Wireless communication data acquisition method

Abstract

The invention discloses a wireless communication data acquisition method, which relates to the technical field of wireless communication, and comprises the following steps of: deploying an edge acquisition node which supports dual-band communication and is internally provided with a network address translation gateway, performing logic subnet division on the edge acquisition node, and dynamically allocating a private IP (Internet Protocol) address; the normalized data packet in the JSON format is uploaded to a central scheduling server through an IPsec encryption tunnel; frequency bands are automatically switched or multi-path redundancy transmission is started in combination with link quality evaluation; data integrity verification, time sequence alignment and clock deviation compensation are carried out at the server side; and meanwhile, an electromagnetic interference suppression module and a lightweight anomaly detection model are integrated, so that the real-time state evaluation of the edge side is realized. According to the method, the stability, the integrity and the intelligent level of wireless data acquisition in a complex industrial environment are effectively improved, and low-cost digital upgrading of stock equipment is supported.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and in particular to a method for acquiring wireless communication data. Background Technology

[0002] With the deep integration of modern information technology, especially artificial intelligence and the Industrial Internet of Things, digital management of production equipment has become a core path for manufacturing enterprises to improve operational efficiency, ensure product quality, and achieve intelligent decision-making. Real-time acquisition and structured storage of equipment operation data not only provide crucial support for process parameter optimization, quality defect tracing, and predictive maintenance, but also constitute the data foundation of the intelligent manufacturing system. Against this backdrop, how to efficiently and cost-effectively acquire the operating status information of heterogeneous equipment scattered throughout the workshop has become a key link in promoting the digital transformation of SMEs. Currently, most SMEs still widely adopt a stand-alone operation mode, relying on paper work orders and manual recording in their production processes. They lack unified data interface standards, and early equipment selection did not reserve digital upgrade capabilities, resulting in fragmented equipment layouts, diverse communication protocols, and chaotic network configurations, severely restricting the feasibility of plant-wide data integration.

[0003] For industrial equipment with Ethernet interfaces or serial communication ports, traditional data acquisition solutions typically rely on wired networks for physical connections, connecting the equipment to a central monitoring system via switches or industrial gateways. While this approach offers high stability in newly built production lines, it faces significant bottlenecks in existing infrastructure upgrades. Firstly, older workshops have limited wiring space, making conduit and cable routing complex and prone to interfering with existing production lines, resulting in high downtime and upgrade costs. Secondly, conflicting IP addresses from different equipment manufacturers are common, especially when control system firmware is closed, source code is unavailable, or original manufacturer technical support is lacking. This makes it impossible to modify network parameters using conventional methods, hindering the coexistence of multiple devices on the same local area network and creating data silos. Furthermore, even with physical connectivity, stable and reliable data aggregation cannot be achieved without resolving address conflicts and protocol compatibility issues.

[0004] Existing technologies exhibit multiple limitations in addressing these challenges. First, data acquisition architectures based on fixed cabling lack deployment flexibility, making it difficult to adapt to frequent equipment adjustments or temporary additions to production needs, resulting in poor scalability. Second, while mainstream industrial gateways support some protocol conversion functions, they generally lack Network Address Translation (NAT) and custom subnetting capabilities, making it impossible to isolate and remap conflicting IPs without altering the original device configuration, leading to network failures. Furthermore, existing wireless acquisition solutions largely focus on sensor-level low-power wide-area network (LPWAN) applications, whose bandwidth and real-time performance are insufficient to meet the high-frequency data interaction requirements of control devices such as PLCs and CNCs, and they lack robust optimization for the electromagnetic environment and metal shielding effects of workshops. Summary of the Invention

[0005] The purpose of this invention is to provide a wireless communication data acquisition method to solve the problems of data silos and poor transmission reliability in industrial equipment caused by IP address conflicts, unstable wireless links, and inconsistent data timing in the prior art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A wireless communication data acquisition method includes the following specific steps: Step 1: Deploy edge acquisition nodes with multi-channel wireless communication capabilities within the workshop. Each edge acquisition node is equipped with a wireless transceiver module supporting dual-band 2.4GHz and 5.8GHz, and has a built-in embedded gateway unit with network address translation function. The edge acquisition nodes establish a wired connection with surrounding industrial equipment through Ethernet or serial communication interfaces, collect equipment operation data in real time, and then upload the aggregated data to the central control system or adjacent nodes through a wireless link to realize remote data transmission and centralized monitoring. Step 2: Logically subnet each edge acquisition node and assign an independent Virtual Local Area Network (VLAN) identifier to each edge acquisition node, ranging from 1 to 4094. For multiple industrial devices within the same edge acquisition node that use the same static IP address due to manufacturer restrictions, such as 192.168.0.1, multiple routing modules with NAT functionality can be deployed inside the edge acquisition node to map each device to a different private IP address within the subnet to which the edge acquisition node belongs, such as 192.168.1.100, 192.168.1.101, etc., thereby achieving parallel access and address uniqueness for devices within the edge acquisition node. By combining VLAN isolation and NAT translation within the edge acquisition node, the system can simultaneously avoid IP address conflicts at two levels: across edge acquisition nodes and within edge acquisition nodes, ensuring network connectivity and communication isolation. Step 3: Automatically assign IP addresses within each logical subnet using the Dynamic Host Configuration Protocol (DHCP) service, with a lease time of 7200 seconds. When a device connects for the first time, the address allocation process is triggered. The edge acquisition node records the mapping relationship between the device's MAC address and the assigned IP address and establishes a local address resolution table. Step 4: Configure a unified data aggregation agent on the central dispatch server. The data aggregation agent listens to the encrypted data tunnels from each edge acquisition node. The encrypted data tunnels are established using the IPsec protocol with a pre-shared key. The security association lifecycle is 3600 seconds. Key updates are negotiated using Diffie-Hellman Group 14 parameters. Step 5: The edge acquisition node performs protocol parsing and format normalization on the raw equipment data it collects, identifies the data frame structure of three mainstream industrial protocols: Modbus-TCP, EtherNet / IP and Profinet, extracts the process variable values, equipment status codes and timestamp information, and encapsulates them into a unified JSON format data packet with a timestamp accuracy of 1 millisecond. Step 6: Perform wireless link quality assessment. The edge acquisition node periodically sends probe signals to measure signal strength, signal-to-noise ratio and retransmission rate. The signal strength threshold is set to -75dBm. When the signal-to-noise ratio is lower than 20dB, the link optimization mechanism is activated. The optimization mechanism includes automatic switching of operating frequency band, adjustment of modulation order and activation of forward error correction coding. Step 7: When the main link quality is detected to be continuously lower than the threshold for more than 30 seconds, the multi-path redundancy transmission strategy is activated, and the same data packet is forwarded to the central scheduling server through two or more different edge acquisition nodes. The transmission delay difference between each path is controlled within 50 milliseconds. The server performs data deduplication and reassembly based on the sequence number. Step 8: The central scheduling server performs integrity verification and time series alignment on the received data packets, uses the SHA-256 algorithm to verify the authenticity of the data source, and synchronizes the clocks of each node in combination with the network time protocol. The clock deviation compensation algorithm is based on the least squares method to fit the time offset of each node. Step 9: Establish an equipment data index library, using the unique identifier of the equipment as the index key, to store its historical operating data sequence. The data storage period shall not be less than 365 days, and it shall support multi-dimensional query and retrieval by time range, process stage and equipment type. Step 10: Integrate an electromagnetic interference suppression module into the edge acquisition node. The module monitors the spectrum occupancy of the 2.4GHz band in real time, and uses an adaptive notch filter to eliminate the influence of fixed frequency interference sources. The notch depth is not less than 25dB, and the bandwidth adjustment resolution is 100kHz.

[0007] Furthermore, in step 1, the maximum wireless coverage radius of the edge acquisition node is 50 meters, and the minimum effective communication distance in a dense metal structure environment is no less than 20 meters. The nodes support wireless cascading expansion and can form up to 3 levels of relay network. A single node can connect to no less than 16 heterogeneous industrial devices at the same time.

[0008] Furthermore, in step 2, the virtual LAN identifier is uniformly assigned by the central configuration management system, and the subnet status is managed using a binary bit marking method. Subnet creation and deletion operations are initiated through out-of-band management channels, and the configuration change response time is less than 500 milliseconds.

[0009] Furthermore, in step 3, the Dynamic Host Configuration Protocol (DHCP) service sets the address pool capacity to 250 available addresses, adopts the first-match priority strategy for address allocation, and implements address conflict detection through gratuitous ARP packet probing, with 3 probes and an interval of 200 milliseconds.

[0010] Furthermore, in step 4, the IPsec tunnel adopts the ESP encapsulation mode, the encryption algorithm is AES-256, the authentication algorithm is HMAC-SHA1, the tunnel establishment process includes the IKEv2 negotiation phase, and identity authentication is completed based on the pre-installed device certificate, which is valid for 730 days.

[0011] Furthermore, in step 5, the JSON format data packet contains the device ID, collection time, data point list, and verification field. Each entry in the data point list contains a variable name, value, and unit. The unit is represented using the International System of Units (SI) standard symbols, and the numerical precision is retained to 4 decimal places.

[0012] Furthermore, in step 6, the modulation order is dynamically adjusted according to the link quality, supporting four modes: BPSK, QPSK, 16-QAM, and 64-QAM. The forward error correction coding uses (7,5) convolutional code with a coding rate of 1 / 2, and the decoding uses the Viterbi algorithm.

[0013] Furthermore, in step 7, the multi-path redundancy transmission strategy is controlled by a routing strategy table issued by the central scheduling server. The table entries include node priority, path weight, and failure detection period. The path weight is dynamically updated based on the historical transmission success rate, with an update period of 600 seconds.

[0014] Furthermore, the time offset calculation model for the least squares fitting in step 8 is as follows: The coefficients a and b are solved using at least 5 sets of time samples, and the root mean square of the fitting residual is less than 0.5 milliseconds.

[0015] Furthermore, in step 9, the unique device identifier is generated by concatenating the device type code, serial number, and deployment location code, with a total length of 32 characters. It adopts the Base32 encoding rule, and the index library uses a columnar storage structure to optimize query efficiency, with a single complex query response time of less than 200 milliseconds.

[0016] Furthermore, in step 10, the center frequency of the adaptive notch filter is determined by the spectral energy detection module, the detection period is 100 milliseconds, the filter order is 8, the coefficient update adopts the least mean square algorithm, and the convergence factor is set to 0.01.

[0017] Furthermore, a lightweight containerized data preprocessing engine is run on the edge acquisition node. This engine loads a pre-trained anomaly detection model. The model input is a sequence of sensor data for 128 consecutive time steps, and the output is a device operating status score with a score range of 0 to 1 and a threshold set to 0.85.

[0018] Furthermore, the anomaly detection model is built on a long short-term memory network, containing two hidden layers with 64 neurons in each layer, using the tanh activation function, and the output layer uses the sigmoid function. The model training dataset contains no less than 100,000 normal and abnormal working condition samples.

[0019] Furthermore, a dynamic scheduling mechanism for wireless channel occupancy is established. Each edge acquisition node performs carrier sensing before sending data, and the idle waiting time is randomly set between 10 and 30 milliseconds. Burst data transmission adopts a fragmentation mechanism, with a maximum frame length of 1024 bytes.

[0020] Furthermore, in the fragmentation mechanism, each data frame contains a frame sequence number, a total number of fragments, and a current fragment index. If the receiving end fails to collect all fragments within 100 milliseconds, it triggers a retransmission request. The retransmission request is sent through a dedicated control channel, which occupies a bandwidth of 200 kHz.

[0021] Furthermore, the central dispatch server is configured with a dual-machine hot standby architecture. The primary and standby servers synchronize status information through a dedicated heartbeat link with a heartbeat cycle of 1 second and a failover time of less than 3 seconds. Data synchronization adopts incremental log replication with a latency of no more than 100 milliseconds.

[0022] Compared with the prior art, the beneficial technical effects of the present invention are as follows: This invention achieves logical isolation and unified access for multi-source heterogeneous devices by deploying edge acquisition nodes with network address translation capabilities, without modifying the original device IP configuration, effectively solving the data silo problem caused by IP address conflicts.

[0023] This invention employs dual-band wireless communication and a dynamic link optimization mechanism, which significantly improves the stability and anti-interference capability of data transmission in complex electromagnetic environments and overcomes the shortcomings of traditional wireless solutions that are prone to interruption in metal workshops.

[0024] This invention introduces multi-path redundant transmission and time series alignment technology to ensure the integrity and temporal consistency of key process data and improve the reliability of upper-level data analysis.

[0025] This invention combines a lightweight model to perform real-time status assessment at the edge, reducing the computational load on the central server and realizing an integrated closed loop from data acquisition to preliminary intelligent analysis.

[0026] The overall solution of this invention does not require large-scale wiring modifications, is flexible in deployment, and has strong scalability, making it particularly suitable for the low-cost digital upgrade needs of existing equipment in small and medium-sized manufacturing enterprises. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the overall technical solution architecture of a wireless communication data acquisition method proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of multi-path redundant transmission and adaptive optimization of link quality between edge acquisition nodes and central scheduling server proposed in this invention. Figure 3 This is a schematic diagram illustrating the specific steps of a wireless communication data acquisition method proposed in this invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0029] Currently, with the deep integration of modern information technology, especially artificial intelligence and the Industrial Internet of Things, digital management of production equipment has become a core path for manufacturing enterprises to improve operational efficiency, ensure product quality, and achieve intelligent decision-making. Real-time acquisition and structured storage of equipment operation data not only provide crucial support for process parameter optimization, quality defect tracing, and predictive maintenance, but also constitute the data foundation of the intelligent manufacturing system. In this context, how to efficiently and cost-effectively acquire the operating status information of heterogeneous equipment scattered throughout the workshop has become a key link in promoting the digital transformation of SMEs. To address the aforementioned technical issues, this invention proposes deploying edge acquisition nodes with network address translation capabilities. This achieves logical isolation and unified access for multi-source heterogeneous devices without modifying the original device IP configuration, effectively solving the data silo problem caused by IP address conflicts. Employing dual-band wireless communication and dynamic link optimization mechanisms significantly improves data transmission stability and anti-interference capabilities in complex electromagnetic environments, overcoming the shortcomings of traditional wireless solutions prone to interruptions in metal workshops. Introducing multi-path redundant transmission and time series alignment technology ensures the integrity and temporal consistency of key process data, improving the reliability of upper-layer data analysis. Combining a lightweight model for real-time status assessment at the edge reduces the computational load on the central server, achieving an integrated closed loop from data acquisition to preliminary intelligent analysis. The overall solution requires no large-scale cabling modifications, is flexible in deployment, and highly scalable, making it particularly suitable for the low-cost digital upgrade needs of existing equipment in small and medium-sized manufacturing enterprises. It is also applied to a wireless communication data acquisition method.

[0030] refer to Figure 1The overall technical architecture of this invention includes a central dispatch server, multiple edge acquisition nodes, and industrial equipment distributed throughout the workshop. The industrial equipment includes, but is not limited to, programmable logic controllers (PLCs), computer numerical control (CNC) machine tools, human-machine interfaces (HMIs), and other automation control units equipped with Ethernet or serial communication interfaces. The edge acquisition nodes, as the core hardware carrier of this invention, are deployed within the workshop. Each edge acquisition node is equipped with a wireless transceiver module supporting both 2.4GHz and 5.8GHz dual-band frequencies and has a built-in embedded gateway unit with network address translation functionality. The edge acquisition nodes establish wired connections with surrounding industrial equipment via Ethernet or serial communication interfaces, collecting real-time equipment operation data. The collected data is then uploaded to the central control system or adjacent nodes via wireless links, achieving remote data transmission and centralized monitoring.

[0031] In step 1 of the aforementioned wireless communication data acquisition method, edge acquisition nodes with multi-channel wireless communication capabilities are deployed within the workshop. Their maximum wireless coverage radius is 50 meters, and the minimum effective communication distance in environments with dense metal structures is no less than 20 meters. Nodes support wireless cascading expansion, forming up to a 3-level relay network. A single node can simultaneously connect to no fewer than 16 heterogeneous industrial devices. Specifically, the physical deployment of the edge acquisition nodes follows a grid-like principle, with the distance between adjacent nodes not exceeding 40 meters to ensure that the signal overlap coverage area is greater than 10 meters, thus providing a physical basis for subsequent multi-path redundant transmission. The wireless transceiver module adopts a MIMO antenna array design, including four spatial diversity antennas, supporting the IEEE 802.11ac standard, with a theoretical peak rate of 867 Mbps. The embedded gateway unit is built on an ARM Cortex-A53 quad-core processor with a main frequency of 1.2 GHz, equipped with 1GB DDR3 memory and 8GB eMMC flash memory, running a customized Linux kernel, and integrating iptables and netfilter frameworks to implement Network Address Translation (NAT) functionality. For industrial equipment that only has RS-232 or RS-485 serial interfaces, the edge acquisition node completes the protocol conversion through the onboard UART to Ethernet bridge chip, encapsulates the serial data stream into TCP / IP data packets, and then incorporates them into the unified processing flow.

[0032] In step 2 of the aforementioned wireless communication data acquisition method, logical subnetting is performed on each edge acquisition node, and an independent virtual local area network (VLAN) identifier is assigned to each edge acquisition node, ranging from 1 to 4094. For multiple industrial devices within the same edge acquisition node that use the same static IP address due to manufacturer restrictions, such as 192.168.0.1, multiple routing modules with NAT functionality can be deployed within the edge acquisition node to map each device to a different private IP address within the subnet to which the edge acquisition node belongs, such as 192.168.1.100, 192.168.1.101, etc., thereby achieving parallel access and address uniqueness for devices within the edge acquisition node. By combining VLAN isolation and NAT translation within the edge acquisition node, the system can simultaneously avoid IP address conflicts at two levels: across edge acquisition nodes and within edge acquisition nodes, ensuring network connectivity and communication isolation. Specifically, logical subnetting is performed on each edge acquisition node. The central configuration management system uses a 4096-bit bitmap to uniformly allocate virtual LAN identifiers, i.e., VLAN IDs, and issues subnet creation and deletion commands through an out-of-band management channel. Its configuration change response time is less than 500 milliseconds. The system assigns a unique VLAN ID to each edge acquisition node. Each edge acquisition node creates a corresponding virtual network interface in its Linux kernel and binds all downlink device interfaces to this interface. Within this logical subnet, the device IP address range is strictly limited to 192.168.x.0 / 24, where x equals the lower 8 bits of the assigned VLAN ID. For example, when the VLAN ID is 101, the address range is 192.168.101.0 / 24. This design ensures complete isolation of the IP address spaces of different nodes, fundamentally eliminating the possibility of IP conflicts between edge acquisition node devices. For multiple devices with the same static IP address (e.g., all 192.168.0.1) within the same edge acquisition node due to manufacturer restrictions, multiple routing modules with NAT functionality can be deployed within the edge acquisition node to map each device to a different local IP address within the edge acquisition node's address range (e.g., 192.168.x.0 / 24), such as 192.168.x.100 or 192.168.x.101. This achieves parallel access and address uniqueness for multiple devices with the same IP address within the edge acquisition node. By centrally planning VLANs and IP address ranges to achieve cross-edge acquisition node isolation, and by using NAT translation within the edge acquisition node to resolve local IP address conflicts, the system can avoid IP address conflicts across the entire network, ensuring stable access and communication for various industrial devices.

[0033] Specifically, the virtual LAN identifier is uniformly assigned by the central configuration management system, and a binary bit-based method is used to manage the subnet status. Subnet creation and deletion operations are initiated via out-of-band management channels, with a configuration change response time of less than 500 milliseconds. The central configuration management system maintains a 4096-bit bitmap array, where each bit corresponds to a VLAN ID. A value of 1 indicates that the ID is occupied, and a value of 0 indicates that it is available. When a new edge acquisition node is deployed, the system scans the bitmap to find the first available bit, sets it to 1, and sends the corresponding VLAN ID to the node. After receiving the VLAN ID, the edge acquisition node creates the corresponding virtual network interface in its Linux kernel and binds all downlink device interfaces to this virtual interface. Within this logical subnet, device IP address allocation is strictly limited to the 192.168.x.0 / 24 subnet, where x equals the lower 8 bits of the VLAN ID. For example, when the VLAN ID is 101, the allocated address range is 192.168.101.0 / 24. This design ensures that the IP address spaces of devices managed by any two different edge acquisition nodes do not overlap, fundamentally eliminating the possibility of cross-node IP conflicts.

[0034] In step 3 of the aforementioned wireless communication data acquisition method, an IP address is automatically allocated within each logical subnet via the Dynamic Host Configuration Protocol (DHCP) service, with a lease time of 7200 seconds. When a device first connects, the address allocation process is triggered. The edge acquisition node records the mapping relationship between the device's MAC address and the allocated IP address and establishes a local address resolution table. Specifically, the DHCP service sets the address pool capacity to 250 available addresses. Address allocation adopts a first-match priority strategy, and address conflict detection is achieved through gratuitous ARP packet probing, with three probes conducted at 200 millisecond intervals. The DHCP server process built into the edge acquisition node listens on UDP port 67. When a DHCP Discover broadcast packet sent by a new device is detected, the first unallocated address is selected from the preset address pool, and a DHCP Offer packet is constructed in response. Before formal allocation, the server sends three gratuitous ARP requests to the candidate address. If no response is received within the 200-millisecond interval, the address is confirmed as available and allocation is completed; if a response is received on any of the three requests, an address conflict is determined, and the search for that address is skipped. After successful allocation, the server writes the device's MAC address, allocated IP address, and lease start and end times into an SQLite database, forming a persistent address resolution table. This table is also used for reverse lookups; that is, when a data packet is received destined for the local subnet, the target device's MAC address can be quickly obtained by looking up the table, completing Layer 2 frame encapsulation.

[0035] In step 4 of the aforementioned wireless communication data acquisition method, a unified data aggregation agent is configured on the central dispatch server. This agent listens for encrypted data tunnels from each edge acquisition node. The encrypted data tunnels are established using a pre-shared key IPsec protocol, with a security association lifespan of 3600 seconds. Key updates are negotiated using Diffie-Hellman parameter group 14. Specifically, the IPsec tunnel uses ESP encapsulation mode, AES-256 encryption algorithm, and HMAC-SHA1 authentication algorithm. The tunnel establishment process includes an IKEv2 negotiation phase, and authentication is based on a pre-configured device certificate with a validity period of 730 days. The central dispatch server runs a StrongSwan IPsec daemon configured in responsive mode, listening on UDP port 500. Each edge acquisition node is pre-configured with a unique X.509 device certificate and corresponding private key at the factory. The certificate subject name includes the node serial number and deployment location code. When an edge acquisition node starts up, it actively initiates IKEv2 negotiation with the central dispatch server to exchange certificates for two-way authentication. After successful authentication, both parties generate shared key material based on Diffie-Hellman key set 14, deriving the AES-256 key and HMAC-SHA1 authentication key for ESP encryption. Once the Security Association (SA) is established, all data streams from the edge acquisition nodes to the central dispatch server are encapsulated within the ESP header. The source address in the external IP header is the public IP address of the edge node, and the destination address is the public IP address of the server. The SA lifecycle is set to 3600 seconds, with automatic key renewal negotiation triggered 300 seconds before expiration to ensure the security of long-term communication.

[0036] In step 5 of the aforementioned wireless communication data acquisition method, the edge acquisition node performs protocol parsing and format normalization on the acquired raw device data. It identifies the data frame structures of three mainstream industrial protocols: Modbus-TCP, EtherNet / IP, and Profinet, extracting process variable values, device status codes, and timestamp information, and encapsulating them into a unified JSON format data packet with a timestamp accuracy of 1 millisecond. Specifically, the JSON format data packet includes the device ID, acquisition time, data point list, and verification field. Each entry in the data point list includes a variable name, value, and unit, with units represented using the International System of Units (SI) and values ​​retained to four decimal places. The edge acquisition node has a built-in protocol parsing engine and uses a state machine model to achieve multi-protocol identification. Upon receiving an Ethernet data frame, it first checks the TCP destination port: port 502 corresponds to Modbus-TCP, port 44818 corresponds to EtherNet / IP, and port 34962 or 34963 corresponds to Profinet. For Modbus-TCP, the function code, starting address, and register count are parsed, and 16-bit integer values ​​are extracted from the response data and converted to floating-point numbers according to the engineering range. For EtherNet / IP, the attribute IDs and data types in the CIP object model are parsed, and structured data is extracted. For Profinet, the slot numbers and sub-slot numbers in the IO data frames are parsed and mapped to a predefined variable table. All extracted variable values, along with high-precision timestamps, are organized into JSON objects. The high-precision timestamps are provided by a hardware real-time clock chip with an accuracy of ±1 millisecond. For example, a typical data packet structure is as follows: {"device_id":"CNC_001","timestamp":"2023-10-05T14:23:01.123Z","data_points":[{"name":"spindle_speed"," value":1250.0000,"unit":"rpm"},{"name":"tool_temp","value":85.5000,"unit":"°C"}],"checksum":"a1b2c3d4"} The verification field is calculated using the CRC32 algorithm on all content except itself, and is used by the receiving end to quickly verify the integrity of the data.

[0037] In step 6 of the aforementioned wireless communication data acquisition method, a wireless link quality assessment is performed. Edge acquisition nodes periodically send probe signals to measure signal strength, signal-to-noise ratio (SNR), and retransmission rate. The signal strength threshold is set to -75dBm. When the SNR falls below 20dB, a link optimization mechanism is activated. This mechanism includes automatic switching of the operating frequency band, adjustment of the modulation order, and activation of forward error correction coding. Specifically, the modulation order is dynamically adjusted based on link quality, supporting four modes: BPSK, QPSK, 16-QAM, and 64-QAM. Forward error correction coding uses a (7,5) convolutional code with a coding rate of 1 / 2, and decoding uses the Viterbi algorithm. Edge acquisition nodes perform a link quality assessment every 5 seconds, reading physical layer statistics, including Received Signal Strength Indicator (RSSI), Signal-to-Noise Ratio (SNR), and MAC layer retransmission count, through the driver layer API. The optimization mechanism is triggered when the RSSI remains below -75dBm or the SNR falls below 20dB for 10 seconds. First, a switch from the 5.8 GHz band to the 2.4 GHz band is attempted, as the latter has a longer wavelength, stronger diffraction capability, and better penetration performance in environments with metal obstructions. If the 2.4 GHz band also degrades, the modulation order is reduced, for example, from 64-QAM to 16-QAM, sacrificing data rate for link robustness. Simultaneously, (7,5) convolutional codes are enabled, with a coding rate of 1 / 2, meaning that for every 1 bit input, 2 bits of coded output are generated. The receiver uses the Viterbi algorithm for maximum likelihood decoding to correct burst errors. This three-stage optimization strategy ensures that basic communication capabilities are maintained even under adverse channel conditions.

[0038] refer to Figure 2The core principles of multi-path redundant transmission and adaptive link quality optimization between edge acquisition nodes and the central scheduling server proposed in this invention are clearly demonstrated. In step 7 of the aforementioned wireless communication data acquisition method, when the main link quality is detected to be continuously lower than the threshold for more than 30 seconds, a multi-path redundant transmission strategy is activated. The same data packet is forwarded to the central scheduling server through two or more different edge acquisition nodes. The transmission delay difference between each path is controlled within 50 milliseconds. The server performs data deduplication and reassembly based on the sequence number. Specifically, the multi-path redundant transmission strategy is controlled by a routing strategy table issued by the central scheduling server. The table entries include node priority, path weight, and failure detection period. The path weight is dynamically updated based on the historical transmission success rate, with an update period of 600 seconds. The central scheduling server maintains a global routing strategy table, initially assigning one main path and two backup paths to each edge acquisition node. When the main path quality deteriorates and times out, the edge acquisition node queries the strategy table and selects the backup path with the highest weight. Data packets are copied multiple times before transmission, with each copy appended with a unique sequence number and path identifier. After receiving all copies, the central scheduling server caches them in a sliding window queue, with the window size corresponding to a 50-millisecond latency tolerance. When data packets with the same sequence number appear in the queue, only the earliest arriving copy is retained, and the rest are discarded, thus completing deduplication. The reassembled data stream is then submitted to the upper-layer application in sequence number order to ensure timing correctness.

[0039] In step 8 of the aforementioned wireless communication data acquisition method, the central scheduling server performs integrity verification and time series alignment on the received data packets. It uses the SHA-256 algorithm to verify the authenticity of the data source and synchronizes the clocks of each node using the network time protocol. The clock offset compensation algorithm is based on the least squares method to fit the time offset of each node. Specifically, the least squares method-fitted time offset calculation model is as follows: Here, coefficient 'a' represents the frequency offset, and coefficient 'b' represents the phase offset. Coefficients 'a' and 'b' are calculated using at least five sets of time sample pairs, with the root mean square of the fitting residual less than 0.5 milliseconds. The server first verifies the HMAC-SHA256 signature of the data packet to confirm that it originates from a legitimate edge acquisition node and has not been tampered with. Subsequently, an independent time calibration model is established for each node. The server periodically sends NTP requests to each node every 10 minutes, and the nodes return their local timestamps. The server records the time of reception. Collect at least 5 sets ( , After sampling, an overdetermined system of equations is constructed, and the coefficients a and b in the model are calculated using the least squares method to fit the time offset. Subsequently, when a data packet is received from this node, this model is used to timestamp the packet. The time is converted to a unified time base for the server. This method effectively compensates for crystal oscillator drift and network latency jitter, resulting in a system-wide time synchronization accuracy better than 0.5 milliseconds.

[0040] In step 9 of the aforementioned wireless communication data acquisition method, a device data index is established, using the device's unique identifier as the index key to store its historical operational data sequence. The data storage period is no less than 365 days, supporting multi-dimensional queries and retrieval by time range, process stage, and device type. Specifically, the device's unique identifier is generated by concatenating the device type code, serial number, and deployment location code, with a total length of 32 characters. Base32 encoding is used, and the index uses a columnar storage structure to optimize query efficiency, with a single complex query response time of less than 200 milliseconds. The device's unique identifier generation rules are as follows: the first 8 digits are the device type code, the middle 16 digits are the Base32 encoding of the manufacturer's serial number, and the last 8 digits are the Base32 representation of the workshop-production line-workstation three-level location code. The index is built based on the Apache Parquet columnar storage format, partitioned by device ID, with each partition further divided into buckets by date. When data is written, the JSON data packet is parsed into structured records, and each field is stored in an independent column file. The query engine utilizes predicate pushdown and column pruning techniques to read only the relevant column data, significantly reducing I / O overhead. For complex queries such as "querying all records of CNC equipment with spindle speeds greater than 1000 rpm in the past 7 days", the system can return results within 200 milliseconds.

[0041] In step 10 of the aforementioned wireless communication data acquisition method, an electromagnetic interference suppression module is integrated into the edge acquisition node. This module monitors the spectrum occupancy of the 2.4GHz band in real time, employs an adaptive notch filter to eliminate the influence of fixed-frequency interference sources, with a notch depth of no less than 25dB and a bandwidth adjustment resolution of 100kHz. Specifically, the center frequency of the adaptive notch filter is determined by the spectrum energy detection module, with a detection period of 100 milliseconds, an 8th-order filter, and coefficient updates using the least mean square algorithm, with a convergence factor set to 0.01. The electromagnetic interference suppression module includes a dedicated spectrum sensing coprocessor that performs fast Fourier transform analysis on the 2.4GHz ISM band every 100 milliseconds, with a scan step size of 100kHz. When a frequency point is detected to have an energy continuously 10dB higher than the background noise for 3 cycles, it is identified as a fixed interference source, and its center frequency f0 is recorded. Subsequently, the 8th-order IIR notch filter is activated, with its transfer function zeros configured at f0 and its poles close to the unit circle to control the bandwidth. The filter coefficients are updated online using the least mean square algorithm, and the convergence factor is set to 0.01. This ensures rapid frequency drift tracking even in non-stationary interference environments. Real-world testing shows that this module can suppress the effects of common interference sources such as Bluetooth and microwave ovens by more than 25dB, significantly improving the stability of the 2.4GHz link.

[0042] In addition, a lightweight containerized data preprocessing engine runs on the edge acquisition nodes. This engine loads a pre-trained anomaly detection model. The model input is a sequence of sensor data from 128 consecutive time steps, and the output is a device operating status score, ranging from 0 to 1, with a threshold set to 0.85. Specifically, the anomaly detection model is built on a Long Short-Term Memory (LSTM) network, containing two hidden layers with 64 neurons per layer, using the tanh activation function, and employing the sigmoid function in the output layer. The model training dataset contains no fewer than 100,000 samples of normal and abnormal operating conditions. The data preprocessing engine is deployed as a Docker container, loading an ONNX format LSTM model upon startup. The engine continuously caches sensor data from the most recent 128 time steps, forming... The input matrix is ​​3D. The time step is 100 milliseconds per step, totaling 12.8 seconds; N represents the number of sensors. The input, after normalization, is fed into an LSTM network, which outputs a score between 0 and 1, reflecting the degree of anomaly in the device's current operating status. When the score exceeds the 0.85 threshold five consecutive times, the engine generates an anomaly alarm event, attaches a fragment of the original data, and uploads it to the central scheduling server via a high-priority channel. This edge intelligence mechanism filters over 90% of routine data locally, uploading only critical anomaly information, significantly reducing uplink bandwidth pressure.

[0043] In addition, a dynamic scheduling mechanism for wireless channel occupancy is established. Each edge acquisition node performs carrier sensing before transmitting data, with the idle waiting time randomly set between 10 and 30 milliseconds. Burst data transmission adopts a fragmentation mechanism, with a maximum frame length of 1024 bytes. Specifically, each data frame in the fragmentation mechanism includes a frame sequence number, the total number of fragments, and the current fragment index. If the receiver does not collect all fragments within 100 milliseconds, it triggers a retransmission request. The retransmission request is sent through a dedicated control channel with a bandwidth of 200kHz. The edge acquisition nodes follow an improved CSMA / CA protocol: they listen to the channel before transmission; if busy, they back off; if idle, they generate a random backoff time between 10 and 30 milliseconds, listen again after the period expires, and transmit if still idle. For burst data exceeding 1024 bytes, it is automatically divided into multiple fragments. Each fragment header contains a 32-bit frame sequence number, an 8-bit total number of fragments, and an 8-bit current fragment index, where the frame sequence number uniquely identifies a frame. The receiving end starts a 100-millisecond timer. If all fragments are received within the time limit, they are reassembled and delivered. If the timeout occurs and all fragments are not received, a NACK retransmission request is sent through a dedicated 200kHz control channel, specifying the index of the missing fragment. This mechanism ensures reliable transmission of large blocks of data while avoiding long frames monopolizing the channel, thus improving channel utilization.

[0044] Furthermore, the central dispatch server is configured with a dual-machine hot standby architecture. The primary and standby servers synchronize status information via a dedicated heartbeat link with a 1-second heartbeat cycle and a failover time of less than 3 seconds. Data synchronization employs incremental log replication, with a latency not exceeding 100 milliseconds. The primary and standby servers are directly connected via gigabit Ethernet, running Keepalived and DRBD software stacks. Keepalived sends a VRRP heartbeat packet every 1 second to check the primary server's liveness; DRBD synchronizes data writing in real-time at the block device layer, using protocol C synchronous replication to ensure strong consistency between primary and standby data. When the primary server fails, the standby server takes over the virtual IP address within 3 seconds and continues to provide services. Application layer data is incrementally synchronized through a WAL (Write-Ahead Logging) mechanism. Each write operation is first logged and then asynchronously replicated to the standby machine, with end-to-end latency controlled within 100 milliseconds, minimizing data loss during failover.

[0045] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A wireless communication data acquisition method, characterized in that, The specific steps include the following: Step 1: Deploy edge acquisition nodes with multi-channel wireless communication capabilities in the workshop. Each edge acquisition node is equipped with a wireless transceiver module and has a built-in embedded gateway unit with network address translation function. Step 2: Divide each edge acquisition node into logical subnets and assign an independent virtual local area network identifier to each edge acquisition node; Step 3: Automatically assign IP addresses within each logical subnet through the Dynamic Host Configuration Protocol (DHCP) service. When a device connects for the first time, the address allocation process is triggered. The edge acquisition node records the mapping relationship between the device's MAC address and the assigned IP address and establishes a local address resolution table. Step 4: Configure a unified data aggregation agent on the central dispatch server; Step 5: The edge acquisition node performs protocol parsing and format normalization on the acquired raw device data; Step 6: Perform wireless link quality assessment, with edge acquisition nodes periodically sending probe signals; Step 7: When the main link quality is detected to be continuously lower than the threshold, the multi-path redundancy transmission strategy is activated, and the same data packet is forwarded to the central scheduling server through two or more different edge acquisition nodes. The server performs data deduplication and reassembly based on the sequence number. Step 8: The central scheduling server performs integrity verification and time series alignment on the received data packets, verifies the authenticity of the data source, and synchronizes the clocks of each node in conjunction with the network time protocol. Step 9: Establish a device data index library, using the device's unique identifier as the index key, to store its historical operating data sequence; Step 10: Integrate an electromagnetic interference suppression module into the edge acquisition node. The module monitors the spectrum occupancy of the 2.4GHz band in real time and uses an adaptive notch filter to eliminate the influence of fixed frequency interference sources.

2. The wireless communication data acquisition method according to claim 1, characterized in that, In step 1: the edge acquisition node establishes a wired connection with surrounding industrial equipment via Ethernet or serial communication interface, collects equipment operation data in real time, and then uploads the aggregated data to the central control system or adjacent nodes via wireless link to realize remote data transmission and centralized monitoring; the maximum wireless coverage radius of the edge acquisition node is 50 meters, and the minimum effective communication distance in dense metal structure environments is ≥20 meters. The edge acquisition nodes support wireless cascading expansion, forming a maximum of 3 levels of relay network, and a single node can connect to ≥16 heterogeneous industrial devices at the same time.

3. The wireless communication data acquisition method according to claim 1, characterized in that, In step 2: the virtual LAN identifier is uniformly assigned by the central configuration management system, which manages the subnet status. Subnet creation and deletion operations are initiated through out-of-band management channels, and the configuration change response time is less than 500 milliseconds. For multiple industrial devices within the same edge acquisition node that use the same static IP due to manufacturer restrictions, multiple routing modules with NAT functionality are deployed inside the edge acquisition node to map each device to a different private IP address within the subnet to which the edge acquisition node belongs. By combining virtual LAN isolation with NAT translation within the edge acquisition node, the system avoids IP address conflicts at both the cross-edge acquisition node and intra-edge acquisition node levels.

4. The wireless communication data acquisition method according to claim 1, characterized in that, In step 3: the Dynamic Host Configuration Protocol service sets the address pool capacity to 250 available addresses, the address allocation adopts the first-match priority strategy, and the address conflict detection is achieved through gratuitous ARP packet detection.

5. The wireless communication data acquisition method according to claim 1, characterized in that, In step 4: the data aggregation agent listens to the encrypted data tunnels from each edge acquisition node; the encrypted data tunnels are established using the pre-shared key IPsec protocol, with a security association lifecycle of 3600 seconds, and key updates are negotiated using Diffie-Hellman Group 14 parameters; the IPsec tunnels use ESP encapsulation mode, with AES-256 encryption algorithm and HMAC-SHA1 authentication algorithm. The tunnel establishment process includes an IKEv2 negotiation phase, and identity authentication is completed based on a pre-installed device certificate, with a certificate validity period of 730 days; the central scheduling server is configured with a dual-machine hot standby architecture, with the primary and standby servers synchronizing status information via a dedicated heartbeat link. The heartbeat cycle is 1 second, the failover time is <3 seconds, and data synchronization uses incremental log replication with a latency of no more than 100 milliseconds.

6. The wireless communication data acquisition method according to claim 1, characterized in that, In step 5: the data frame structures of three mainstream industrial protocols, Modbus-TCP, EtherNet / IP and Profinet, are identified, and process variable values, device status codes and timestamp information are extracted and encapsulated into a unified JSON format data packet with a timestamp accuracy of 1 millisecond. The JSON format data packet contains device ID, acquisition time, data point list and verification field. Each entry in the data point list contains variable name, value and unit.

7. The wireless communication data acquisition method according to claim 1, characterized in that, In step 6: the signal strength, signal-to-noise ratio (SNR), and retransmission rate are measured, and a signal strength threshold is set. When the SNR is lower than the SNR threshold, a link optimization mechanism is activated. The optimization mechanism includes automatic switching of the operating frequency band, adjustment of the modulation order, and activation of forward error correction coding. The modulation order is dynamically adjusted according to the link quality.

8. The wireless communication data acquisition method according to claim 1, characterized in that: In step 7, the multi-path redundancy transmission strategy is controlled by a routing strategy table issued by the central scheduling server. The table entries include node priority, path weight, and failure detection cycle. The path weight is dynamically updated based on the historical transmission success rate. In step 9, the unique device identifier is generated by concatenating the device type code, serial number, and deployment location code. The index uses a columnar storage structure to optimize query efficiency. In step 10, the center frequency of the adaptive notch filter is determined by the spectrum energy detection module.

9. The wireless communication data acquisition method according to claim 1, characterized in that: The wireless communication data acquisition method also includes running a lightweight containerized data preprocessing engine on the edge acquisition node. The engine loads a pre-trained anomaly detection model, with sensor data sequences as input and device operating status scores as output.

10. The wireless communication data acquisition method according to claim 1, characterized in that: The wireless communication data acquisition method further includes establishing a dynamic scheduling mechanism for wireless channel occupancy. Each edge acquisition node performs carrier sensing before sending data, and burst data transmission adopts a fragmentation mechanism. In the fragmentation mechanism, each data frame contains a frame sequence number, a total number of fragments, and a current fragment index. If the receiving end fails to collect all fragments within 100 milliseconds, it triggers a retransmission request, which is sent through a dedicated control channel.

Images