Underwater Wireless Sensor Networks: Current Research and Challenges

Authors: Wang Jing, Chen Jianfeng, Zhang Lijie, Huang Jianguo, Source: Streams of Ocean Life

1. Introduction

Underwater wireless sensor networks (UWSNs) utilize aircraft, submarines, or surface vessels to randomly deploy a large number of inexpensive miniature sensor nodes (ranging from hundreds to thousands) into the water of interest. These nodes form a multi-hop, self-organizing network system through underwater acoustic wireless communication, collaboratively sensing, collecting, and processing information about the objects of interest within the network coverage area, and transmitting it to receivers. In recent years, UWSN technology has gained widespread attention both domestically and internationally, being extensively used for ocean data collection, pollution prediction, offshore mining, disaster avoidance, and ocean monitoring.

UWSNs possess advantages that traditional sensor technologies cannot match: the sensor network is composed of dense, low-cost, randomly distributed nodes. Its self-organization and fault tolerance ensure that the failure of certain nodes due to malicious attacks does not lead to a collapse of the entire system; the multi-angle and multi-directional information fusion from distributed nodes can enhance data collection efficiency and yield more accurate information; sensor nodes operating in close proximity to the target improve the signal-to-noise ratio of received signals, thereby enhancing the system’s detection performance; the mixed application of various sensors within nodes allows for a more comprehensive reflection of the target’s characteristics, beneficial for improving system positioning and tracking performance; the sensor network expands the spatial and temporal coverage capabilities of the system; and the ability of individual mobile nodes to adjust the network’s topology can effectively eliminate shadows and blind spots within the detection area. Therefore, sensor networks can be applied in harsh battlefield environments. In the military domain, through the close coordination of multi-sensor systems, a multi-sensor complementary surveillance network composed of air-ship-land-based sensors can capture, track, and identify targets.

Due to the unique application environments of UWSNs, factors such as seawater salinity, pressure, ocean currents, marine life, and acoustic wave attenuation must be considered, making the design of UWSNs more challenging than that of terrestrial wireless sensor networks, with higher hardware requirements. A comparative overview of the various modules of UWSNs and terrestrial wireless sensor networks is shown in Table 1.

Table 1: Comparison of Underwater and Terrestrial Wireless Sensor Networks

2. Current Research Status of Underwater Wireless Sensor Networks

Due to the immense application value of UWSNs, they have attracted significant attention from military departments in many countries around the world. The development of underwater sensor network technology is even influencing changes in naval military strategies. With the advancement of underwater sensor network technology, future naval battles can fully leverage coastal spatial advantages.

The United States was the first country to conduct research on UWSNs, investing heavily in the 1950s to build a large underwater acoustic monitoring system (SOSUS) in the Atlantic and Pacific Oceans. In recent years, significant UWSN projects in the United States include: the SeaWeb program by the U.S. Navy Research Office from 1999 to 2004; the CodeBlue platform research program initiated by Harvard University in 2004; and the PLUSNet program, which was disclosed at the 2006 Submarine Technology Forum held by the U.S. Navy Submarine League, expected to achieve full operational capability around 2015; Lockheed Martin’s advanced deployment system (ADS) developed for the U.S. Navy, designed to adapt to nearshore environments and to be rapidly deployed for detecting underwater enemy submarines.

As UWSN technology matures, it is gradually shifting towards civilian applications. In the civilian sector, it is mainly used for monitoring marine environments and marine biology, as well as underwater scientific experiments. Major civilian engineering projects for UWSNs include the marine biochemistry monitoring system (LOBO) and the marine monitoring system (MOOS) established by MBARI, the NEPTUNE project laying a cable monitoring system in the North Pacific, and the Frontiers Analysis Observation Network and Telemetry (FRONT) system located off the southern coast of Long Island, New York.

In 2006, the National Natural Science Foundation of China listed key technologies for underwater mobile sensor networks as a major research direction. Several universities and research institutions, including the University of Science and Technology of China, Shenyang Institute of Automation, and the Institute of Computing Technology at the Chinese Academy of Sciences, have begun research in the field of wireless sensor networks. In 2007, the project “Key Technologies of Wireless Sensor Networks and Their Application Demonstration in Road Traffic,” jointly undertaken by the Shanghai Institute of Microsystem and Information Technology of the Chinese Academy of Sciences and other institutions, passed acceptance.

With the development of UWSN technology, the concept of nodes has expanded from traditional sensors to include new node forms such as AUVs, divers, vessels, submarines, torpedoes, mines, and surface buoys. The new UWSNs overcome the shortcomings of traditional underwater sensor networks, which include single performance, poor timeliness, high costs, and deployment difficulties, featuring more flexible deployment, stronger adaptability to environments, and faster information collection.

The new generation of underwater networks is not just an acoustic network but a comprehensive physical network that includes acoustic, magnetic, and electrostatic fields. It integrates reconnaissance, vigilance, command, communication, navigation, positioning, target attack, and comprehensive combat capabilities.

3. Basic Structure and Characteristics of Underwater Sensor Networks

Sensor network systems typically comprise sensor nodes, aggregation nodes (receivers), and management nodes (servers). The data monitored by sensor nodes is transmitted multi-hop along sensor backbone nodes (cluster heads). During transmission, monitored data may be processed by multiple nodes, routed to aggregation nodes after several hops, and finally reach management nodes via the Internet or satellites. Users can configure and manage the sensor network through management nodes, publish monitoring tasks, and collect monitoring data. The schematic architecture is illustrated in Figure 1.

Figure 1: Structure of Wireless Sensor Networks

Underwater sensor nodes primarily consist of a main controller or CPU. This controller connects to the sensors via interface circuits. The controller receives data from the sensors, stores it in memory, processes the data, and sends it to other network nodes via an acoustic modem. Its internal structure is shown in Figure 2.

Figure 2: Internal Structure of an Underwater Sensor Node

The topology of underwater sensor networks is an open research area. Currently, studies on underwater wireless sensor networks include two-dimensional, three-dimensional static networks, and three-dimensional networks with AUVs. Figures 3 and 4 illustrate the schematic diagrams of two-dimensional static networks and seabed-based three-dimensional static networks, respectively.

Figure 3: Architecture of Two-Dimensional Static Networks

Figure 4: Architecture of Seabed-Based Three-Dimensional Static Networks

In both structures, sensor nodes are arranged on the seabed. In the two-dimensional static network, the sensor network nodes are fixed to the seabed, self-organizing into clusters, and the information collected by the nodes is transmitted directly or via multiple hops to the cluster head, which then relays it to surface relay stations or ship-based receiving stations, and finally communicates with shore-based receiving stations to reach the data processing center. The nodes in the three-dimensional static network are anchored to the seabed via anchor chains, forming a three-dimensional network distributed on the seabed by adjusting the length of the anchor chains. Each sensor node must be able to relay the collected information to the surface aggregation node, thus requiring at least one link to the surface relay station, which allows for better underwater sampling than the two-dimensional network. Both structures feature non-intrusiveness to navigation and are difficult to detect, but they pose challenges for deployment and maintenance.

Figure 5 illustrates a three-dimensional static network based on surface buoys. In this underwater wireless sensor network, each sensor node is equipped with a surface buoy. Unlike the buoys in the three-dimensional static network in Figure 4, the initial sensor nodes are positioned within floating buoys on the water’s surface. After deployment, the sensor nodes are placed underwater at predetermined depths, forming a three-dimensional distribution of sensor nodes. Each sensor node consists of a group of sensors, which are located at the same depth underwater. The sensors communicate with the buoy via a tether, while sensor nodes communicate wirelessly with each other, and data-collecting buoys gather data from the sensor nodes. The characteristics of this structure include ease of deployment and maintenance, low cost, high visibility, and sensitivity to external weather changes, with sensor nodes prone to movement.

Figure 5: Architecture of Three-Dimensional Static Networks Based on Surface Buoys

The three-dimensional underwater communication network with AUVs extends the functionality of static networks. AUVs can operate without cables or remote control, making them widely applicable in oceanography, environmental monitoring, and underwater resource development. The use of AUVs can enhance the capabilities of underwater acoustic communication networks. This network adapts through commands to AUVs at certain nodes within the network for adaptive sampling and self-configuration.

One of the goals in designing AUVs is to enable them to rely on their own intelligence with minimal reliance on online control from shore-based control centers. The NEPTUNE cable ocean monitoring system in the northeast Pacific employs AUVs to extend the wireless sensor networks in the ocean and on the seabed. Figure 6 provides a schematic representation of its basic elements.

Figure 6: Basic Elements of the Cable Underwater Monitoring System

4. Challenges Faced in Research

1. Characteristics of Underwater Acoustic Propagation

Terrestrial sensor networks primarily rely on radio frequency (RF) signals for information transmission, whereas underwater, high-frequency RF signals experience significant attenuation. In fact, the frequency of radio waves used underwater is only in the low-frequency band (30Hz to 300Hz), requiring relatively high transmission power. Light signals do not experience as high attenuation as RF signals, but they scatter significantly. Therefore, underwater wireless sensor networks utilize acoustic signals for wireless communication.

The limitations of underwater acoustic channels, including bandwidth and various delays, necessitate the development of new reliable communication protocols. The main challenges in designing underwater acoustic wireless networks include: 1) the achievable bandwidth of acoustic signals is strictly limited; 2) underwater channels suffer severe fading due to multipath propagation, and the attenuation underwater is variable; 3) the mobility of sensor nodes and energy depletion leading to node ‘death’ can result in unreliable communication connections, high bit error rates, and temporary communication interruptions; 4) the small communication range of nodes prevents real-time data processing; 5) solar energy cannot be utilized underwater, and battery energy is limited and non-replaceable; 6) the real-time and energy issues of underwater wireless sensor networks are contradictory, requiring software redesign to accommodate changes in information processing structures; 7) high density of underwater sensor nodes can lead to congestion and errors.

Based on the characteristics of underwater acoustic signal propagation, the main issues in the design of underwater wireless sensor networks are summarized.

2. Issues in Underwater Wireless Sensor Network Design

1) Sensor Node Localization Issues

Sensor nodes in underwater networks are randomly deployed. Due to the large number of nodes, their small size, and cost considerations, it is impractical to equip each node with a GPS localization system. Moreover, after deployment, the location of each node is not fixed and may drift with ocean currents, posing challenges for self-localization of the nodes.

Localization is fundamental for military applications. The localization problem includes two parts: self-localization of nodes and localization of external targets, with the former being the basis for the latter.

For self-localization based on reference nodes, many scholars have proposed various localization methods aimed at enabling sensor nodes to be randomly deployed within a given area and automatically detect their positions based on some reference nodes.

In wireless sensor networks, nodes that require localization are referred to as unknown nodes, while nodes with known positions through manual deployment or GPS systems are termed anchor nodes. Localization of nodes includes: absolute localization and relative localization, centralized localization and distributed localization, range-based localization and range-free localization. Absolute localization yields a standard coordinate position, such as latitude and longitude. Relative localization generally establishes a relative coordinate system for the entire network based on some nodes within the network. Centralized localization involves transmitting information to a central node for localization calculations, facilitating a global perspective for obtaining relatively accurate position estimates. Distributed localization aims to distribute the computational load among the nodes in the network to reduce demands on the computational capacity of the central node.

The vast majority of existing localization methods in wireless sensor networks consist of two fundamental steps: distance (or angle) measurement and localization calculation. Current localization methods include: RSSI-based localization algorithms, TOA-based localization algorithms, TDOA-based localization algorithms, and AOA-based localization techniques. Methods for calculating node positions include triangulation, trilateration, multilateration, centroid localization algorithms, multi-hop localization algorithms (APS, APIT), maximum likelihood estimation, Hop-Euclidean localization algorithms, and multi-dimensional localization algorithms (MSD). In underwater wireless sensor networks, due to the underwater placement of sensors, network nodes often lack definitive location information, necessitating a distributed localization mechanism for anchor-free WSN nodes.

Given that many mobile sensor nodes, such as AUVs and mobile robots, may exist in underwater WSNs, numerous localization methods for mobile nodes have emerged.

The performance of WSN localization algorithms directly impacts their usability. The evaluation criteria for localization algorithms primarily include: localization accuracy, scalability, anchor node density, node density, coverage, fault tolerance, adaptability, power consumption, and cost.

2) System Energy Issues

Due to the application characteristics of underwater wireless sensor networks, replacing batteries of underwater nodes is very challenging, making energy consumption a focal point for researchers. The primary energy consumption for sensor network nodes occurs during information transmission and data processing, with information transmission consuming more energy than data processing. Employing good topology control algorithms, efficient communication protocols, and routing protocols in software, as well as rational design of internal structures, use of low-power chips, and improved battery performance in hardware, are key to conserving energy.

Currently, many topology algorithms have been developed to effectively compute energy in real-time applications for sensors. Examples include the Least-Bit Energy Routing algorithm (LBMER), HCAN, CDS (Connected Dominating Set), JSD, and Ant Colony algorithms. All these algorithms are based on energy conservation principles.

In wireless sensor networks, medium access control (MAC) protocols dictate the usage of wireless channels, allocating limited wireless communication resources among sensor nodes to construct the foundational infrastructure of the sensor network system. Based on the sequential reduction of energy consumption in the four states of wireless communication modules (sending, receiving, listening, and sleeping), MAC protocols in sensor networks typically adopt an alternating strategy of ‘listening/sleeping’ to reduce energy consumption. Employing simple, efficient, and energy-saving MAC protocols is a critical consideration in the design of wireless sensor networks.

Unlike traditional network routing protocols, routing protocols in wireless sensor networks prioritize energy efficiency. The design goal of routing protocols is to minimize energy consumption, enhance the effectiveness of energy usage, and avoid excessive use of low-energy nodes, thus extending the network’s lifespan. Additionally, routing protocols should not occupy large amounts of storage space and should minimize computational load.

In hardware, due to the limited transmission capabilities and distances of sensor nodes, a large number of nodes often need to be deployed. Consequently, inexpensive and low-power sensor nodes are continually being developed. New experimental platforms designed to reduce energy consumption are being researched, such as Mica2 and MicaDot2, as well as the Sensor Web developed by NASA JPL. Mica2 features a new microprocessor and RF chip, correcting some technical flaws of Mica. The fourth generation of the Mote series, Telos, has seen significant improvements in design by selecting new microprocessors to reduce sleep current and system wake-up time, leading to substantial energy reductions. Many domestic research institutions have also developed new nodes according to their application needs. Advances in battery technology have gradually alleviated the energy bottleneck in sensor networks. The University of California, Los Angeles, the University of Florida, the University of Utah, and the Naval Research Laboratory have developed three-dimensional microbatteries using micro-electro-mechanical systems (MEMS) technology, which are expected to break through this limitation. Utilizing wind and solar energy in underwater wireless sensor networks is also feasible; MBARI has designed a system to collect wind and solar energy through surface buoys and transmit it to each network node via underwater cables.

3) Target Localization and Tracking

Acoustic signal target localization methods can be broadly categorized into several types based on physical processing: AOA, TDOA, DOA, and RSSI (or SRP). AOA-based localization not only determines the coordinates of targets but also provides angle information; however, AOA ranging techniques are susceptible to environmental influences and require additional hardware, making them unsuitable for large-scale sensor networks in terms of hardware size and power consumption. Currently, the most commonly used acoustic source localization methods are TDOA, DOA, and RSSI. A comparative summary of these three localization methods is presented in Table 2.

Table 2: Summary of Three Localization Methods

Tracking underwater targets often involves data fusion, which integrates information from multiple sensors or sources to arrive at more accurate and reliable conclusions. Target tracking includes several stages: detection, classification, localization, prediction, and tracking. Detection requires establishing appropriate thresholds; exceeding the threshold indicates the presence of a target, while falling below it indicates no target; classification determines the presence or absence of targets and the number of targets; the most commonly used method for localization is the maximum likelihood (ML) localization algorithm, which can yield good localization results but is sensitive to parameter variations and has high complexity in multi-target tracking environments; prediction involves establishing motion equations based on collected data to forecast the target’s trajectory.

4. Others

1) Security Issues of Underwater Wireless Sensor NetworksIn military applications, the security of sensor networks cannot be overlooked, as there are high demands for data sampling transmission and physical node distribution. The inherent characteristics of sensor networks mean that the approaches and methods for addressing security issues differ from those of traditional networks. Key differences include: limited storage space and computational capacity; lack of prior knowledge regarding post-deployment node distribution; physical security of deployment areas cannot be guaranteed; limited bandwidth and communication energy; and the security issues of the entire network being application-dependent. The security of sensor networks is an open and active research field.

2) Real-Time Performance of NetworksDue to their military application in target tracking, sensor networks must be able to reflect target motion information in real-time.

3) Optimal Number of Sensor Nodes in Fixed AreasIdentifying different types and positions of sensors to maximize coverage while minimizing costs is an NP problem. One approach is to find a coverage path that maximizes the sum of distances to each node while minimizing the distance to the nearest sensor node.

5. Conclusion

Underwater wireless sensor networks represent a large system with numerous technical issues related to hardware and software. This article summarizes existing research on underwater sensor networks, focusing on self-localization of sensor network nodes, energy issues of the system, and target localization methods based on acoustic signals, explaining several important factors that affect sensor network performance. Future work should consider all factors significantly impacting sensor network performance, seeking a balance between computational load and localization accuracy to achieve optimal localization tracking methods for the best positioning results.

Related posts

Leave a Comment Cancel reply