A gpgpu-based industrial wireless sensor network management system and method

Abstract

The application discloses an industrial wireless sensor network management system and method based on GPGPU, belongs to the technical field of GPGPU application, and is used for solving the technical problems of the current industrial wireless sensor network, such as heavy load loss, high calculation difficulty, complex CPU control, and insufficient load balancing capability. The system comprises an FPGA industrial wireless gateway, a GPGPU computing resource pool, a plurality of sensor nodes, an industrial wireless sensor network, and an industrial wireless sensor network management device. The FPGA industrial wireless gateway is used for scheduling a sensing data processing task to the GPGPU computing resource pool. The sensor nodes and the FPGA industrial wireless gateway communicate and exchange data through the industrial wireless sensor network. The GPGPU computing resource pool comprises a plurality of sensing signal processing terminals based on GPGPU. The sensing signal processing terminals are connected with the FPGA industrial wireless gateway through a CXL switch, are used for receiving the sensing data processing task from the FPGA industrial wireless gateway, and write back the calculation results in real time.

Description

Technical Field

[0001] This invention relates to the field of GPGPU applications, and more particularly to an industrial wireless sensor network management system and method based on GPGPU. Background Technology

[0002] Wireless Sensor Networks (WSNs) are distributed sensor networks whose endpoints can sense and inspect various sensors in the external world. Different sensors communicate wirelessly, allowing for flexible network deployment, easy relocation of devices, and wired or wireless connectivity to the internet. Currently, industrial wireless sensor networks are a key component of the Industrial Internet of Things (IIoT), and their applications in the industrial sector are becoming increasingly widespread.

[0003] However, with the rapid advancement of technology, the types, quantities, and application complexity of wireless sensors connected to industrial wireless sensor networks have undergone tremendous changes. Traditional industrial wireless sensor network computing, management, and control models are increasingly unable to meet the demands of the rapid development of the Industrial Internet of Things (IIoT). Specifically: 1. The number of wireless sensors connected to industrial wireless sensor networks has surged. Traditional wireless sensor sampling data calculations are often performed by the management terminal CPU. Such a massive amount of data inevitably leads to CPU computational overhead, increased load, and impacts the execution of other CPU processes. 2. The types of wireless sensors connected to industrial wireless sensor networks are diverse. The calculation of sampling data from many sensors often requires the collaborative processing of sampling data from other types of sensors (e.g., energy storage motor control requires the collaborative calculation of sampling data from several speed and flow sensors). This increases the difficulty of sensor data calculation and complicates CPU control. 3. Industrial wireless sensor networks must meet the real-time sampling and calculation requirements of a considerable amount of sensor data, with extremely high concurrency. Traditional CPU solutions are not adept at performing such high-concurrency, highly repetitive calculations. Without scientific and reasonable load balancing, excessive latency in sensor data calculation and processing will occur. Summary of the Invention

[0004] This invention provides an industrial wireless sensor network management system and method based on GPGPU to solve the following technical problems: current industrial wireless sensor networks suffer from heavy computing power consumption and load, high computational difficulty, complex CPU control, and insufficient load balancing capabilities.

[0005] The embodiments of the present invention adopt the following technical solutions:

[0006] On one hand, embodiments of the present invention provide an industrial wireless sensor network management system based on GPGPU, the system including: an FPGA industrial wireless gateway, a GPGPU computing resource pool, several sensor nodes, an industrial wireless sensor network, and an industrial wireless sensor network management device;

[0007] The FPGA industrial wireless gateway is used to coordinate sensor data operations in the industrial wireless sensor network and schedule sensor data processing tasks to the GPGPU computing resource pool.

[0008] The sensor nodes include several types, and each node has a unique identification IP; the sensor nodes communicate and exchange data with the FPGA industrial wireless gateway through the industrial wireless sensor network.

[0009] The GPGPU computing resource pool contains several GPGPU-based sensor signal processing terminals, each with a unique device ID. The sensor signal processing terminals are connected to the FPGA industrial wireless gateway via a CXL Switch to receive sensor data processing tasks from the FPGA industrial wireless gateway and write back the calculation results in real time.

[0010] In one feasible implementation, the hardware structure of the FPGA industrial wireless gateway includes at least: an FPGA chip, peripheral electronic devices, DDR4 memory, a wireless signal receiver, peripheral interfaces, and a PCB board.

[0011] The peripheral electronic devices include at least a power interface, a power chip, and a clock chip;

[0012] The peripheral interfaces include at least a JTAG debugging interface, a PCIe interface, and a gigabit network port;

[0013] The PCIe interface is connected to the CXL switch and is used to send wireless sensor signal processing requests to the GPGPU computing resource pool and receive processing results.

[0014] The wireless signal receiver is used to receive wireless sensor signals deployed in the industrial wireless sensor network and transmit the received sampling data to the FPGA chip.

[0015] The PCB board is used to provide a carrier for FPGA chips and peripheral electronic devices.

[0016] In one feasible implementation, the gigabit network port includes at least a first gigabit network and a second gigabit network port;

[0017] The first gigabit Ethernet port is connected to the industrial wireless sensor network management device and is used to send signal sampling and processing results to the user;

[0018] The second gigabit network port is a backup port that supports simultaneous access by multiple users.

[0019] In one feasible implementation, the software structure of the FPGA industrial wireless gateway includes at least: a data caching module, two sensor network protocol decoding modules, a source IP reading module, a hash decoding module, a sensor combined data identification FSM state machine, a sensor data processing confirmation module, and a GPGPU execution unit allocation module.

[0020] In one feasible implementation, the data caching module is used to cache the sampled data from the sensor and send the sampled data to the first sensor network protocol decoding module in a preset format;

[0021] The first sensor network protocol decoding module determines the sensor protocol type based on the input data, decodes the input data according to the protocol type requirements, and sends the decoding result to the source IP reading module.

[0022] The source IP reading module extracts specific bytes from the header of the input data, performs format conversion and concatenation to obtain the source sensor IP of the input data, and sends it to the hash decoding module.

[0023] The hash decoding module performs a hash function calculation on the received source sensor IP to obtain the address of the source sensor's sensor data combination directory stored in the DDR4 memory, and sends it to the sensor combination data identification FSM state machine.

[0024] The sensor combination data identification FSM state machine reads the sensor combination directory address, determines the sensor sampling data to be processed and the data processing method, and packages and sends it to the sensor data processing confirmation module.

[0025] The sensor data processing confirmation module verifies the input sensor sampling data to be processed and the data processing method to confirm that the deployed GPGPU supports the execution of such sensor data calculation tasks; then it sends the sensor sampling data to be processed and the data processing method to the GPGPU execution unit allocation module.

[0026] The GPGPU execution unit allocation module executes the optimal allocation and scheduling algorithm based on the device occupancy status of the sensor signal processing terminals in the GPGPU computing resource pool.

[0027] In one feasible implementation, the software structure of the FPGA industrial wireless gateway includes at least: a DDR read control module, a sensor data secondary read module, a target sensor data extraction module, and a sensor data merging module;

[0028] The DDR read control module is used to read the sensor data combination directory address of the source sensor stored in the DDR4 memory, determine whether the current sensor needs to be combined with the sampled data of other sensors in the sensor network for calculation, and obtain the target sensor IP that needs to be combined for calculation.

[0029] The sensor data secondary reading module is used to perform a secondary reading operation of sensor data based on the target sensor IP.

[0030] The target sensor data extraction module is used to extract the sensor sampling data and data processing method corresponding to the target sensor IP;

[0031] The sensor data merging module is used to perform secondary confirmation on the data extracted by the target sensor data extraction module and execute the corresponding merging operation to obtain the sensor sampling data to be processed and the data processing method.

[0032] In one feasible implementation, the software architecture of the FPGA industrial wireless gateway includes at least: a CXL transceiver buffer module, a CXL transceiver control module, a GPGPU processing result buffer module, a GPGPU processing log generation module, and a network port protocol conversion module.

[0033] In one feasible implementation, the CXL transceiver buffer module is used to convert the sensor sampling data to be processed into CXL protocol format data, and send it to the target sensor signal processing terminal ID through the CXL Switch via the CXL transceiver control module.

[0034] After the target sensing signal processing terminal completes the sensing data calculation, the sensing data calculation result is sent to the CXL transceiver control module via the CXL Switch, and then sent to the GPGPU processing result caching module to cache the data.

[0035] The GPGPU processing log generation module is used to record task processing time, processing method, source sensor type, and sensor signal processing terminal ID.

[0036] The network port protocol conversion module converts the received CXL protocol format data into network port protocol data, and sends the conversion result to the industrial wireless sensor network management device through the network port, thus completing the task of calculating and processing industrial wireless sensor sampling data.

[0037] On the other hand, embodiments of the present invention also provide a method for managing industrial wireless sensor networks based on GPGPU, the method comprising:

[0038] The FPGA industrial wireless gateway receives real-time sampling data from each sensor node in the industrial wireless sensor network through a wireless signal receiver; and based on the data information of the real-time sampling data, it checks the preset sensor sampling combination, reads the target sensor data a second time, and completes the merging of the sensor sampling data to be processed.

[0039] The FPGA industrial wireless gateway assigns corresponding target sensor signal processing terminals to the sensor sampling data to be processed, and sends the corresponding data to the GPGPU of the target sensor signal processing terminal.

[0040] The GPGPU processes the received sensor sampling data and sends the processing result back to the FPGA industrial wireless sensor gateway.

[0041] The FPGA industrial wireless sensor gateway sends the processing result to the industrial wireless sensor network management device through the network port to complete the sensor signal processing.

[0042] In one feasible implementation, based on the data information of the real-time sampling data, the preset sensor sampling combination is checked, the target sensor data is read a second time, and the merging of the sensor sampling data to be processed is completed, specifically including:

[0043] The FPGA industrial wireless gateway reads the sensor data combination directory address of the source sensor stored in the DDR4 memory, determines whether the current sensor needs to be combined with the sampled data of other sensors in the sensor network for calculation, and obtains the target sensor IP that needs to be combined for calculation.

[0044] The FPGA industrial wireless gateway performs a secondary reading operation on the target sensor IP to obtain the sensor sampling data and data processing method corresponding to the target sensor IP, and performs a secondary confirmation and corresponding merging operation to obtain the sensor sampling data and data processing method to be processed.

[0045] The FPGA industrial wireless gateway receives the sensor source IP address decoding result from the hash decoding module, and controls the DDR storage control module to read the current source sensor's sensor data combination list from the DDR4 memory based on the address decoding result.

[0046] The FPGA industrial wireless gateway determines whether the current sensor data needs to be processed in conjunction with sampling data from other types of sensors in the sensor network. If so, it determines the target sensor IP that needs to be processed in conjunction with the data and reads the real-time sampling data of the source sensor IP and one or more target sensor IPs.

[0047] After the sensor data is read, the FPGA industrial wireless gateway sends the read real-time sampling data to the second sensor network protocol decoding module for data protocol decoding, and extracts the valid sensor sampling data based on the decoding result; the valid sensor sampling data is decomposed, spliced ​​and merged to obtain the data merging result;

[0048] The FPGA industrial wireless gateway compares the data merging result with the sensor IPs in the sensor data combination directory, confirms the data a second time, obtains the sensor sampling data to be processed, and verifies the data format and decoding correctness.

[0049] Compared with the prior art, the industrial wireless sensor network management system and method based on GPGPU provided in this embodiment of the invention have the following beneficial effects:

[0050] This invention provides an in-depth analysis of the operating mechanism and core architecture design concept of contemporary industrial wireless sensor networks, and proposes a GPGPU-based industrial wireless sensor network computing management system and its application method to meet the needs of large-scale and high-efficiency data processing and application control of industrial wireless sensor networks in the context of the Industrial Internet era.

[0051] The GPGPU-based industrial wireless sensor network computing management system provided by this invention adopts a novel core design concept of "unified control of a heterogeneous GPGPU computing resource pool by an FPGA industrial wireless gateway." Using wireless network transmission technology and CXL high-speed interconnect technology as its architectural carriers, the FPGA industrial wireless gateway serves as the sensor data sampling and computing scheduling terminal, enabling centralized management and coordination of multiple distributed GPGPUs to complete complex computational tasks involving large-scale wireless sensor sampling signals in parallel within a single time period. This invention designs a GPGPU-based industrial wireless sensor network computing management platform composed of an FPGA industrial wireless gateway, a GPGPU computing resource pool, various types and random numbers of sensor nodes, an industrial wireless sensor network management device, a CXL switch carrying them, and an industrial wireless sensor network. The distributed GPGPUs in the GPGPU computing resource pool are connected to the CXL switch via CXL3.1, and their number can be arbitrarily expanded. The types and numbers of various sensor nodes are unlimited and support arbitrary expansion.

[0052] Furthermore, this invention provides a detailed practical application method for a GPGPU-based industrial wireless sensor network computing management platform. The core of this design lies in allocating the large-scale sensor signal processing computational demands, which would otherwise be executed by the CPU of the industrial wireless sensor network management terminal, to a distributed GPGPU computing resource pool through reasonable scheduling and preprocessing by an FPGA industrial wireless gateway. This enables parallel processing of sensor signals within a single time period, significantly improving the signal processing efficiency of the wireless sensor network and facilitating timely adjustments to industrial equipment by users. In addition, this design utilizes "wireless sensor network + CXL hardware communication" technology to achieve communication between different devices within the wireless sensor network, resulting in higher data transmission and communication efficiency and stronger stability. It can better achieve unified scheduling and control of massive sensor sampling signals, significantly improving system performance. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0054] Figure 1 A schematic diagram of an industrial wireless sensor network management system architecture based on GPGPU provided for an embodiment of the present invention;

[0055] Figure 2 A schematic diagram of the hardware structure of an FPGA industrial wireless gateway provided in an embodiment of the present invention;

[0056] Figure 3 A schematic diagram of the software structure of an FPGA industrial wireless gateway provided in an embodiment of the present invention;

[0057] Figure 4 A flowchart illustrating an industrial wireless sensor network management method based on GPGPU, provided for an embodiment of the present invention;

[0058] Figure 5 The present invention provides a state machine operation flowchart for a sensor combination data recognition FSM state machine. Detailed Implementation

[0059] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.

[0060] First, this invention provides an industrial wireless sensor network management system based on GPGPU. Figure 1 A schematic diagram of an industrial wireless sensor network management system architecture based on GPGPU is provided for an embodiment of the present invention, as shown below. Figure 1 As shown, the system consists of at least an FPGA industrial wireless gateway, a GPGPU computing resource pool, several sensor nodes, an industrial wireless sensor network management device, and a CXL switch and industrial wireless sensor network that carry them.

[0061] Among them, the FPGA industrial wireless gateway is responsible for coordinating all wireless sensor data reception, IP identification, protocol conversion, data preprocessing and other operations carried out on the industrial wireless sensor network computing management platform, and scheduling sensor data processing tasks to the GPGPU computing resource pool.

[0062] Sensor nodes can be categorized into several types, with any number of nodes allowed per category. Each node is assigned a unique identification IP address, allowing users to pinpoint its exact location within a scene and identify which sensor node it should be paired with, by referring to a directory. The sensor nodes and the FPGA industrial wireless gateway communicate and exchange data through the industrial wireless sensor network.

[0063] The GPGPU computing resource pool contains several GPGPU-based sensor signal processing terminals, the number of which can be expanded arbitrarily as needed. Each sensor signal processing terminal is assigned a unique device ID. The GPGPU-based sensor signal processing terminal connects to the FPGA industrial wireless gateway via a CXL switch to receive sensor data computation and processing tasks from the gateway and write back the computation results in real time.

[0064] The industrial wireless sensor network management device is equipped with software APIs for related functions, allowing sensor network managers to monitor the real-time sensor data sampled by each target sensor from the software level, thereby enabling management and overall control of the industrial wireless sensor network. The PC terminal communicates with the FPGA industrial wireless gateway via wired / wireless network.

[0065] Furthermore, the FPGA industrial wireless gateway serves as the task control and scheduling core of the GPGPU-based industrial wireless sensor network computing management platform. The FPGA industrial wireless gateway comprises both hardware and internal software components.

[0066] Figure 2 This is a schematic diagram of the hardware structure of an FPGA industrial wireless gateway provided in an embodiment of the present invention, as shown below. Figure 2 As shown, the FPGA industrial wireless gateway hardware board includes an FPGA chip, peripheral electronic components, DDR4 memory, wireless signal receiver, peripheral interfaces, and a PCB board.

[0067] The FPGA chip contains the application engineering module structure and specific information processing methods necessary for the normal operation of the industrial wireless sensor network computing management platform.

[0068] Peripheral electronic components include power interfaces, power chips, clock chips, and necessary resistors and capacitors. Peripheral interfaces include a JTAG debug interface, a PCIe interface, and two Gigabit Ethernet ports (one of which is a spare). The PCB board provides the platform for the FPGA chip and peripheral electronic components.

[0069] The FPGA industrial wireless gateway hardware board is responsible for the acquisition, processing, and transmission of real-time sampling information from industrial wireless sensor networks. It also receives computation results from the GPGPU computing resource pool and sends these results to the industrial wireless sensor network management device. Furthermore, the board is responsible for powering the FPGA core chip, providing clock input, handling various forms of I / O communication, and exchanging information with peripheral interfaces.

[0070] Gigabit Ethernet port 1 (i.e., the first gigabit Ethernet port) connects to the industrial wireless sensor network management PC and is responsible for sending signal sampling and processing results to users. Gigabit Ethernet port 2 (i.e., the second gigabit Ethernet port) is a backup port that supports simultaneous access by multiple users.

[0071] The wireless signal receiver is responsible for receiving wireless sensor signals deployed in the wireless sensor network and transmitting the received sampling data to the FPGA chip. The PCIe interface connects to the CXL switch and is responsible for sending sensor signal processing requests to the GPGPU computing resource pool, which contains GPGPU-based sensor signal processing terminals, and receiving the processing results.

[0072] Furthermore, Figure 3 This is a schematic diagram of the software structure of an FPGA industrial wireless gateway provided in an embodiment of the present invention, as shown below. Figure 3As shown, the software structure includes at least the following components: an input data caching module, two sensor network protocol decoding modules, a source IP reading module, a hash decoding module, a sensor combined data recognition FSM state machine, a DDR reading control module, a sensor data secondary reading module, a target sensor data extraction module, a sensor data merging module, a sensor data processing confirmation module, a GPGPU execution unit allocation module, a CXL transceiver caching module, a CXL transceiver control module, a GPGPU processing result caching module, a GPGPU processing log generation module, and a network interface protocol conversion module.

[0073] After receiving sampling data from a sensor, the industrial wireless gateway buffers the sampling data from the sensor in the FPGA's input data buffer and sends the data to the Sensor Network Protocol Decoding 1 module (i.e., the first Sensor Network Protocol Decoding module) according to a specific format. The Sensor Network Protocol Decoding 1 module determines the sensor protocol type (WirelessHART, ISA100.11a, WIA-PL, WIA-FA, etc.) based on the received data, decodes the input data according to the protocol requirements, and sends the decoding result to the source IP reading module.

[0074] The source IP reading module extracts specific bytes from the header of the input data, performs format conversion and concatenation to obtain the source sensor IP of the current data, and sends it to the hash decoding module. The hash decoding module performs a hash function calculation on the received source sensor IP address, thereby achieving hash decoding and obtaining the address of the user-preset sensor data combination list stored in the DDR4 memory (this list is user-preset and stores whether the current sensor needs to perform combination calculations with the sampled data of other sensors in the sensor network; if so, it also needs to read the real-time sampled data of other sensors).

[0075] Finally, the address of the source sensor's data combination directory is sent to the state machine of the sensor combination data recognition FSM. The state machine reads the corresponding address in DDR4, determines whether the current sensor needs to be combined with the sampled data of other sensors in the sensor network for calculation based on the read directory, obtains the corresponding target sensor IP, and performs a second reading operation of the sensor data.

[0076] After receiving the sensor sampling data and data processing calculation method from the sensor for secondary reading, the sensor combined data recognition FSM state machine confirms the data for secondary reading and performs the corresponding merging operation, packaging the data and the corresponding calculation method and sending them to the sensor data processing confirmation module.

[0077] The sensor data processing verification module verifies the format and calculation method of the input sensor sampling data to be processed, confirming that the deployed GPGPU supports executing this type of sensor data calculation task. It then sends the sampled data and calculation method to the GPGPU execution unit allocation module.

[0078] The GPGPU execution unit allocation module executes the optimal allocation scheduling algorithm based on the recorded device occupancy status of the GPGPU computing resource pool sensor signal processing terminals, and adds a tag to the task data to allocate it to the corresponding sensor signal processing terminal ID.

[0079] The CXL transceiver buffer module converts the sensor sampling data format, calculation method, and sensor signal processing terminal ID data to be processed into the CXL 3.1 protocol format, buffers it, and sends it to the target sensor signal processing terminal ID through the CXL Switch via the CXL transceiver control module.

[0080] After the target sensor signal processing terminal completes the sensor data calculation, the CXL Switch sends the received sensor data calculation result to the CXL transceiver control module, which then sends it to the GPGPU processing result caching module to cache the data. The GPGPU processing log generation module records information such as task processing time, processing method, source sensor type, and sensor signal processing terminal ID, and then sends the processed data calculation result to the network protocol conversion module. The network protocol conversion module converts the received CXL protocol format data into the network protocol data format and sends the processing result to the industrial wireless sensor network administrator's PC terminal via wired / wireless network, thus completing the industrial wireless sensor sampling data calculation and processing task.

[0081] On the other hand, embodiments of the present invention also provide a method for managing industrial wireless sensor networks based on GPGPU. Figure 4 A flowchart of an industrial wireless sensor network management method based on GPGPU provided for an embodiment of the present invention is shown below. Figure 4 As shown, the method specifically includes the following steps:

[0082] S101, The FPGA-based industrial wireless gateway receives real-time sampling data from various sensor nodes in the industrial wireless sensor network via a wireless signal receiver.

[0083] S102, the wireless signal receiver sends the sampled data to the FPGA chip of the industrial wireless gateway.

[0084] S103 and FPGA decode the read signal, verify the user-preset sensor sampling combination according to the type of received data, sensor source IP and other information, read the target sensor data a second time and complete the data to be processed and merge.

[0085] S104: The FPGA chip assigns the corresponding sensor signal processing terminal to the data to be processed, and sends the corresponding data to the GPGPU of the target terminal device ID through the CXL switch.

[0086] S105. The GPGPU of the target sensor signal processing terminal completes the sensor data processing as required and sends it back to the industrial wireless sensor gateway. The FPGA chip sends the data to be processed to the PC terminal of the industrial wireless sensor network administrator through the network port to complete the sensor signal processing.

[0087] As a feasible implementation method, Figure 5 A flowchart of the state machine operation of a sensor-integrated data recognition FSM state machine is provided for an embodiment of the present invention, as follows: Figure 5 As shown, the specific operation flow of the sensor combination data recognition FSM state machine is as follows:

[0088] 1. The sensor combination data identification FSM state machine receives the sensor source IP address decoding result from the hash decoding module.

[0089] 2. The sensor combination data recognition FSM state machine controls the DDR storage control module to read the sensor data combination list of the current source sensor from DDR4 based on the address decoding result (this list is preset by the user and stored at the address corresponding to DDR4).

[0090] 3. The sensor combination data identification FSM state machine determines whether the current sensor data needs to be processed in conjunction with the sampling data of other types of sensors in the sensor network; if so, it jumps to step 4, otherwise it jumps to step 8.

[0091] 4. Determine the target sensor IP that needs to be computed collaboratively. The sensor combination data identification FSM state machine controls the sensor data secondary reading module to read real-time sampling data from the source sensor IP and one or more target sensor IPs from the wireless sensor signal receiver.

[0092] 5. After the sensor data is read, the read data is sent to the sensor network protocol decoding module 2 to perform data protocol decoding according to the specific protocol specifications, and the decoding result is sent to the target sensor data extraction module to extract the corresponding valid sensor sampling data.

[0093] 6. The effective sensor sampling data is decomposed, spliced ​​and merged by the sensor data merging module, and the sensor combined data identification FSM state machine receives the sensor sampling data merging result required by the sensor data merging module.

[0094] 7. The sensor combination data identification FSM state machine compares the merged result with the sensor IP of the corresponding sensor data combination list read from DDR4 for secondary confirmation, and then jumps to step 11.

[0095] 8. Sensor combination data recognition FSM state machine control sensor data secondary reading module reads real-time sampling data of a single source sensor IP from the wireless sensor signal receiver.

[0096] 9. After the sensor data is read, the read data is sent to the sensor network protocol decoding module 2 to perform data protocol decoding according to the specific protocol specifications, and the decoding result is sent to the target sensor data extraction module to extract the corresponding valid sensor sampling data.

[0097] 10. The FSM state machine compares the sampled data with the sensor IPs in the corresponding sensor data combination list read from DDR4 to confirm whether the list is preset to only require data acquisition and processing of a single sensor.

[0098] 11. The sensor combination data identification FSM state machine sends the data to be processed to the sensor data processing confirmation module to verify the data format and decoding correctness.

[0099] The various embodiments in this invention are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0100] The foregoing has described specific embodiments of the present invention. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0101] The above description is merely an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the embodiments of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the embodiments of the present invention should be included within the protection scope of the present invention.

Claims

1. A GPGPU-based industrial wireless sensor network management system, characterized by, The system comprises at least: an FPGA industrial wireless gateway, a GPGPU computing resource pool, a plurality of sensor nodes, an industrial wireless sensor network, and an industrial wireless sensor network management device; The FPGA industrial wireless gateway is used to plan sensor data operations in the industrial wireless sensor network and schedule sensor data processing tasks to the GPGPU computing resource pool; The sensor nodes include a plurality of types, each node having a unique identity IP; the sensor nodes communicate and exchange data with the FPGA industrial wireless gateway through the industrial wireless sensor network; The GPGPU computing resource pool includes a plurality of GPGPU-based sensor signal processing terminals, each having a unique device ID; the sensor signal processing terminals are connected to the FPGA industrial wireless gateway through a CXL switch, used to receive sensor data processing tasks from the FPGA industrial wireless gateway and write back the calculation results in real time; The software structure of the FPGA industrial wireless gateway comprises at least: a data cache module, a first sensor network protocol decoding module, a second sensor network protocol decoding module, a source IP reading module, a hash decoding module, a sensor combination data identification FSM state machine, a sensor data processing confirmation module, and a GPGPU execution unit allocation module; The software structure of the FPGA industrial wireless gateway further comprises at least: a DDR reading control module, a sensor data secondary reading module, a target sensor data extraction module, and a sensor data merging module; The DDR reading control module is used to read the sensor data combination directory address of the source sensor stored in the DDR4 memory, determine whether the current sensor needs to be combined with other sensor sampling data in the sensor network for calculation, and obtain the target sensor IP that needs to be combined for calculation; The sensor data secondary reading module is used to perform a secondary reading operation of sensor data according to the target sensor IP; the read real-time sampling data is sent to the second sensor network protocol decoding module for data protocol decoding, and the decoding result is sent to the target sensor data extraction module; The target sensor data extraction module is used to extract the sensor sampling data and data processing method corresponding to the target sensor IP; The sensor data merging module is used to perform secondary confirmation and corresponding merging operation on the data extracted by the target sensor data extraction module, to obtain the sensor sampling data and data processing method to be processed.

2. The GPGPU-based industrial wireless sensor network management system of claim 1, wherein, The hardware structure of the FPGA industrial wireless gateway comprises at least: an FPGA chip, peripheral electronic devices, a DDR4 memory, a wireless signal receiver, a peripheral interface, and a PCB board; The peripheral electronic devices comprise at least a power supply interface, a power supply chip, and a clock chip; The peripheral interface comprises at least a JTAG debugging interface, a PCIe interface, and a gigabit network port; The PCIe interface is connected to a CXL switch, used to send a wireless sensor signal processing request to the GPGPU computing resource pool and receive a processing result; The wireless signal receiver is used for receiving the wireless sensor signal of the industrial wireless sensor network deployment and transmitting the received sampling data to the FPGA chip; The PCB card is used for providing a carrier for the FPGA chip and peripheral electronic devices.

3. The GPGPU-based industrial wireless sensor network management system of claim 2, wherein, The gigabit network port at least includes a first gigabit network port and a second gigabit network port; The first gigabit network port is connected with the industrial wireless sensor network management device and is used for transmitting the signal sampling processing result to the user; The second gigabit network port is a standby network port and supports simultaneous access of multiple users.

4. The GPGPU-based industrial wireless sensor network management system of claim 1, wherein, The data cache module is used for caching the sampling data from the sensor and transmitting the sampling data to the first sensor network protocol decoding module according to a preset format; The first sensor network protocol decoding module judges the sensor protocol type according to the input data and decodes the input data according to the protocol type requirement, and transmits the decoding result to the source IP reading module; The source IP reading module extracts the header specific byte of the input data, obtains the source sensor IP of the input data through format conversion and splicing, and transmits the source sensor IP to the hash decoding module; The hash decoding module performs hash function calculation on the received source sensor IP, obtains the sensor data combination directory address of the source sensor stored in the DDR4 memory, and transmits the sensor data combination directory address to the sensor combination data identification FSM state machine; The sensor combination data identification FSM state machine reads the sensor data combination directory address, determines the sensor sampling data to be processed and the data processing mode, and packages and transmits the sensor sampling data to be processed and the data processing mode to the sensor data processing confirmation module; The sensor data processing confirmation module verifies the sensor sampling data to be processed and the data processing mode, confirms that the deployed GPGPU supports executing the sensor data calculation task, and then transmits the sensor sampling data to be processed and the data processing mode to the GPGPU execution unit distribution module; The GPGPU execution unit distribution module executes the optimal distribution scheduling algorithm according to the device occupation condition of the sensor signal processing terminal of the GPGPU calculation resource pool.

5. The GPGPU-based industrial wireless sensor network management system of claim 1, wherein, The software structure of the FPGA industrial wireless gateway at least further includes a CXL transceiving cache module, a CXL transceiving control module, a GPGPU processing result cache module, a GPGPU processing log generation module, and a network port protocol conversion module.

6. The GPGPU-based industrial wireless sensor network management system of claim 5, wherein, The CXL transceiving cache module is used for converting the sensor sampling data to be processed into CXL protocol format data, and transmitting the CXL protocol format data to the target sensor signal processing terminal ID through the CXL transceiving control module and the CXL Switch; After the target sensor signal processing terminal completes the sensor data calculation, the sensor data calculation result is transmitted to the CXL transceiving control module through the CXL Switch, and then the sensor data calculation result is transmitted to the GPGPU processing result cache module for data caching; The GPGPU processing log generation module is used for recording the task processing time, the processing mode, the source sensor category, and the sensor signal processing terminal ID; The second gigabit network port is a standby network port and supports simultaneous access of multiple users. The network port protocol conversion module converts the received CXL protocol format data into network port protocol data, and sends the conversion result to the industrial wireless sensor network management device through the network port, thereby completing the industrial wireless sensor sampling data calculation and processing task.

7. A GPGPU-based industrial wireless sensor network management method applied to the GPGPU-based industrial wireless sensor network management system of any one of claims 1-6, characterized in that, The method comprises: The FPGA industrial wireless gateway receives real-time sampling data of each sensor node in the industrial wireless sensor network through a wireless signal receiver; According to the data information of the real-time sampling data, a preset sensor sampling combination is checked, target sensor data is read again, and merging of the processed sensor sampling data is completed, specifically comprising: The FPGA industrial wireless gateway reads the address of the sensor data combination directory of the source sensor stored in the DDR4 memory, determines whether the current sensor needs to be combined with other sensor sampling data in the sensor network, and obtains the target sensor IP that needs to be combined; The FPGA industrial wireless gateway performs a secondary reading operation of the sensor data according to the target sensor IP, sends the read real-time sampling data to a second sensor network protocol decoding module for data protocol decoding, and sends the decoding result to a target sensor data extraction module; The target sensor data extraction module extracts the sensor sampling data corresponding to the target sensor IP and the data processing method; The sensor data merging module performs secondary confirmation on the data extracted by the target sensor data extraction module and performs corresponding merging operation, thereby obtaining the processed sensor sampling data and the data processing method; The FPGA industrial wireless gateway assigns a corresponding target sensor signal processing terminal to the processed sensor sampling data, and sends the corresponding data to the GPGPU of the target sensor signal processing terminal; The GPGPU performs data processing on the received processed sensor sampling data, and sends the processing result back to the FPGA industrial wireless sensor gateway; The FPGA industrial wireless sensor gateway sends the processing result to the industrial wireless sensor network management device through the network port, thereby completing the sensor signal processing.

Images