarine fiber cable and operates across an underwater observation range of 17–2660 m.
- EMSODEV develops the EGIM instrument module to support accurate, consistent, and comparable long-term oceanographic measurements.The measurements target challenges including climate change, marine ecosystem disturbances, and marine disasters.
- DONET supports earthquake and tsunami early warning with 5 nodes and 20 observation points concentrated south of the Kii Peninsula.Observation points are spaced 15–20 km apart.
- IORS combines 33 types of instruments for ocean, atmosphere, weather, safety, and environmental observations that provide information for global change.
2. Issues of Recent Underwater Networks
Underwater networks connect smart objects and sensors for monitoring, tracking, localization, and energy supply, but they operate under severe communication, power, and environmental constraints. Recent work therefore combines acoustic, optical, cable-based, software-defined, and AI-related approaches while recognizing unresolved practical limitations.
- Communication Technologies: Underwater wireless channels are affected by noise, limited bandwidth, scarce power resources, and harsh environmental conditions.
- Communication Technologies: Acoustic communication supports long-distance transmission but has low bandwidth, high transmission losses, time-varying multipath propagation, and high latency.Available acoustic technology can transfer data up to tens of kbps over distances from 10 m to 10 km.
- Communication Technologies: Software-defined acoustic modems use SDR capabilities to provide flexible and easily reconfigurable underwater IoT communication.
- Communication Technologies: Traditional acoustic communication cannot meet data-rate requirements of up to tens of Mbps for many underwater robots and sensors.Fiber optic or copper cables are used to achieve high data rates.
- Tracking Technologies: Existing underwater tracking methods work in some situations, but a more practical and applicable framework is needed for real-world use.
- Energy Harvesting Technologies: Ocean energy harvesting research emphasizes ocean-current power generation and salinity-gradient energy for supplying underwater networks.A floating ocean-current turbine system was reported as a low-cost mooring system and a possible main electricity source.
- Ocean Network Characteristics: Ocean networks are sparser and nonuniform because their underwater energy and communication infrastructure is costly and challenging.Underwater vehicles use acoustic communication and sensors to detect and track phenomena of interest.
- Localization Technologies: Ocean localization differs from land IoT localization because GPS or artificial arrangements cannot generally locate unknown underwater nodes.Algorithms are organized into distributed or centralized categories and into estimation-based or prediction-based approaches.
3. Cognitive Ocean Network (CONet)
CONet is organized as a four-layer cognitive architecture spanning sensing, local processing, cloud computing, and applications. The section describes edge anomaly detection, fog-based image enhancement, cloud transmission, and YOLO-based ocean monitoring applications.
- 3.1 Definition: The CONet architecture comprises Perception Sensor (Edge), Local Processing (Fog), Cloud-Computing, and Application layers, each using cognitive artificial intelligence.The layers sense or compute data across the system.
- 3.2.1 Perception Sensor (Edge) Layer: The edge layer integrates underwater sensors, AUVs, ROVs, buoys, ships, and ASVs for ocean observation.The listed platforms support sensing and vehicle operations in the perception layer.
- 3.2.1 Perception Sensor (Edge) Layer: CONet uses deep reinforcement learning and motor-temperature sensing to detect vehicle operating abnormalities and return underwater vehicles to the surface when temperatures exceed generated thresholds.DS18B20 sensors monitor motor status, while a Raspberry Pi processes the signal.
- 3.2.2 Local Processing (Fog) Layer: The fog layer preprocesses edge data, reducing cloud-computing burden while collecting, interacting with, analyzing, and computing information from multiple edge devices.Fog nodes operate at the local-zone-network level and can provide security for some information.
- 3.2.2 Local Processing (Fog) Layer: CONet proposes deep-learning image descattering followed by super-resolution processing, with a fusion scheme selecting constraints to produce clearer images.Captured underwater images are first descattered by deep neural networks before super-resolution processing.
- 3.2.3 Cloud-Computing Layer: Processed information moves from local and sensor layers to cloud computing, then reaches users and onshore monitoring centers through satellite, GPRS, or wideband CDMA communications.The architecture also draws on ship-based and vessel-based cloud-computing approaches for oceanographic research.
- 3.2.4 Application Layer: YOLO divides images into an S×S grid, predicts B bounding boxes and confidence scores, and retains boxes whose confidence P exceeds threshold T.The method supports rapid detection and tracking of multiple marine organisms and other ocean-related targets.
4. Applications
CONet applications span ocean exploration, resource detection, security, disaster prevention, environmental monitoring, and industrial oversight. The paper describes cognitive sensing and autonomous systems for these uses, including tsunami detection before arrival.
- CONet applications are classified into five categories spanning practical and potential ocean uses.The supplied passages introduce the classification but do not enumerate all five category names.
- Ocean Exploration: Ocean mining remains difficult because locating deep-sea mines and recognizing mine-like objects are challenging.Automation targets mine localization and recognition without human intervention through search-classify-map and reacquire-and-identify stages.
- Ocean Exploration: Search-classify-map rapidly surveys large areas, while reacquire-and-identify uses close-range magnetic, acoustic, or electro-optic sensing for final classification.The passages contrast rapid area search with close-range target reacquisition and identification.
- Security: CONet supports security applications including sea-bottom discovery, mineral-resource detection, and harbor monitoring.One sea-bottom discovery approach combines sensors and Particle Swarm Optimization to measure water depth and maximize surveillance coverage.
- Disaster Prevention: 150 seafloor sensor nodes connected by fiber-optic cable enabled offshore tsunami sensors to detect a 5-meter-high tsunami approximately 10 minutes before arrival.The network is described as an NIED seafloor network along the Japan Trench.
- Other Applications: Other applications include visitor education with PIT tags, fish-farm monitoring with water-quality sensors and cameras, and deep-sea pipeline oil-spill control.These uses are presented as useful applications for humanity.
5. Challenges and Future Trends
The paper frames CONet’s future around cognitive intelligence and five research challenges, including self-management, energy efficiency, communications coverage, and fog/edge and cloud computing. It also identifies general-purpose intelligence cognition technology and distributed cloud platforms as future directions.
- Current AI models are described as complicated, overly dependent on big data, and lacking a defined function.
- The paper introduces general-purpose intelligence cognition technology called the Cognitive Ocean Network.
- CONet research challenges include self-management, energy efficiency, communications coverage, fog/edge computing, and cloud computing.The passages specifically discuss autonomous malfunction detection, energy-saving devices, disruption-tolerant routing, and distributed computation.
- Communications Coverage: Sparse CONets require automatic routing that satisfies application requirements despite nonstable connectivity.
- Cloud Computing: Parallel and distributed cloud computing is identified as a future trend for handling large amounts of data.
6. Conclusions
The conclusion positions CONet as an AI-enabled framework for intelligent ocean observation distinct from underwater sensor networks and the Internet of Underwater Networks. It also emphasizes that substantial challenges and open problems remain.
- CONet is introduced as a framework combining Internet of Things, artificial intelligence, robotics, and blockchains for ocean observation.These technologies are described as core technologies of Industry 4.0.
- CONet differs from underwater sensor networks and the Internet of Underwater Networks in characteristics and definition.
- CONet aims to use artificial intelligence to build intelligent ocean-observing frameworks.
- The paper reviews CONet challenges and states that many open problems remain for future investigation.