Dual-Link Synchronous Acquisition and Transmission System for Cabled Seafloor Earthquake Observatory

Abstract

Seafloor observatories play a crucial role in acquiring continuous and precise submarine monitoring data, thereby holding significant implications for advancing major scientific advancements in marine science, particularly in the field of seafloor earthquake observation. This work mainly builds a dual-link observation system designed for observing seismic information on the seafloor based on a Zynq7000 system-on-chip and time synchronization module. The system is based on Zynq7000 SoC(MILIANKE; Changzhou, China) and AD7768(Analog Devices, Inc.; Norwood, MA, USA) to achieve eight-channel data (24 bit) synchronous acquisition, and the robustness of the system is improved by upgrading the link to full-duplex transmission and adding node data self-storage function. The P88 1588 PTP time synchronization single board(CoolShark; Beijing, China) is used to provide PPS (Pulse per second) signals for the system to realize microsecond timestamps to support subsequent seismic data inversion. An upper computer system based on the Qt framework is also developed to monitor the network condition in real time while visualizing the data transmission. For the acquisition of seismic signals, we employed triaxial seismic sensors. Additionally, a temperature and humidity monitoring module, along with an attitude detection module, was designed to enable real-time monitoring of the nodes. These modules not only facilitate the real-time monitoring of the nodes but also contribute to seismic data inversion. The experimental results indicate that the system provides a good synchronization of data acquisition, high accuracy, and reliability of inter-node transmission, which has good application prospects.

1. Introduction

Submarine earthquakes are intense manifestations of seafloor tectonic activities. In addition, activities such as submarine landslides, volcanic eruptions, and underwater explosions can also generate seismic signals similar to submarine earthquakes. Submarine earthquake monitoring is crucial for national marine disaster management, ensuring the safety of human lives and properties. It is also an important means for studying Earth system science. Around 85% of global natural earthquakes occur in the oceans, yet their monitoring has remained relatively weak. Rational development and utilization of marine resources can yield significant economic benefits [1,2]. However, frequent occurrences of submarine microseisms and the presence of marine geological hazards not only pose threats to offshore operations such as drilling and transportation but also have adverse effects on climate. To address challenges such as climate change and marine geological hazards, significant improvements have been made in marine sensing technologies such as submarine communication cables over the past few decades [3]. Monitoring the oceans can provide fundamental data for early warning of submarine earthquakes, assessment of marine engineering safety, the study of mechanisms related to underwater mineral resources, and monitoring of marine tectonic activities. In the context of gas emissions, considerable attention has been given to the investigation of radon gas (222Rn) and its daughter isotopes (214Bi, 214Pb) as potential seismic precursors [4,5,6], particularly in groundwater environments over the past few decades. Based on this study, Christos Tsabaris deployed gamma-ray monitoring underwater to continuously analyze the total gamma-ray intensity and investigate its relationship with earthquakes [7]. In addition, Claudio Martini conducted research on the groundwater in the Laiozi region based on the model proposed by Chiodini et al. Two significant carbon dioxide emission zones were identified through measurements, and the correlation between water quality changes and seismic activity was studied [8]. Continuous and accurate submarine observations are indispensable for the blue economy [9]. The key technological research on regional submarine microseism chain observation proposed in this project holds significant importance for the development of submarine earthquake monitoring.

Under the trend of long-term and continuous observations dominating marine research, submarine observatories are considered the third Earth science platform following remote sensing. Submarine observatories involve the deployment of sensors on the seafloor for data collection. They enable continuous, real-time bidirectional communication and power supply between the sensing system and onshore data centers [10]. Through this approach, continuous and multidisciplinary measurements of the water column can be conducted. Researchers worldwide are making unremitting efforts in this regard.

With the continuous efforts of scientific researchers all around the world, the undersea geo-observatory has become one of the critical infrastructures for monitoring the submarine environment. Some examples are the Monterey Accelerated Research System (MARS) from the US [11], NEPTUNE from Canada [12], DONET from Japan [13], EMSO from Europe [14], ZRS Island Experimental Research Observation (Z2ERO) from China [15], and East China Sea Seabed Observation Experiment Station from China [16,17,18]. Among them, East China Sea Seabed Observation Experiment Station from Tongji University implemented a series of sensor resource objects as object models to realize standardized undersea detection [16]. To be more specific, the client-server system was created to realize duplex data transmission with control commands. Researchers have proposed various solutions to meet gridded seafloor observation requirements. Bruce M. Howe from the University of Hawaii introduced JTF SMART Subsea Cables, combining cable and sensing technology. By integrating a "repeater" with sensing, digital signal processing, and transceiving functions, traditional cable-laying equipment can be utilized for monitoring essential ocean variables (EOVs) and providing disaster warning data without modifications [19]. The Canadian seafloor observing network relies on SIIM as its core technology, establishing VENUS on an offshore scale and NEPTUNE on a regional scale. The network includes expansion ports and employs the remotely operated platform for ocean sciences (ROPOS) underwater robot [12]. In Europe, Nadine Lantéri et al. developed the EMSO Generic Instrument Module (EGIM) as part of the European Multidisciplinary Seafloor and water column Observatory (EMSO). This distributed research infrastructure fosters collaboration, allowing for collaborative observation of variables and enhanced interoperability, complementing regional facilities [20]. The online seafloor seismometer developed by Shinohara et al. at the University of Tokyo uses conventional cable layouts with Ethernet switches and optical transceivers at the nodes for data transmission and reception, and the deepest node is equipped with an underwater mateable connector (UMC) that can supply power to pressure sensors via Ethernet [21].

In our previous work, we proposed a prototypical chained data acquisition and transmission system [22]. Through prototype validation, we successfully demonstrated the feasibility of its mechanisms, including clock synchronization and link transmission. However, the single-link structure exhibited instability. To achieve long-term and stable monitoring of seismic data, we have made improvements to the system. Firstly, the unidirectional links have been transformed into bidirectional links, with independent transmission on each link. Secondly, to address potential complete communication failures between nodes, an SD card-based real-time data storage system has been implemented to ensure data backup at the local nodes. Thirdly, the AD7768(Analog Devices, Inc.; Norwood, MA, USA) is utilized for multi-channel synchronized conversion and data acquisition, while triaxial seismic sensors are employed for seismic signal collection. The temperature and humidity monitoring module, along with the attitude detection module, has been designed for real-time monitoring of the nodes, enhancing the comprehensiveness of data collection and facilitating seismic data inversion in collaboration with the seismic sensors. Through these improvements, the challenges of “unobservable, unstable, and imprecise” submarine seismic signals have been successfully addressed. Additionally, an intelligent system management platform based on the Qt language has been developed to simplify system maintenance and management, enable data visualization, as well as facilitate link monitoring and fault feedback.

The remaining sections of this paper are structured as follows: The first section introduces the design for enhancing the reliability and robustness of SQATS. The second part presents the implemented hardware and software modules. The third part covers the deployment of the hardware system and the development of the management platform. Following that, the design implementation is demonstrated. Lastly, the experimental results and discussion are provided in the final section.

2. Design

2.1. Overall System Design

To address the issue of "unstable" submarine seismic observation data, we have made improvements in the hardware design of the system. We have implemented a dual-link connection between nodes to ensure stability in submarine seismic observation data. The establishment of the link transmissions is independent of each other. In other words, if one link’s data transmission channel fails to establish, the other link can continue to operate smoothly. We reorganized the transmitting-and-receiving timing logic design based on the Aurora IP core between each node to achieve full-duplex transmission between the nodes. At the same time, considering that dual-link transmissions may also fail, we also added a local storage function at each node. The SD card socket and proprietary communication line between Zynq7000 SoC(MILIANKE; Changzhou, China) and the card socket are installed based on the original hardware control circuit of each node. For the software design of duplex SQATS, the improved system also uses C Programming Language for embedded systems to realize the control of reading and writing the SD card on the PS side. After installing the 64 GB SD card, real-time storage of the integrated submarine monitoring data obtained from the designed node can be realized. According to the relevant software design, the node local storage system can store more than six months of data.

Apart from the improved hardware and software design, the upper computer system serving as a data transmission assistant is also upgraded. The upgraded version realizes an intelligent data link management system based on the Qt framework. Inheriting the major functions of the network port transmission assistant, the link management system can visualize real-time measurement data. The system also automatically detects the network transmission state. In emergency states, the backup link system is kicked in.

2.2. Components of Duplex SQATS

In summary, as shown in the following

Table 1. Device composition table of different components in the node.

Control and Transmission Module	Data Acquisition and Conversion Module	Clock Synchronization Module	Power Supply Module
Zynq7000 SoC Main Control Board (MILIANKE; Changzhou, China)	seismic sensor: MTSS-2003-OBS(R-Sensors LLC; Moscow, Russia); AD conversion: AD7768(Analog Devices, Inc.; Norwood, MA, USA); Temperature and humidity sensor: AM2311A(ASAIR; Guangzhou, China); Compass: SEC295(Beiwei; Wuxi, China)	Master Clock: EGM Multi-function Clock Synchronization Master Clock(CoolShark; Beijing, China); Slave Clock: P88 1588 PTP Clock Synchronization Board(CoolShark; Beijing, China);	48 V to 5 V; 48 V to 9 V; 48 V to 12 V

, our system consists of the following components: a control and transmission module, a data acquisition and conversion module, a clock synchronization module, and a power supply module.

The control and transmission module utilizes the Zynq-7000 SoC, which consists of programmable logic (PL) and processing system (PS) components. The Zynq-7000 SoC offers abundant GPIO (General Purpose Input/Output) resources, providing high scalability for various applications.

The next component is the data acquisition and conversion module. We have employed the MTSS-2003-OBS(R-Sensors LLC; Moscow, Russia), a highly sensitive three-axis seismic sensor, for the collection of seismic data, facilitating the conversion of physical signals into analog electrical signals. Considering the susceptibility of analog electrical signals to attenuation and distortion in long-distance transmission, we have utilized the AD7768(Analog Devices, Inc.; Norwood, MA, USA) to convert the analog electrical signals into digital signals that are more suitable for transmission. Additionally, we have incorporated the AM2311 temperature and humidity sensor and the SEC295(Beiwei; Wuxi, China) high-precision compass. Both sensors exhibit stable operation and low power consumption, allowing real-time monitoring of the node’s working environment. Through the integration of these components, the data we collect will be reliable, stable, comprehensive, and well-suited for seismic data inversion.

The clock synchronization module utilizes the EGM (Ethernet Grand Master) (CoolShark; Beijing, China) as the master clock and the P88 1588 PTP (Precision Time Protocol) as the slave clock. The two clocks are connected via optical fiber to ensure time synchronization. This clock synchronization structure supports time synchronization among multiple nodes and exhibits high scalability. The slave clock provides the control module with PPS (Pulse Per Second) and ToD (Time of Day) signals, facilitating precise timing control. In the event of external time synchronization failure, the P88 1588 PTP will enable its own TCXO hold time with a time accuracy of up to 1.5 μs over 8–30 min. Due to the satellite signal-based clock synchronization system and the programmed configuration to correct for time deviations due to transmission, the P88 1588 PTP can still guarantee time deviations in the microsecond range for a certain period of time when the satellite signal cannot be received, thus solving the problem of inaccurate measurements that we often encounter when monitoring microseism signals on the seafloor.

The power supply module enables voltage conversion from 48 V to stable outputs of 5 V, 9 V, and 12 V, providing power to various components within the node.

2.3. Design of Duplex SQATS

2.3.1. Hardware Design of Duplex SQATS

In the duplex SQATS system, each node consists of two optical ports and is connected to neighboring nodes via two SFP+ optical cables. Data transmission follows the high-speed serial SerDes communication protocol. The implementation of data transmission relies on Gigabit transceivers as the underlying hardware support for the transmission.

In this paper, the 8B/10B data encoding method is used. The encoder in the SerDes transceiver of a node adopts the 8B/10B data encoding method, encoding every parallel byte of data into 10-bit data. The encoder then converts the 10-bit data into a pair of serial differential signals through the serializer and sends it to the SerDes transceiver of the next node. After receiving the serial data from the previous node, the serial differential signal is reconstructed into multipath parallel signals through the deserializer. Parallel signals are finally decoded into the original data through the decoder. The whole process is shown in Figure 1.

2.3.2. Software Design of Duplex SQATS

At the software level, a full-duplex data transmission link is realized by changing the sequential logic of data transmission based on Aurora IP. Based on previous settings of data transmission, the Aurora IP core is set to full-duplex mode. The detailed design of the sequential logic of data transmission is described in the following paragraph.

The data transmission link contains two independent data transmission paths. The first data transmission path is named Path A. Path A starts from the front node to the end node, from which data are sent to the data center through Gigabit Ethernet. The counter path is named Path B. Path B starts from the end node to the front node, from which data are also sent to the data center through Gigabit Ethernet (as shown in Figure 2). The order of composition of the packets between the nodes is Header, Separator, Compass Data, Spacer, Temperature and humidity data, Spacer, Sampling data, MAC address, Timestamp, Seismometer number, and its format is defined as follows in

Table 2. Data packet format transmitted between nodes.

Function	Form	Bytes	Example
Header	0x	4	AA 55 AA 55
Separator	0x	2	FF FF
Compass Data	0x	14	77 0D 00 84 10 04 33 10 00 64 00 73 20 00
Spacer	0x	1.5	FF F
Temperature and humidity data	0x	4	50 E7 01 C0
Spacer	0x	1.5	FF F
Sampling data	0x	21	95 FF FF D7 F4 FF 47 07 00 BB FF FF 2D F3 FF DB FB FF AE FA FF
MAC address	0x	6	00 00 00 00 00 01
Time stamp	ASCII	17	5C FE 08 32 31 35 38 33 36 32 35 30 39 31 39 36 39
Seismometer number	0x	1	01

.

Take Path A in the full-duplex data transmission link as an example. For the front node in Path A, the sequential logic of data transmission only needs to send data packets to its upper-level node. So, whenever submarine monitoring data are integrated, data packets can be transmitted to the neighboring upper-level node.

As for the intermediate nodes in Path A, the sequential logic of data transmission should take both receiving and sending data packets into consideration. To be more specific, the intermediate node first receives data packets from its lower-level node, merges the received data with its sampling data, and then sends the merged data to its upper-level node. To correctly organize the sequential logic of receiving and sending data, the counter data_a_recv_cnt is defined. During the receiving process, counter data_a_recv_cnt increases its value by 1 every clock period. When data_a_recv_cnt reaches a set value, meaning the data packet has been well received, the system can then start the sending process.

The end node takes charge of receiving data packets from its lower-level node and sending data packets to the data center. Therefore, the sequential logic of data transmission ought to receive data from its lower-level node, integrate the data with its own sampled data, and then send the integrated data packet to the data center.

2.4. Data Storage on SD Cards

To further enhance the system’s robustness, we have designed a storage feature based on SD cards. Using the PS (Processing System) side of the control module, we utilize relevant library functions provided by Xilinx, such as f_open, f_seek, f_write, and f_close, to implement the logic design for data storage that meets our requirements. It is worth mentioning that since the storage function is used as a supplement for the failure of link transmission establishment, here we will only store the data of the local node.

2.5. Intelligent Data Link Management System

Considering the lack of intuitive data visualization and network transmission state in the network debugging assistant, a more intelligent link management system is designed based on the QT framework.

After opening the upper computer system, users need to enter their user names and passwords in the login interface (as shown in Figure 3a). Then, the system displays a data link start-up interface (as shown in Figure 3b). To ensure the security of the management system, users are required to register their private accounts. If their private account is authorized, the management system can be operated. After users click the start button of network transmission, the transmission of Path A begins and the message box displays a popup message indicating that Path A has started transmitting data.

In the main interface of the intelligent link management system, real-time data from the link nodes are visualized (as shown in Figure 4a). In addition, the system also monitors the network transmission status. When a malfunction of network transmission in Path A is detected, data transmission in Path A will be shut down. Transmission through the backup path (as shown in Figure 4b) will be launched. A prompting message indicating that Path B has started working will be displayed consequently (as shown in Figure 4c). If errors also occur during data transmission in Path B, data transmission in Path B will be stopped. A corresponding prompting message will also be displayed (as shown in Figure 4d). Submarine monitoring data with 30 s intervals of time will be saved for further data analysis.

3. Results

3.1. Experimental Architecture

The 36 GPIO ports of the Zynq7000 SoC are led out by the adapter plate. The adapter plate is connected to the 4 SPI control wires and 8 data wires of the AD7768(Analog Devices, Inc.; Norwood, MA, USA) and 2 data wires of the P88 1588 PTP time synchronization board(CoolShark; Beijing, China). The three hardware devices, the Zynq7000 development board, AD7768(Analog Devices, Inc.; Norwood, MA, USA), and P88 1588 PTP time synchronization board(CoolShark; Beijing, China) are stacked vertically to form nodes in Duplex SQATS. The whole Duplex SQATS system consists of three nodes representing the front node, the intermediate node, and the end node.

The duplex SQATS is shown in Figure 5. The front node, the intermediate node, and the end node are tagged separately in Figure 5. The PC side simulates the data center and uses a sliding rheostat to simulate the sampled signal source. Each node includes a Zynq7000 SoC main control board(MILIANKE; Changzhou, China), an AD7768(Analog Devices, Inc.; Norwood, MA, USA) sampling card, and a P88 1588 PTP clock synchronization board(CoolShark; Beijing, China). The nodes are connected by optical cables to form a transmission link. The P88 1588 PTP clock synchronization board(CoolShark; Beijing, China) of each node and the EGM master clock(CoolShark; Beijing, China) are also connected by optical cables to form a master–slave clock synchronization network to provide a synchronized time stamp. The end node finally sends the sampled data to the PC via Gigabit Ethernet. Through the host computer program on the PC, the received link data can be read out. The node sequence of Path A is defined as the front node, the intermediate node, and the end node, while the sequence of Path B is defined as the end node, the intermediate node, and the front node.

3.2. Experimental Results and Discussion

3.2.1. Data Analysis of Duplex SQATS

With the help of a network debugging assistant, the sampled data from Path A and Path B are shown in Figure 6a,b. Take data from Path A as an example. The yellow box represents the frame header of the data packet. The following is the number of the data packet, which is marked out by the orange box. Next, data from the front node, the intermediate node, and the end node are bounded separately with the blue box, the green box, and the purple box. Similarly, the components of data from Path B are marked out in Figure 6b.

From what has been demonstrated above, both data packets from Path A and Path B are composed of Header, Seismometer number, Timestamp, MAC address, Acquisition data, and Separator. The components in the packets are the same as in the previous design, indicating that the sampled data are correctly merged and transmitted, which proves our data to be reliable.

After running duplex SQATS for a long time, no data bit loss or packet loss is observed, proving the stability of duplex SQATS. In addition, the two links are completely independent of the transmission point of view, so when one link fails, the other link can still continue to work. When we simulate in our experiments that one transmission path fails to be established while the other is still working properly, this result is consistent with the original design intention, which largely improves the fault tolerance rate of the system and enhances the high performance of the system from the macro point of view. It is worth noting that in Path A, the order of the data is data from the front node, the intermediate node, and the end node. The order of the data in Path B is data from the end node, the intermediate node, and the first node. The order of the data in both paths also caters to the design of the data transmission link.

By verifying the consistency of the dual-link transmission system design and the experimental results, we can conclude that our work has improved the reliability of the system, an improvement that solves our problem of unstable measurements when solving for the fundamental position of the seafloor seismic source.

3.2.2. Data Storage Results of Duplex SQATS

A card reader is used to save the local storage data stored in the SD cards of the three nodes to the PC. For the front node and the end node, the first 100 bytes of data are flashed in the PS side buffer into the Text file on the SD card. So, the data structure stored each time should consist of Header (yellow), Packet, Index (brown), Length (the first black block), and Data. The Data part should include the Seismometer number (the second black block), Timestamp (orange), MAC address (pale yellow), Sampled data (pale blue), Separator(purple), temperature and humidity sensor data (green), and compass data (blue). The content of the Text file from the front node and end node is shown in Figure 7a,b. Data packets stored by these two nodes are correctly organized and match with the data received in the Network Debugging Assistant. For the intermediate node, data are transferred from the PL side into the buffer. Therefore, the data structure stored on the PL side is defined as Header (yellow), Separator (purple), Compass data (blue), Temperature and humidity sensor data (green), Sampled data (pale blue), MAC address (pale yellow), Timestamp (orange) and Seismometer number (black). The content of the Text file from the intermediate node is shown in Figure 7c. Data packets stored by intermediate nodes are also correctly organized and match with the corresponding data received in the Network Debugging Assistant.

Under the premise that the control system and acquisition system of nodes can work normally, each node will continuously store its respective data on an SD card to realize data backup. Once the SD card data storage space is insufficient, the data will overwrite the earliest stored data. Taking node 1 as an example, the storage capacity of the SD card is 64GB, the data of a single storage node is 100 Bytes, and the AD sampling rate is 50 Hz, it can be calculated from the following formula that the SD card of node 1 can continuously store at most 159 days of sampling data, node 3 is the same as node 1, and due to the different frame header settings, the data of a single node of node 2 is 72 Bytes, which can store about 220 days of data.

Time_store_total = Capacity of Memory Card/(Single-node Data × Sampling Rate)

(1)

Since the node can store data for at least 159 days, this means that if the transmission of both links is interrupted, the researchers concerned only need to repair the links and read the data in the SD card within 159 days, then the seismic data collected can be regarded as uninterrupted.

4. Conclusions

This paper introduces a dual-link synchronous acquisition and transmission system aiming at achieving high-precision data acquisition and high-speed transmission from offshore nodes to a shore-based data center. The proposed architecture enables full-duplex transmission between nodes, independent transmission across links, and enhanced storage capacity within nodes for local observation data. These advancements significantly enhance fault tolerance in link data transmission and greatly improve system reliability and robustness. For individual node data acquisition, a seismometer is utilized as the analog data acquisition device, and the AD7768(Analog Devices, Inc.; Norwood, MA, USA) facilitates synchronous acquisition across eight channels. To ensure time synchronization between nodes, each node incorporates a high-precision timestamp based on clock synchronization technology. The synchronization of data channels within and between nodes enables analyzable data acquisition, making the cable-based seafloor seismic observation system practical for seismic data inversion and early warning operations. Temperature and humidity sensors, along with a compass, have been integrated into each node to monitor the internal working environment and serve as supplementary data for seismic data inversion, ultimately enhancing the reliability of inversion results. Furthermore, an intelligent data chain management system has been developed to replace the original upper computer system, enabling real-time visualization of transmission data and network monitoring. This system effectively addresses the common issues of unmeasured, inaccurate, and unstable measurements in seabed microseism data. Compared to other contemporary methods, our proposed system offers simplicity of implementation, stability, and expandability. Based on promising experimental results, its application is anticipated in future seafloor observatories.

In our future work, we plan to collaborate with other research teams to optimize the interface of the upper computer and implement data preprocessing. We will also integrate MEMS vector hydrophones, incorporate their data into the link, and transform the nodes from laboratory-level to watertight enclosures suitable for practical testing environments. This will further validate the reliability of the system and enable monitoring and inversion of small and microseism data, enhancing our ability to provide early warnings for underwater seismic hazards. Moreover, due to the inherent scalability of the system, our underwater sensor network can accommodate additional sensors such as gamma-ray spectrometers for monitoring gamma-ray emissions and other sensors for monitoring inorganic carbon components. This expansion allows for the construction of a more diverse and comprehensive underwater sensor network with a broader range of parameters. Such an enhanced network enables the realization of a more extensive range of underwater detection capabilities, facilitating more accurate monitoring and early warning systems for natural phenomena, particularly focusing on submarine activities.