A data transmission method for digital phased array

Abstract

The application discloses a data transmission method of a digital phased array, and particularly relates to the field of communication transmission, and is used for solving the problem of inconsistent configuration change effective boundary of multiple receiving ends under link disorder; a sending end encapsulates sampling data into a data frame with a write frame serial number and frame integrity check information and sends the data frame according to the frame serial number, and inserts a barrier frame with a write barrier serial number before configuration change; a receiving end records a barrier landing frame serial number, a continuous tail frame serial number and generates a neighborhood frame serial number event string after entering a barrier frozen state, and sends a barrier reply; a sending end forms a submission judgment according to the barrier reply, determines a unified submission frame serial number, and sends a submission frame with a write configuration version number and the unified submission frame serial number; and the receiving end writes a new configuration into a submission area, loads the new configuration when the data frame serial number reaches the unified submission frame serial number, and returns a submission reply.

Description

Technical Field

[0001] This invention relates to the field of communication transmission, and more specifically, to a data transmission method for a digital phased array. Background Technology

[0002] In a digital phased array data transmission system, the array is often composed of multiple distributed transceiver units. Each unit digitizes the received signal to form continuous sampled data. Simultaneously, the system needs to distribute control configuration information such as amplitude and phase compensation, beam weighting, and operating mode switching to each unit to ensure coordinated processing across channels within the same operating cycle. To meet distance, cabling, and real-time requirements, existing solutions typically use high-speed links such as fiber optic cables to aggregate the sampled data to a central processing device, or employ a hierarchical processing architecture where the central processing unit calculates compensation parameters and then transmits them back to the sub-units. Transmission stability and channel consistency are maintained through encapsulation, alignment, and synchronization techniques.

[0003] Existing technologies, such as CN109633568A "Design Method of All-Digital Array Radar Beamformer Based on Fiber Optic Interface," propose that amplitude and phase compensation be calculated centrally and transmitted back to the sub-units, which then complete the compensation and subsequent processing. However, in real-world engineering, a hidden problem still easily arises: although both sampling data and control configuration can be "transmitted," there is a lack of a mechanism at the transmission layer to firmly bind "when the configuration takes effect" with "which data segment it corresponds to." The root cause is that the two types of information often undergo different buffering, queuing, and forwarding paths within the link and device. The transmission system focuses more on whether the bits are correct and whether the link is stable, but does not make semantics such as configuration version, frame boundary, and effective time into constraints that can be mutually observed by both parties. As a result, the configuration may be loaded earlier in one unit and later in another, or the configuration may take effect on adjacent data segments. When the problem occurs, the link may appear normal on the surface and the device may continue to output processing results, but the coherence between the channels may have been quietly disrupted. This manifests as unstable beamforming and a greater likelihood of anomalies during mode switching. Moreover, since the transport layer does not leave clear semantic evidence, it is often difficult to quickly locate the real cause of "configuration and data misalignment" during troubleshooting, which affects the maintainability of the system and the reliability of engineering implementation.

[0004] To address the aforementioned problems, a technical solution is provided. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a data transmission method for a digital phased array. The transmitting end encapsulates sampled data into data frames containing frame sequence numbers and frame integrity verification information, and transmits them according to the frame sequence numbers. Before configuration changes, a barrier frame containing a barrier sequence number is inserted. After entering a barrier frozen state, the receiving end records the barrier landing frame sequence number and the sequence number of consecutive tail frames, generates a neighboring frame sequence event string, and sends a barrier receipt. Based on the barrier receipt, the transmitting end forms a submission determination and, after determining the unified submission frame sequence number, sends a submission frame containing the configuration version number and the unified submission frame sequence number. The receiving end writes the new configuration into the pending submission area and loads and activates it when the data frame sequence number reaches the unified submission frame sequence number, then sends back a submission receipt, thus solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] S1: The transmitting end continuously encapsulates the sampling data generated by the digital phased array into data frames, writes the frame sequence number into the data frames, and sends the data frames in order according to the frame sequence number.

[0008] S2: Before the configuration change, the sending end sends a barrier frame. The barrier frame is written with the barrier sequence number and is inserted into the data frame sequence. It is sent using the same transmission link as the data frames.

[0009] S3: After receiving the barrier frame, the receiving end stops the configuration loading action, records the barrier landing frame sequence number and the consecutive tail frame sequence number, writes the barrier sequence number, barrier landing frame sequence number, consecutive tail frame sequence number and neighboring frame sequence event string into the barrier receipt, and sends the barrier receipt to the sending end.

[0010] S4: After receiving the barrier receipt, the sending end forms a submission decision and determines the unified submission frame sequence number. Based on the submission decision, it chooses to send a submission frame with the configuration version number and unified submission frame sequence number written in it, or chooses to retransmit the barrier frame and collect the barrier receipt again.

[0011] S5: After receiving the submission frame, the receiving end writes the new configuration into the submission area. When the local data frame sequence number reaches the unified submission frame sequence number, it loads the new configuration into the submission area and takes effect from the corresponding data frame. It writes the configuration version number and the unified submission frame sequence number into the submission receipt and sends the submission receipt to the sending end.

[0012] Furthermore, the transmitting end encapsulates the continuous sampling data generated by the digital phased array into data frames according to a fixed frame structure, generates frame integrity verification information for each data frame, and in the generation process, the frame header and payload are accumulated by taking words with a fixed word width and then truncated by the word width, the accumulated result is then accumulated with the frame sequence number and truncated by the word width, and finally the result is inverted bit by bit to obtain the frame integrity verification information and written to the frame tail.

[0013] Furthermore, the sending end maintains the sequence number of the largest sent frame on the sending side, writes data frames into the sending queue in ascending order of frame sequence number, reads the sequence number of the data frame at the head of the sending queue and performs a continuity check with the sequence number of the largest sent frame. If the continuity check satisfies that the sequence number of the frame at the head of the queue is equal to the sequence number of the largest sent frame plus one, the link is injected and the sequence number of the largest sent frame is updated; if the continuity check does not satisfy, the link injection is paused and the data frame corresponding to the missing frame sequence number is reconstructed. The reconstructed data frame is written with the missing frame sequence number and the frame integrity verification information is rewritten.

[0014] Furthermore, before the configuration change, the sending end generates a barrier frame and shares the same sending queue and transmission link with the data frame. In the sending queue, continuous segment identification is performed to obtain the number of consecutive acknowledgment frames. The maximum sequence number of the sent frame is added to the number of consecutive acknowledgment frames and then one is added to obtain the target insertion frame sequence number. The barrier frame is inserted before the data frame whose frame sequence number is equal to the target insertion frame sequence number. The frame type identifier and barrier sequence number are written into the barrier frame header, and the target insertion frame sequence number is written into the barrier frame payload.

[0015] Furthermore, after the barrier frame is inserted, the sending end performs an adjacent frame consistency check. The adjacent frame consistency check includes a forward check and a backward check. The forward check determines that the frame sequence number of the data frame immediately preceding the barrier frame is equal to the target inserted frame sequence number minus one. The backward check determines that the frame sequence number of the data frame immediately following the barrier frame is equal to the target inserted frame sequence number. When both the forward check and the backward check are true, the sending end sends the data frame and the barrier frame in the order of the sending queue and updates the sequence number of the largest sent frame. When either the forward check or the backward check is false, the sending end cancels the barrier frame insertion position and re-executes the continuous segment identification.

[0016] Furthermore, the receiving end recalculates the frame integrity check information for each data frame and compares it with the frame tail. Data frames that match are written into the rearrangement buffer. The consecutive tail frame sequence number is used as the sequential output pointer. The expected frame sequence number is obtained by adding one to the consecutive tail frame sequence number. The data frame whose frame sequence number is equal to the expected frame sequence number is searched in the rearrangement buffer and output sequentially. The consecutive tail frame sequence number is updated to the expected frame sequence number. Only one copy of the data frame is retained for the same frame sequence number.

[0017] Furthermore, after the receiving end identifies the barrier frame, it enters the barrier frozen state and stops the configuration loading action. When the expected frame number cannot be matched, it reads the minimum frame number in the rearrangement buffer. When the minimum frame number in the rearrangement buffer is greater than the expected frame number, it records the barrier landing frame number and simultaneously records the consecutive tail frame number. It writes a skip number event or a backtracking event according to the size relationship between the arrived frame number and the expected frame number. It writes a repeat event according to the arrived frame number being less than or equal to the consecutive tail frame number or the existence of the same frame number in the rearrangement buffer. It generates a neighboring frame sequence event string and writes it together with the barrier number, the barrier landing frame number, and the consecutive tail frame number into the barrier receipt.

[0018] Furthermore, after receiving barrier receipts from multiple receivers, the sending end merges them into a barrier receipt set according to the barrier sequence number, and verifies the completeness of the barrier landing frame sequence number, consecutive tail frame sequence number, and neighboring frame sequence event string fields. The sending end obtains the landing offset frame value by subtracting the consecutive tail frame sequence number from the barrier landing frame sequence number, sorts the landing offset frame values, takes the convergence anchor point offset frame value, calculates the upper and lower expansion amounts of the anchor point, and forms the barrier reaching the convergence bandwidth.

[0019] Furthermore, the sending end parses the front, boundary, and back segments of the neighboring frame sequence event string, extracts the boundary disturbance intensity characterization value, statistically analyzes the stable state marker of the front segment and the tense state marker of the back segment, and generates the barrier neighboring frame sequence continuous pressure index. Based on the barrier reaching the convergence bandwidth and the barrier neighboring frame sequence continuous pressure index, a submission decision is formed. When the submission decision is a submission frame, the unified submission frame sequence number is determined by adding the maximum barrier landing point frame sequence number to the stable delay amount, written into the configuration version number, and the submission frame is sent. When the submission decision is a retransmission of the barrier frame, the barrier frame is retransmitted and the barrier receipt is collected.

[0020] Furthermore, the receiving end receives the submission frame, recalculates the frame integrity verification information and compares it to be consistent, parses the configuration version number and the unified submission frame sequence number to complete the rollback judgment and sequence boundary judgment, writes the new configuration into the pending submission area and generates the pending submission area self-verification value, and when the sequential output pointer outputs the frame sequence number equal to the unified submission frame sequence number, loads the new configuration in the pending submission area as the current configuration and makes it effective, and sends a submission receipt containing the configuration version number and the unified submission frame sequence number.

[0021] This invention discloses a data transmission method for a digital phased array, which relates to data security technology, and its technical effects and advantages are as follows:

[0022] 1. By continuously encapsulating data frames with frame sequence numbers and writing frame integrity verification information at the sending end, and cooperating with the sending queue to send data in ascending order of frame sequence numbers and the frame sequence number breakage repair mechanism, the transmission link obtains stable sequential semantics and error frame removal entry. The rearranged output of digital phased array sampled data at multiple receiving ends has consistent frame boundaries, reducing the impact of out-of-order arrival on the processing pipeline.

[0023] 2. By inserting a barrier frame with a written barrier sequence number before configuration changes and having the receiver enter the barrier frozen state, the receiver simultaneously writes the barrier landing frame sequence number, the consecutive tail frame sequence number, and the neighboring frame sequence event string into the barrier receipt. The transmitter calculates the barrier reaching convergence bandwidth and the barrier neighboring frame sequence continuity pressure index based on this and uses complementary quadrant decision to form a submission decision and a unified submission frame sequence number. The configuration handover boundary is upgraded from observation by a single receiver to a collaborative constraint of consistency and boundary health of multiple receivers, reducing cross-end handover drift.

[0024] 3. When the sending end determines to submit a frame, it encapsulates a submission frame containing the configuration version number and the unified submission frame sequence number. The receiving end writes the new configuration into the pending submission area and loads and activates it all at once when the local data frame sequence number reaches the unified submission frame sequence number, and sends back a submission receipt. The configuration activation is strictly aligned with the data frame boundary and can be verified in the submission receipt. The barrier receipt-driven decision mechanism and the loading mechanism of the pending submission area form a closed loop, improving the determinism and traceability of configuration switching. Attached Figure Description

[0025] Figure 1 This is a flowchart illustrating a data transmission method for a digital phased array according to the present invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1: Figure 1 The present invention provides a data transmission method for a digital phased array, comprising:

[0028] S1: The transmitting end continuously encapsulates the sampled data generated by the digital phased array into data frames, writes the frame sequence number into the data frames, and sends the data frames in the order of the frame sequence number.

[0029] S2: Before the configuration change, the sending end sends a barrier frame. The barrier frame is written with the barrier sequence number and is inserted into the data frame sequence. It is sent using the same transmission link as the data frames.

[0030] S3: After receiving the barrier frame, the receiving end stops the configuration loading action, records the barrier landing frame sequence number and the consecutive tail frame sequence number, writes the barrier sequence number, barrier landing frame sequence number, consecutive tail frame sequence number and neighboring frame sequence event string into the barrier receipt, and sends the barrier receipt to the sending end.

[0031] S4: After receiving the barrier receipt, the sending end forms a submission decision and determines the unified submission frame sequence number. Based on the submission decision, it chooses to send a submission frame with the configuration version number and unified submission frame sequence number written in, or chooses to retransmit the barrier frame and collect the barrier receipt again.

[0032] S5: After receiving the submission frame, the receiving end writes the new configuration into the submission area. When the local data frame sequence number reaches the unified submission frame sequence number, it loads the new configuration into the submission area and takes effect from the corresponding data frame. It writes the configuration version number and the unified submission frame sequence number into the submission receipt and sends the submission receipt to the sending end.

[0033] The sampled data from a digital phased array is output as a continuous stream. Configuration changes need to fall on repeatably identifiable boundaries; otherwise, the receiver will find it difficult to correlate the configuration's effective point with each sampled data segment. If the transmitter simply pushes the sampled data directly to the link, the receiver can only guess the segmentation based on the arrival time. When encountering link jitter and buffer reordering, the segmentation judgment will drift. The transmitter first organizes the sampled data into data frames with stable boundaries at the link inlet. Simultaneously, it writes marker information describing the sequence and integrity within the data frames. This allows the receiver to locate the frame start and end points using clear rules and identify erroneous frames, thus providing an interpretable reference for barrier frame insertion and unified submission frame sequence number determination.

[0034] The specific implementation method of step S1 is as follows:

[0035] S101 Data Frame Boundary Establishment and Frame Sequence Number Writing.

[0036] Since continuous sampled data does not have natural frame boundaries, the sending end needs to divide the continuous sampled data into data blocks of fixed length and encapsulate them into data frames so that the data frames have a locatable start and end position in the link.

[0037] The transmitting end first constructs a frame header and reserves a frame sequence number field, then fills the payload area with a continuous sample of data, and finally constructs a frame trailer to form a complete data frame. The frame sequence number follows a monotonically increasing rule. Before generating the current data frame, the transmitting end reads the frame sequence number of the previous data frame, increments it by one to obtain the current frame sequence number, and then writes the current frame sequence number into the frame header of the current data frame. The increment operation is performed according to the bit width of the frame sequence number field. Natural wrapping within the bit width serves as the normal loop for the frame sequence number, and the monotonically increasing semantics remain valid within the loop domain even when wrapping occurs.

[0038] After a data frame is encapsulated and its sequence number is written, it enters the transmission queue. The transmitting end always arranges the transmission queue in ascending order of the frame sequence number. This allows the receiving end to use the frame sequence number to determine the order of data frames and stably advance the sequential output pointer and the sequence number of consecutive last frames, ensuring that data frame boundaries remain identifiable even under link jitter.

[0039] S102 Frame Integrity Verification Information Generation and Frame Tail Writing.

[0040] Link transmission may introduce bit flipping, truncation and splicing errors. The receiving end needs a frame integrity check information that can be recalculated and verified to confirm that the data frame content remains consistent during transmission, while avoiding the advancement of consecutive tail frame sequence numbers by erroneous frames.

[0041] When the sending end generates frame integrity verification information for each data frame, it adopts a two-stage cumulative verification, so that the verification value covers both the data content and is bound to the frame sequence number, reducing the risk of different frame contents being misjudged as the same frame.

[0042] The first cumulative check starts from the beginning of the frame header. The frame header and payload are divided into a string of words with a fixed word width. The sending end adds these words one by one in sequence. The accumulation adopts the modulo addition rule, which means that when the accumulation result exceeds the maximum range that the word width can represent, only the low-order part is retained and the high-order carry is discarded to obtain the intermediate check value one.

[0043] The second cumulative check adds the intermediate check value one and the frame sequence number field with the same word width, and then uses the same modulo addition rule to obtain the intermediate check value two. Then, a bit-by-bit inversion operation is performed on the intermediate check value two. The bit-by-bit inversion operation means that the zero and one of each bit within the word width are swapped to obtain the frame integrity check information.

[0044] The sending end writes frame integrity check information to the frame tail. The receiving end recalculates intermediate check value one and intermediate check value two for the received data frame using the same word width and word extraction order, inverts them to generate a recalculated check value, and then compares the recalculated check value bit by bit with the frame integrity check information carried in the frame tail. If they match, the data frame is allowed to enter the rearrangement buffer and participate in the sequential advance of the last frame number; if they do not match, the data frame is discarded and the error frame status is recorded. This allows the receiving end to uniformly intercept data frame boundary errors and content errors, reducing the probability of error frames entering the sequential output pointer, and making it easier to suppress backoff events and repetition events caused by error frames in the neighboring frame sequence event string.

[0045] S103 transmission queue sorting, frame sequence number breakage identification and completion strategy.

[0046] The insertion of barrier frames and the determination of unified submission frame sequence numbers depend on the continuity of the data frame sequence. If a frame sequence number break occurs at the sending end, the receiver's rearrangement buffer will continuously accumulate, hindering the determination of the barrier landing frame sequence number. The barrier landing frame sequence number in the barrier receipt will also separate from the consecutive tail frame sequence number. The sending end maintains the maximum sent frame sequence number, which represents the sequence number of the most recently successfully injected data frame into the link. When preparing to inject into the link, the sending end reads the sequence number of the data frame at the head of the sending queue and performs a continuity check between the head frame sequence number and the maximum sent frame sequence number. The continuity check rule is that the head frame sequence number equals the maximum sent frame sequence number plus one. The operation is performed within the frame sequence number bit width and allows for natural wrap-around. When the continuity check is successful, the sending end injects the head data frame into the link and updates the maximum sent frame sequence number to the newly injected data frame sequence number. When continuity determination fails, the transmitter pauses injection and locates the range of missing frame sequence numbers. The starting point of the missing range is the sequence number of the largest transmitted frame plus one, and the ending point is the sequence number of the first frame in the queue minus one. The transmitter reconstructs the data frames corresponding to the missing frame sequence numbers one by one according to the missing range. The reconstruction process involves resegmenting the sampled data and re-encapsulating the data frames. The missing frame sequence number is written into the frame header, the sampled data segment corresponding to the missing frame sequence number is filled into the payload, and the frame integrity verification information is regenerated at the frame tail, using the same method as the aforementioned frame integrity verification information generation process. The reconstructed data frames are inserted into the head of the transmission queue in ascending order of frame sequence number. The transmitter performs continuity determination again until continuity determination is successful before resuming injection. This processing ensures that the link ingress maintains a continuously increasing sequence of data frames, the receiver's rearrangement buffer does not need to rely on backtracking for a long time, the advancement of the continuous tail frame sequence number is smoother, and the insertion position of the barrier frame is more controllable.

[0047] After the sending end completes data frame encapsulation, frame sequence number writing, frame integrity verification information writing, and frame sequence number continuity assurance, a continuous data frame sequence with stable boundaries, stable order semantics, and verifiable integrity is formed in the link. The receiving end can use the frame header and frame tail fields to complete frame boundary confirmation and erroneous frame removal, and the sequential output pointer and consecutive tail frame sequence numbers can be advanced under stable conditions. When the barrier frame enters the same transmission link in subsequent steps, the records of the barrier landing frame sequence number and consecutive tail frame sequence numbers are more interpretable, and the neighboring frame sequence event string of the barrier receipt is more likely to reflect real link disturbances rather than encapsulation defects.

[0048] In step S1, the data frames of the digital phased array already possess frame sequence numbers and frame integrity verification information. The transmission queue has arranged the data frames in ascending order of frame sequence numbers and constrained continuity with the largest transmitted frame sequence number. When a configuration change occurs, directly inserting a configuration frame can easily lead to misalignment with the data frame boundary. The receiver may misjudge the effective boundary of the configuration when the pressure on the rearrangement buffer increases. Barrier frames are used to bind configuration change actions to a specific position in the data frame sequence. However, if the barrier frame insertion position falls near a frame sequence number break, the barrier landing frame sequence number will be dragged, and the neighboring frame sequence event string will exhibit dense skipped number events and backtracking events. The barrier receipt will be difficult to reflect the true link status, and the unified submission frame sequence number will also lose its interpretability.

[0049] The specific implementation method of step S2 is as follows:

[0050] S201 barrier frame insertion position determination and alignment calculation process.

[0051] Continuously sampled data has already formed frame sequence order semantics in the data frame. If configuration changes are directly mixed with the data frame, the receiving end may easily place the configuration action on the unstable frame boundary when the reflow occurs in the reordering buffer.

[0052] The sending end first performs continuous segment identification on the transmission queue. This identification starts by reading the frame sequence number from the head of the transmission queue, then reads the frame sequence number of the next data frame and checks if the next frame's sequence number equals the previous frame's sequence number plus one. If the check is true, the number of consecutive acknowledgment frames is incremented, and the judgment window is moved to the next pair of adjacent data frames to continue the judgment. If the check is false, the continuous segment identification ends. The sending end combines the number of consecutive acknowledgment frames with the maximum transmitted frame sequence number to obtain the target insertion frame sequence number. The calculation process involves first adding the maximum transmitted frame sequence number to the number of consecutive acknowledgment frames, then incrementing the result by one. The target insertion frame sequence number is the result of this increment operation.

[0053] The transmitting end searches the transmission queue for the position of the data frame whose frame number equals the target insertion frame number, and inserts the barrier frame before this position, so that the frame number of the data frame immediately following the barrier frame equals the target insertion frame number. When the barrier frame falls after a consecutive frame sequence interval, the receiving end experiences fewer segment jumps when advancing the consecutive tail frame numbers, the barrier landing frame number is more likely to maintain a stable difference with the consecutive tail frame numbers, and the neighboring frame sequence event sequence is better able to reflect the link disturbance itself.

[0054] S202 barrier frame construction and barrier number writing operation process.

[0055] The barrier frame serves as the freeze trigger. The receiving end needs to directly identify the barrier frame type and bind the barrier frame to a single configuration change without parsing the payload content. When constructing a barrier frame, the sending end first generates the barrier frame header and writes the frame type identifier. The frame type identifier uses a fixed encoding and is consistent between the sending and receiving ends. The sending end then generates the barrier sequence number. The barrier sequence number generation process involves reading the barrier sequence number used in the previous round of barrier frames, then performing an increment operation on the read barrier sequence number, and taking the result of the increment operation. The barrier sequence number is then written to the corresponding field in the barrier frame header. The sending end writes barrier association information into the barrier frame payload. The barrier association information includes at least the target inserted frame sequence number. The writing order of the barrier association information is: first write the target inserted frame sequence number, then write a duplicate copy of the barrier sequence number. The duplicate copy is used by the receiving end for cross-consistency verification. The sending end generates frame integrity verification information for the barrier frame and writes it to the barrier frame trailer. The frame integrity verification information generation process follows the same rules as the frame integrity verification information generation for data frames. With the barrier frame having a frame type identifier, barrier number, and target inserted frame number, the receiving end can identify the barrier frame more directly. The barrier number in the barrier receipt can strictly correspond to a barrier freeze and a configuration change. When the barrier receipt is grouped by barrier number, cross-wheel information will not be mixed in.

[0056] The process of enqueuing and sending the S203 barrier frame and verifying the consistency with adjacent frames.

[0057] After the barrier frame is inserted, it is still necessary to verify whether the frame sequence number relationship between the barrier frame and the adjacent data frame satisfies the insertion semantics. Otherwise, the receiver may map the barrier frame to an incorrect boundary when recording the barrier landing frame sequence number. After inserting the barrier frame into the transmission queue, the transmitter performs an adjacent frame consistency check, which consists of forward and backward checks. The forward check reads the frame sequence number of the data frame immediately preceding the barrier frame, and then performs a decrement operation on the target inserted frame sequence number. The forward check determines whether the read frame sequence number is equal to the decrement result. The backward check reads the frame sequence number of the data frame immediately following the barrier frame, and then determines whether the read frame sequence number is equal to the target inserted frame sequence number. When both the forward and backward checks are true, the transmitter injects the data frame into the link frame by frame according to the transmission queue order, and updates the maximum transmitted frame sequence number to the newly injected data frame sequence number after each injection. The barrier frame and the data frame arrive at the receiver along the same link path. If either the forward check or the backward check fails, the sending end cancels the barrier frame insertion position and re-executes the continuous segment identification and target insertion frame sequence number calculation until both the forward check and the backward check are successful before sending the barrier frame.

[0058] After the adjacent frame consistency check is successful, the barrier frame forms a stable adjacent frame order relationship with the data frame in the link. When the receiver enters the barrier frozen state, it will not generate a number skipping event and backoff event due to the insertion point drift. The convergence pattern of the barrier reaching the convergence bandwidth is clearer.

[0059] The barrier frame injection before configuration changes is completed through three steps: insertion position alignment, barrier sequence number increment, and adjacent frame consistency verification. The barrier frame is located at a clearly defined boundary of a continuous frame sequence interval in the transmission queue. When the receiver enters the barrier freeze state in step S3, the frame type identifier carried in the barrier frame header triggers the freeze action. The barrier sequence number corresponds one-to-one with the freeze action and a single configuration change, and the target inserted frame sequence number provides a consistency reference for the barrier landing point frame sequence number record. After the barrier frame is in a stable adjacent frame relationship, the neighboring frame sequence event sequence better reflects the link disturbance and rearrangement characteristics, and the calculation results of the barrier reaching convergence bandwidth and the barrier neighboring frame sequence continuity pressure index are also more interpretable.

[0060] After the sending end completes data frame encapsulation and barrier frame injection, the link simultaneously contains a sequence of data frames arranged by frame sequence number and a barrier frame used to trigger freezing. The receiving end needs to restore out-of-order data frames to sequential output under the constraint of frame integrity verification information, and then pause configuration loading and determine the barrier landing frame sequence number when the barrier frame arrives. If the barrier landing frame sequence number is recorded incorrectly, the neighboring frame sequence event string in the barrier receipt will lose its directionality. However, the barrier receipt must simultaneously reflect the sequential output status and the arrival disturbance pattern to support the sending end in forming a submission decision and unifying the submission frame sequence number.

[0061] The specific implementation method of step S3 is as follows:

[0062] S301 sequential output pointer and rearrangement buffer maintenance.

[0063] When data frames arrive at the receiver, out-of-order and gapped frames may occur. The sequential output pointer needs to advance without losing the semantic meaning of the frame sequence number. The reordering buffer needs to absorb out-of-order frames and wait for the gaps to close. For each data frame, the receiver first recalculates the frame integrity check information and compares it bit-by-bit with the frame integrity check information carried at the end of the frame. Data frames that match are written to the reordering buffer; data frames that do not match are discarded and marked as erroneous frames. The receiver uses the sequence number of the consecutive tail frames as the sequential output pointer. The initial value of the consecutive tail frame sequence number is taken from the synchronization value at the receiver's startup or the recorded value from the last stable output. Each time the receiver prepares to advance the sequential output pointer, it performs an increment operation on the consecutive tail frame sequence number to obtain the desired frame sequence number. The increment operation is performed within the bit width of the frame sequence number field and allows natural wrap-around. The receiver searches for a data frame in the reordering buffer whose frame sequence number equals the desired frame sequence number. If found, the corresponding data frame is handed over to the processing pipeline, and the consecutive tail frame sequence number is updated to the desired frame sequence number. If not found, the consecutive tail frame sequence number remains unchanged, and subsequent data frames are received. The receiving end uses the same frame sequence number index to limit the number of data frames with the same frame sequence number to only one copy in the rearrangement buffer, so as to avoid the rearrangement buffer from being bloated due to repeated writing of the same frame sequence number. The sequential output pointer advance process keeps the frame sequence number continuous and the semantics clear.

[0064] S302 Barrier Freeze State Entry and Configuration Load Stop Rule.

[0065] Barrier frames are used to separate configuration changes from the data frame sequence. The receiving end must immediately stop configuration loading upon the arrival of a barrier frame to prevent configuration switching from falling into the out-of-order reordering period. The receiving end parses the frame type identifier of each received frame. When the frame type identifier indicates a barrier frame, the receiving end enters a barrier-frozen state and saves the barrier sequence number. In the barrier-frozen state, configuration loading remains suspended. The stopping rule is that new configurations stored in the pending submission area are not allowed to be loaded into the current configuration; the current configuration remains unchanged. In the barrier-frozen state, frame integrity verification information recalculation and comparison can continue, writing to the reordering buffer is allowed, and the sequence number of consecutive last frames can continue to advance according to the expected frame sequence number and output data frames to the processing pipeline. The barrier-frozen state separates configuration switching from data frame reordering, preventing configuration switching from being affected by transient fluctuations in the reordering buffer.

[0066] S303 Barrier Landing Point Confirmation Rules and Barrier Landing Point Frame Sequence Record.

[0067] The barrier landing frame sequence number needs to reflect the position of the sequential output breakpoint near the arrival of the barrier frame. The breakpoint is indicated when the expected frame sequence number cannot be found in the rearrangement buffer, and the rearrangement buffer has already reached a later data frame. While the barrier is frozen, the receiver advances the sequence number of consecutive tail frames according to the rule in step S301, continuing this advance until the rearrangement buffer can no longer find the expected frame sequence number. When the expected frame sequence number cannot be found, the receiver checks if the rearrangement buffer is empty. If the rearrangement buffer is empty, it continues to receive data frames and repeats the check. When the rearrangement buffer is not empty, the receiver reads the sequence numbers of all data frames in the rearrangement buffer and takes the minimum value as the minimum sequence number of the rearrangement buffer. The receiver compares the minimum sequence number of the rearrangement buffer with the expected frame number. If the minimum sequence number of the rearrangement buffer is greater than the expected frame number, the receiver records the minimum sequence number of the rearrangement buffer as the barrier landing frame sequence number, and also records the current consecutive tail frame sequence numbers. The barrier landing frame number and the consecutive tail frame number come from the same observation instant. The barrier landing frame number indicates the first frame number to arrive after the gap, and the consecutive tail frame number indicates the last consecutive output frame number before the gap. The difference between the two directly reflects the gap span and is consistent with the definition of the landing offset frame value at the transmitting end.

[0068] S304 Neighborhood Frame Sequence Event String Generation and Event Judgment Operation Process.

[0069] The neighborhood frame sequence event string needs to encode the abnormal patterns of the arrival process near the barrier into a finite event set. The sending end can only distinguish between gap closure and out-of-order rollback based on this event set. During the barrier freeze state, the receiving end reads the arrival frame sequence number of each data frame that has passed the frame integrity check information and calculates the expected frame sequence number according to the rule in step S301. The receiving end compares the arrival frame sequence number with the expected frame sequence number. If the arrival frame sequence number is greater than the expected frame sequence number, a skip event is written, and the gap length is obtained by subtracting the expected frame sequence number from the arrival frame sequence number. The gap length serves as the basis for continuous marking of skip events. If the arrival frame sequence number is less than the expected frame sequence number, a backtrack event is written, and the backtrack span is obtained by subtracting the arrival frame sequence number from the expected frame sequence number. The backtrack span serves as the basis for continuous marking of backtrack events.

[0070] Duplicate event detection employs a dual-condition approach: a duplicate event is written when the arriving frame number is less than or equal to the consecutive last frame number; and a duplicate event is written when the arriving frame number is greater than the consecutive last frame number and the reordering buffer already contains the same arriving frame number. The receiver temporarily stores each written event along with its arriving frame number. After the barrier landing point frame number is recorded, a segment classification operation is performed. This operation subtracts the barrier landing point frame number from the arriving frame number to obtain a relative position marker. Events with a relative position marker less than zero are assigned to the preceding segment, events with a relative position marker equal to zero are assigned to the boundary segment, and events with a relative position marker greater than zero are assigned to the following segment. The receiver inserts fixed separator markers between the preceding, boundary, and following segments to form a neighborhood frame sequence event string. This neighborhood frame sequence event string retains both the event type and the relative positional distribution of events near the barrier landing point, allowing the transmitter to distinguish the disturbance characteristics before and after the barrier.

[0071] S305 Barrier Receipt Field Population and Sending Constraints.

[0072] The barrier receipt needs to provide the sender with the barrier freeze round identifier, boundary location result, and arrival perturbation code. Missing fields will cause the sender to lose its basis for packet calculation and quadrant determination. When generating the barrier receipt, the receiver writes the barrier sequence number, the barrier landing frame sequence number, the consecutive tail frame sequence number, and the neighboring frame sequence event string into the fixed fields of the barrier receipt. The receiver does not inject links before the barrier receipt is written; after writing, it injects links according to the sending queue order to avoid ambiguity in the sender's parsing caused by partial updates to the barrier receipt fields. After sending the barrier receipt, the receiver maintains the barrier freeze state, the configuration loading action remains stopped, and the sequential output pointer and rearrangement buffer maintenance continue. The consecutive tail frame sequence number reflected in the barrier receipt maintains the same time source as the barrier landing frame sequence number, and the event segment reflected in the neighboring frame sequence event string is consistent with the barrier landing frame sequence number.

[0073] Within the barrier frozen state, the receiving end completes the frame integrity verification information filtering and rearrangement buffer restoration for data frames. The sequence number of consecutive tail frames advances according to the expected frame sequence number rules. The barrier landing frame sequence number is recorded when it matches the minimum frame sequence number in the rearrangement buffer plus one from the consecutive tail frame sequence number. Neighborhood frame sequence event strings are written and distributed to the front, boundary, and rear segments according to three types of events: skip events, backtracking events, and repetition events. The barrier receipt carries the barrier sequence number, barrier landing frame sequence number, consecutive tail frame sequence number, and neighborhood frame sequence event string to the sending end. The sending end obtains a consistent arrival disturbance description and sequential advancement status within the same barrier sequence number group.

[0074] Data frames have been continuously injected into the link at the sending end according to their frame sequence numbers. Barrier frames were inserted near the target insertion frame sequence number before the configuration change. At the receiving end, under barrier-frozen state, the barrier sequence number, barrier landing frame sequence number, consecutive tail frame sequence number, and neighboring frame sequence event string have been written into the barrier receipt and sent back to the sending end. The sending end needs to unify multiple barrier receipts under the same barrier sequence number semantics to form a submission decision and determine a unified submission frame sequence number. However, the rearrangement buffer states of different receiving ends are not consistent. Therefore, the boundary positioning of the barrier receipt and perturbation coding must be used to jointly constrain the decision, avoiding binding the configuration version number to unstable frame boundaries.

[0075] The specific implementation method of step S4 is as follows:

[0076] S401 Barrier Receipt Grouping and Field Consistency Verification.

[0077] Barrier receipts come from multiple receivers. Barrier sequence numbers are used to distinguish between different rounds of barrier freezing; mixing barrier sequence numbers will overlap the boundary locations of different rounds. After receiving the barrier receipts, the sending end reads the barrier sequence numbers one by one, grouping receipts with the same sequence number into a barrier receipt set. Within the barrier receipt set, the completeness of each field is verified. These fields include the barrier sequence number, barrier landing frame sequence number, consecutive tail frame sequence number, and neighboring frame sequence event string. Barrier receipts missing any field are removed from the set. The sending end then verifies the field relationships: the barrier landing frame sequence number must not be earlier than the consecutive tail frame sequence number, and the neighboring frame sequence event string must contain front, boundary, and rear segment separators. Barrier receipts that do not meet these relationships are removed from the set. After grouping and verification, the barrier landing frame sequence number and consecutive tail frame sequence number within the barrier receipt set are comparable within the same round, and the segment positions of the neighboring frame sequence event string are also comparable within the same round, resulting in more focused input for judgment.

[0078] S402 landing point offset frame value calculation and sorting preparation.

[0079] The barrier landing frame number and the consecutive tail frame number both belong to the frame sequence domain. Subtracting them within the frame sequence domain maintains dimensional consistency and directly expresses the gap crossing. The transmitter reads the barrier landing frame number and the consecutive tail frame number for each barrier receipt in the barrier receipt set. The transmitter subtracts the consecutive tail frame number from the barrier landing frame number to obtain the landing offset frame value. All landing offset frame values ​​are summarized into a single column and sorted in ascending order. The sorting result is used for subsequent convergence anchor point offset frame values ​​and quantile span values. The landing offset frame value maintains the frame dimension, allowing the transmitter to directly compare the boundary closure states of different receivers, and making it easier to trace the source of extreme values ​​back to specific barrier receipts.

[0080] S403 barrier convergence bandwidth calculation and narrowband / broadband state classification.

[0081] Multi-receiver boundary consistency requires simultaneously suppressing the impact of extreme values ​​while preserving the overall unfolding shape. From the sorted landing point offset frame values, the middle landing point offset frame value is selected as the convergence anchor point offset frame value. The middle position follows a rounding-up order rule; if the number of barrier receipt sets is even, the later middle position is used. The upper-side unfolding amount of the anchor point is calculated as the maximum difference between all landing point offset frame values ​​and the convergence anchor point offset frame value. The lower-side unfolding amount of the anchor point is calculated as the maximum difference between the convergence anchor point offset frame value and all landing point offset frame values. The larger of the upper-side and lower-side unfolding amounts is taken as the half-bandwidth. Doubling this half-bandwidth yields the barrier convergence bandwidth, which maintains the frame dimension.

[0082] Narrowband and wideband states are categorized into two cases based on the number of barrier receipt sets. When the number of barrier receipt sets is less than four, the sender does not use the quartile position but instead uses symmetry discrimination. If the difference between the expansion amount above and below the anchor point is not significant, it enters the narrowband state; if the difference is significant, it enters the wideband state. When the number of barrier receipt sets is not less than four, the frame values ​​of the landing point offsets of the quartile position and the third / fourth quartile position are subtracted to obtain the quartile span. When the barrier reaches a convergence bandwidth that is no more than twice the quartile span, it enters the narrowband state; when the barrier reaches a convergence bandwidth that is more than twice the quartile span, it enters the wideband state. The classification rules are derived from the inherent characteristics of the same round of barrier receipt sets. The influence of a single extreme receipt on the classification result decreases, and boundary concentration and boundary dispersion are more easily and stably distinguished.

[0083] S404 Neighborhood Frame Sequence Event String Parsing and Boundary Perturbation Intensity Characterization Value Extraction.

[0084] The neighborhood frame sequence event string records the arrival disturbance pattern near the barrier landing point. The boundary segment reflects the intensity of the sudden disturbance, and the subsequent segment reflects the duration of the disturbance. These two need to be extracted separately and then synthesized into a pressure characterization. Each neighborhood frame sequence event string within the barrier receipt set is split into a front segment, a boundary segment, and a subsequent segment according to a separator mark. Within the boundary segment, the event type and the event continuity mark length are read item by item. The event type is limited to skip events, backtracking events, and repeating events. The event continuity mark length indicates the number of consecutive occurrences of the same type of event. The highest disturbance level of the boundary segment is determined according to a fixed priority order, with skip events higher than backtracking events, and backtracking events higher than repeating events. Within the boundary segment, a set of events corresponding to the highest disturbance level of the boundary segment for each event type is selected. The maximum value of the event continuity mark length in this set is taken as the boundary persistence pattern. The highest disturbance level of the boundary segment and the boundary persistence pattern are recorded together as the boundary disturbance intensity characterization value. The boundary disturbance intensity characterization value retains both the intensity and persistence dimensions; the instantaneous spikes and persistent jitter of the boundary segment are no longer compressed into a single quantity in the recording.

[0085] Calculation of stable morphological markers in the front section and tense morphological markers in the rear section of S405.

[0086] The boundary segment only describes the instantaneous disturbances near the barrier's landing point. The preceding and following segments reflect the density and duration of disturbances, and the number of event clusters can distinguish between discrete and continuous events. In the preceding event sequence, skipped events and backtracking events are considered disturbance events, while repetitive events are considered non-disturbance events. The transmitter scans along the event sequence, and when a disturbance event occurs at a scanned position and no disturbance event occurred at the previous position, this position is recorded as the start point of the event cluster. Counting all the start points of the preceding event clusters yields the preceding stable state marker. The same rules are applied to the following event sequence, and counting all the start points of the following event clusters yields the following tense state marker. The preceding stable state marker and the following tense state marker are dimensionless counts; the count only reflects the number of times the disturbance clusters occur and is not affected by the length of a single event's continuous marker. Short disturbances and continuous disturbances are measured separately at the counting level, and the continuity of the following segment is easier to observe independently.

[0087] S406 barrier neighborhood frame sequence continuous pressure index classification.

[0088] The barrier neighborhood frame sequence continuity pressure index needs to incorporate the highest disturbance level of the boundary segment, the boundary persistence pattern, the preceding stable pattern marker, and the following tense pattern marker to prevent repeated events from triggering a high-pressure level when there is no subsequent continuation. For each barrier receipt, the initial classification direction is first determined based on the highest disturbance level of the boundary segment. When the highest disturbance level of the boundary segment is a skip event, the barrier neighborhood frame sequence continuity pressure index is set to a high-pressure level. When the highest disturbance level of the boundary segment is a backtracking event, it is checked whether the boundary persistence pattern exceeds a single occurrence and whether the following tense pattern marker exceeds the preceding stable pattern marker. If both conditions are met, it is set to a high-pressure level; if neither condition is met, it is set to a medium-pressure level. When the highest disturbance level of the boundary segment is a repeated event, the sender checks whether the following tense pattern marker is zero. If the following tense pattern marker is zero, it is set to a low-pressure level; if the following tense pattern marker is not zero, it is set to a medium-pressure level. The grading process separates boundary bursts from subsequent continuations. Repeated events will not be elevated to a high-pressure level in repeated arrival scenarios after sequential closure, and the pressure level is closer to the actual pressure of the rearrangement buffer.

[0089] S407 Complementary Quadrant Decision and Unified Submission Frame Sequence Number Determination.

[0090] The barrier convergence bandwidth describes the consistency of multiple receiver boundaries, while the barrier neighborhood frame sequence continuity pressure index describes the order health near the boundary. These two metrics require complementary decision-making to avoid missed judgments caused by considering only consistency or only pressure. Narrowband and wideband states are used as consistency dimensions, while low-pressure, medium-pressure, and high-pressure levels are used as health dimensions. A complementary quadrant rule is used when forming a submission decision: in narrowband state and with a pressure level no higher than medium-pressure, the submission decision points to sending a submission frame; in wideband state or with a high-pressure level, the submission decision points to retransmitting the barrier frame and recollecting barrier receipts.

[0091] When the submission decision indicates sending a submission frame, a unified submission frame sequence number is determined. The maximum barrier landing frame sequence number is found in the barrier receipt set and used as the base frame sequence number. Each subsequent event sequence in the barrier receipt set is analyzed, and the maximum value of the consecutive marker length for skip events and the maximum value of the consecutive marker length for backoff events are found. The maximum of these two is used as the subsequent continuous perturbation length for this barrier receipt. If there are no skip events or backoff events in the subsequent segment, the subsequent continuous perturbation length is set to zero. The maximum value of all subsequent continuous perturbation lengths in the barrier receipt set is used as the stable delay. The base frame sequence number and the stable delay are added to obtain the unified submission frame sequence number. The unified submission frame sequence number maintains the frame order domain dimension, while the stable delay comes from the continuity of the subsequent perturbation. The unified submission frame sequence number is closer to the common stable interval of multiple receivers. When the submission decision indicates retransmission of the barrier frame, the unified submission frame sequence number is uncertain, the barrier frame is retransmitted, and the barrier receipts are collected again.

[0092] S408 submission frame encapsulation transmission and retransmission barrier frame processing.

[0093] When the commit decision indicates the need to send a commit frame, the commit frame is encapsulated, with the configuration version number and unified commit frame sequence number written into it. The commit frame is then injected into the same transmission link as the data frame and sent in the order of the sending queue. When the commit decision indicates the need to retransmit a barrier frame, a new barrier frame is generated and a new barrier sequence number is written into it. The target insertion frame sequence number is re-determined according to the alignment rule of the largest sent frame sequence number and inserted into the sending queue. After passing the adjacent frame consistency check, the barrier frame is sent and a new set of barrier acknowledgments is awaited. Both the commit frame and the barrier frame are transmitted on the same link. The receiving end can complete the freeze and load state transition without switching parsing channels. The sending end has a clear execution path for the commit decision, and the transmission rhythm of repeatedly sampled data will not be interrupted by the decision action.

[0094] After group verification, the barrier receipts form a barrier receipt set. The landing point offset frame value, convergence anchor point offset frame value, and barrier convergence bandwidth are calculated within the frame dimension. The neighboring frame sequence event string is decomposed into boundary disturbance intensity characterization value, front-segment stable state marker, and rear-segment tense state marker in the boundary segment and the subsequent segment, and synthesized into a barrier neighboring frame sequence continuous pressure index. The submission decision is output under the joint constraints of narrowband state and pressure index. The unified submission frame sequence number is determined using the maximum barrier landing point frame sequence number and the stable delay. The sender decides whether to send the submission frame or retransmit the barrier frame based on this. The configuration version number is bound to the interpretable frame boundary.

[0095] The barrier acknowledgment has been grouped and decided at the sending end, and the commit frame has been written with the configuration version number and unified commit frame sequence number and sent to the receiving end along the same transmission link. The receiving end continuously outputs data frames while the barrier is frozen, and the configuration loading action remains suspended. Upon arrival of the commit frame, the receiving end must confirm that the content of the commit frame is not affected by link disturbances, load the new configuration into the commit area, and strictly align the effective boundary of the new configuration to the data frame corresponding to the unified commit frame sequence number. However, out-of-order reordering and duplicate arrivals exist at the receiving end. If the loading timing is misjudged, the new configuration may fall on the boundaries of different data frames at different receiving ends, causing momentary inconsistencies in the array processing link.

[0096] The specific implementation method of step S5 is as follows:

[0097] S501 submits frame reception and frame integrity verification information recalculation.

[0098] Submission frames and data frames share the same transmission link. Link jitter, burst errors, and buffer contention can perturb the content of submission frames, turning them into erroneous frames. The receiving end needs to exclude erroneous frames before parsing fields to avoid introducing incorrect configuration version numbers and unified submission frame sequence numbers into the loading process. After receiving a submission frame, the receiving end separates the frame header, payload, and frame trailer according to a fixed frame structure. It reads the frame integrity verification information from the frame trailer and concatenates the frame header and payload in a pre-defined byte order to form a verification input sequence. Cyclic redundancy check (CRC) is performed on the verification input sequence. The CRC process uses rule-driven shift and XOR operations. Starting from the first byte of the verification input sequence, each byte is fed into the shift register structure. After each byte is fed in, the register structure is XORed in a fixed order according to the generation rules until the last byte of the verification input sequence is processed. The remaining value in the register structure is used as the CRC result. The recalculation verification result is compared bit by bit with the frame integrity verification information carried at the end of the frame. If they do not match, the submitted frame is discarded while the current configuration remains unchanged. At the same time, a verification failure flag is written to the submission receipt and a submission receipt is sent. If they match, the field parsing begins. The submitted frame only enters the parsing path when the verification is successful. The source of the unified submitted frame sequence number and configuration version number is kept reliable, and erroneous frames will not trigger configuration loading.

[0099] S502 Field Parsing and Field Consistency Verification.

[0100] Even after a submitted frame passes verification, unreasonable field combinations may still occur. For example, the unified submission frame sequence number might fall into an already output frame sequence range, or duplicate configuration version numbers might lead to unclear loading semantics. The receiving end reads the configuration version number and unified submission frame sequence number from the submission frame payload, reads the current configuration version number, and performs a consistency check. If the configuration version number matches the current configuration version number, it does not directly enter the pending submission area for writing. Instead, it first verifies whether the unified submission frame sequence number matches the effective boundary already confirmed by the receiving end. The verification method uses a combination of rollback discrimination and sequence boundary discrimination. Sequence boundary discrimination uses the consecutive tail frame sequence number as a benchmark to confirm that the unified submission frame sequence number does not fall before the consecutive tail frame sequence number. Rollback discrimination utilizes the maximum frame sequence position recorded in the rearranged buffer. The receiving end interprets the order relationship between the unified submission frame sequence number and the consecutive tail frame sequence number within the observation context of the maximum frame sequence position. Once the unified submission frame sequence number shows a past position in the rollback arrival interval, the field verification is marked as failed. When field verification passes, a commit receipt is still sent. The commit receipt includes the configuration version number, the unified commit frame sequence number, and a consistency flag. When field verification fails, a commit receipt is sent along with a field failure flag. The receiving end explicitly provides feedback on duplicate versions and boundary consistency, allowing the sending end to distinguish between duplicate commits and ineffective ones. Link retransmission will not cause version state drift.

[0101] S503 Write to the area to be submitted and generate the self-verification value of the area to be submitted.

[0102] After the new configuration is written to the commit area, a waiting period occurs. During this period, data frames continue to be processed according to the current configuration. Storage disturbances and write interruptions can corrupt the content of the commit area. The receiving end needs consistent self-checking mechanisms before and after loading. When the field consistency verification passes and the configuration version number is different from the current configuration version number, the receiving end completely copies the new configuration from the commit frame payload to the commit area data area. At the same time, it writes the configuration version number and the unified commit frame sequence number to the commit area metadata area. After the commit area is written, the receiving end performs the same cyclic redundancy check recalculation as the commit frame in the commit area data area. The recalculation process also sends the data to the shift register structure in byte order and performs XOR correction according to the generation rules until the new configuration byte sequence is processed. The remaining value of the register structure is used as the commit area self-check value. The receiving end writes the commit area self-check value to the commit area metadata area and sets the commit area status to "written but not loaded". The commit area self-check value and the write status together form a verifiable snapshot to be loaded. When the loading action encounters storage disturbances, it can be identified in time, avoiding the entry of half-written configuration into the current configuration.

[0103] S504 sequential output pointer alignment and unified submission frame sequence number trigger determination.

[0104] The unified commit frame sequence number defines the effective boundary of the new configuration. The receiving end must place the loading action before the data frame corresponding to the unified commit frame sequence number enters the processing pipeline; otherwise, the same data frame will be processed with different configurations at different receiving ends. While the barrier is frozen, the receiving end continuously maintains the rearrangement buffer and the sequential output pointer. Each time it prepares to output the next data frame from the rearrangement buffer, the receiving end first obtains the frame sequence number of the next frame and performs an equality check with the unified commit frame sequence number. If the equality check fails, the next data frame is processed according to the current configuration, and the sequence number of the consecutive last frames is advanced. If the equality check succeeds, the next data frame is temporarily suspended from entering the processing pipeline, and the loading action branch is initiated. The trigger check pins the loading action before the frame corresponding to the unified commit frame sequence number, ensuring that the effective boundary remains consistent with the unified commit frame sequence number determined by the sending end, and the cross-end effective point is not disturbed by the output clock.

[0105] S505 One-time loading action and effective boundary locking.

[0106] Configuration loading involves multiple parameter fields. If the loading process is executed in segments, some parameters may be mixed up (old and new) within a short interval of the array processing chain. After the trigger determination is successful, the receiving end first reads the configuration version number and unified submission frame sequence number from the metadata area of ​​the area to be submitted, confirms that they are consistent with the parsing result of the submission frame, and then recalculates the cyclic redundancy check result of the data area of ​​the area to be submitted and compares it bit by bit with the self-check value of the area to be submitted. If the comparison is inconsistent, the loading is terminated and a submission receipt is sent to write a loading failure flag. If the comparison is consistent, the new configuration of the data area of ​​the area to be submitted is copied to the current configuration storage area at once. After the copy is completed, the current configuration version number is updated to the configuration version number of the metadata area of ​​the area to be submitted, and the status of the area to be submitted is set to submitted, and the unified submission frame sequence number and self-check value in the metadata area of ​​the area to be submitted are cleared. After the receiving end completes the loading, it immediately sends the previously deferred next frame data frame into the processing pipeline and starts to apply the new configuration to this frame. The loading action is completed before the frame corresponding to the unified submission frame sequence number, and the loading process is completed at once. The effective boundary is locked on the same frame sequence number, and parameter switching will not have interleaved segments.

[0107] S506 submission receipt generation, receipt verification writing and sending.

[0108] The sending end needs to confirm whether each receiving end has completed loading at the unified submission frame sequence number. The submission receipt must carry the configuration version number and the unified submission frame sequence number and can be verified by the sending end using the same verification rules. After successful loading, the receiving end generates a submission receipt. The submission receipt payload contains the configuration version number and the unified submission frame sequence number, as well as the receiving end's effective frame sequence number. The effective frame sequence number is taken as the sequence number of the next data frame that was temporarily suspended when the trigger judgment was established. The receiving end performs cyclic redundancy check recalculation on the submission receipt frame header and payload. The recalculation process is consistent with that of the submission frame. The receiving end writes the recalculation result to the submission receipt frame tail as frame integrity verification information. The receiving end injects the submission receipt into the same transmission link as the data frame and sends it. The sending order follows the rule of writing before sending; the submission receipt cannot be injected into the link before the fields are written. The submission receipt simultaneously transmits the version, boundary, and verification, allowing the sending end to directly verify that multiple receiving ends have completed the activation at the same unified submission frame sequence number. Retransmissions and duplicate receipts can also be parsed consistently.

[0109] At the receiving end, the submitted frame is first filtered by frame integrity verification information, and then filtered by logical consistency verification between the configuration version number and the unified submission frame sequence number. The new configuration is written to the pending submission area and the pending submission verification value is saved. The receiving end performs a one-time loading operation when the candidate output frame sequence number driven by the sequential output pointer equals the unified submission frame sequence number. Verification remains consistent before and after loading. After loading, the configuration takes effect from the data frame corresponding to the unified submission frame sequence number. The submission receipt, carrying the configuration version number, unified submission frame sequence number, effective frame sequence number, and frame integrity verification information, is sent to the sending end. The sending end uses this information to verify the consistency of multiple receiving end boundaries.

[0110] Specifically, the above are merely preferred embodiments of this application and are not intended to limit this application.

[0111] In the description of this specification, references to terms such as "an embodiment," "example," and "specific example" indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0112] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A data transmission method for a digital phased array, characterized in that, Including the following steps: S1: The transmitting end continuously encapsulates the sampling data generated by the digital phased array into data frames, writes the frame sequence number into the data frames, and sends the data frames in order according to the frame sequence number. S2: Before the configuration change, the sending end sends a barrier frame. The barrier frame is written with the barrier sequence number and is inserted into the data frame sequence and sent using the same transmission link as the data frames. S3: After receiving the barrier frame, the receiving end stops the configuration loading action, records the barrier landing frame sequence number and the consecutive tail frame sequence number, writes the barrier sequence number, barrier landing frame sequence number, consecutive tail frame sequence number and neighboring frame sequence event string into the barrier receipt, and sends the barrier receipt to the sending end. S4: After receiving the barrier receipt, the sending end forms a submission decision and determines the unified submission frame sequence number. Based on the submission decision, it chooses to send a submission frame with the configuration version number and unified submission frame sequence number written in it, or chooses to retransmit the barrier frame and collect the barrier receipt again. S5: After receiving the submission frame, the receiving end writes the new configuration into the submission area. When the local data frame sequence number reaches the unified submission frame sequence number, it loads the new configuration into the submission area and takes effect from the corresponding data frame. It writes the configuration version number and the unified submission frame sequence number into the submission receipt and sends the submission receipt to the sending end.

2. The data transmission method for a digital phased array according to claim 1, characterized in that, Step S1 includes the following: The transmitting end encapsulates the continuous sampling data generated by the digital phased array into data frames according to a fixed frame structure. For each data frame, frame integrity verification information is generated. In the generation process, the frame header and payload are accumulated by taking words with a fixed word width and then truncated by the word width. Then, the accumulated result is accumulated with the frame sequence number and truncated by the word width. Finally, the result is inverted bit by bit to obtain the frame integrity verification information and written to the frame tail.

3. The data transmission method for a digital phased array according to claim 2, characterized in that, Step S1 also includes the following: The sending end maintains the maximum sequence number of the sent frame on the sending side, writes data frames into the sending queue in ascending order of frame sequence number, reads the frame sequence number of the data frame at the head of the sending queue and performs a continuity determination with the maximum sequence number of the sent frame. When the continuity determination satisfies that the sequence number of the frame at the head of the queue is equal to the maximum sequence number of the sent frame plus one, the link is injected and the maximum sequence number of the sent frame is updated. If the continuity determination is not met, the injection link is paused and the data frame corresponding to the missing frame sequence number is reconstructed. The reconstructed data frame is written with the missing frame sequence number and the frame integrity verification information is rewritten.

4. The data transmission method for a digital phased array according to claim 3, characterized in that, Step S2 includes the following: Before the configuration change, the sending end generates a barrier frame and shares the same sending queue and transmission link with the data frame. It performs continuous segment identification in the sending queue, obtains the number of consecutive acknowledgment frames, adds the maximum sent frame sequence number to the number of consecutive acknowledgment frames and then adds one to obtain the target insertion frame sequence number, and inserts the barrier frame before the data frame whose frame sequence number is equal to the target insertion frame sequence number. The barrier frame header is written with the frame type identifier and the barrier sequence number, and the barrier frame payload is written with the target insertion frame sequence number.

5. The data transmission method for a digital phased array according to claim 4, characterized in that, Step S2 also includes the following: After the barrier frame is inserted, the sending end performs an adjacent frame consistency check. The adjacent frame consistency check includes forward check and backward check. The forward check determines that the frame sequence number of the data frame immediately before the barrier frame is equal to the target inserted frame sequence number minus one. The backward check determines that the frame sequence number of the data frame immediately after the barrier frame is equal to the target inserted frame sequence number. When both forward and backward checks are true, the sending end sends data frames and barrier frames in the order of the sending queue and updates the sequence number of the largest sent frame; when either forward or backward check is false, the sending end cancels the barrier frame insertion position and re-executes continuous segment identification.

6. The data transmission method for a digital phased array according to claim 5, characterized in that, Step S3 includes the following: The receiving end recalculates the frame integrity check information for each data frame and compares it with the frame tail. Data frames that match are written to the rearrangement buffer. The consecutive tail frame sequence number is used as the sequential output pointer. The expected frame sequence number is obtained by adding one to the consecutive tail frame sequence number. The receiving end searches for data frames whose frame sequence number is equal to the expected frame sequence number in the rearrangement buffer and outputs them sequentially. The consecutive tail frame sequence number is updated to the expected frame sequence number. Only one copy of the data frame is retained for the same frame sequence number.

7. The data transmission method for a digital phased array according to claim 6, characterized in that, Step S3 also includes the following: After the receiving end identifies the barrier frame, it enters the barrier frozen state and stops the configuration loading action. When the expected frame number cannot be matched, it reads the minimum frame number in the rearrangement buffer. When the minimum frame number in the rearrangement buffer is greater than the expected frame number, it records the minimum frame number in the rearrangement buffer as the barrier landing frame number and records the current consecutive tail frame number synchronously. When the arriving frame number is greater than the expected frame number, it writes a skip event. When the arriving frame number is less than the expected frame number, it writes a backtrack event. When the arriving frame number is less than or equal to the consecutive tail frame number, it writes a repeat event, or when the arriving frame number is greater than the consecutive tail frame number and the same arriving frame number already exists in the rearrangement buffer, it writes a repeat event. It generates a neighboring frame sequence event string and writes it together with the barrier number, the barrier landing frame number, and the consecutive tail frame number into the barrier receipt.

8. The data transmission method for a digital phased array according to claim 7, characterized in that, Step S4 includes the following: After receiving barrier receipts from multiple receivers, the transmitting end reads the barrier sequence number one by one and merges barrier receipts with the same barrier sequence number into a barrier receipt set. Within the barrier receipt set, the completeness of the fields is verified one by one. The fields include the barrier sequence number, the barrier landing frame sequence number, the consecutive tail frame sequence number, and the neighboring frame sequence event string. Barrier receipts with any missing field are removed from the barrier receipt set. The transmitting end obtains the landing offset frame value by subtracting the consecutive tail frame sequence number from the barrier landing frame sequence number. The landing offset frame values ​​are sorted and the convergence anchor point offset frame value is obtained. The upper and lower expansion amounts of the anchor point are calculated and the barrier reaches the convergence bandwidth.

9. A data transmission method for a digital phased array according to claim 8, characterized in that... Step S4 also includes the following: The sending end parses the front, boundary, and back segments of the neighboring frame sequence event string, extracts the boundary disturbance intensity characterization value, and statistically analyzes the stable state marker of the front segment and the tense state marker of the back segment to generate the barrier neighboring frame sequence continuous pressure index. Based on the barrier reaching the convergence bandwidth and the barrier neighboring frame sequence continuous pressure index, a submission decision is formed. When the submission decision is a submission frame, the unified submission frame sequence number is determined by adding the maximum barrier landing point frame sequence number to the stable delay amount, written into the configuration version number, and the submission frame is sent. When the submission decision is a retransmission of the barrier frame, the barrier frame is retransmitted and the barrier receipt is collected.

10. A data transmission method for a digital phased array according to claim 9, characterized in that, Step S5 includes the following: The receiving end receives the submission frame, recalculates the frame integrity verification information and compares it with the frame integrity verification information, parses the configuration version number and the unified submission frame sequence number to complete the rollback judgment and sequence boundary judgment, writes the new configuration into the submission area and generates the self-verification value of the submission area. The receiving end first obtains the frame sequence number of the next frame, performs an equality check between the frame sequence number of the next frame and the unified submission frame sequence number. If the equality check is successful, the next frame data frame is temporarily suspended from being sent into the processing pipeline and the loading action branch is entered. The receiving end reads the configuration version number and unified submission frame sequence number from the metadata area of ​​the area to be submitted, confirms that they are consistent with the parsing result of the submission frame, and then recalculates the cyclic redundancy check result of the data area of ​​the area to be submitted and compares it bit by bit with the self-check value of the area to be submitted. If they match, the new configuration of the data area of ​​the area to be submitted is copied to the current configuration storage area in one go. After the copying is completed, the current configuration version number is updated to the configuration version number of the metadata area of ​​the area to be submitted, and the previously deferred next frame data frame is sent into the processing pipeline so that the new configuration takes effect according to the corresponding next frame data frame. After successful loading, the receiving end generates a submission receipt. The submission receipt payload is written with the configuration version number and unified submission frame sequence number, and the effective frame sequence number of the receiving end is also written. The effective frame sequence number of the receiving end is taken as the frame sequence number of the deferred next frame data frame when the equality judgment is true.

Images