Point-to-point pre-negotiation adaptive image transmission system and method

Abstract

The invention discloses a point-to-point pre-negotiation adaptive image transmission system and method, and the method comprises the following steps: S1, a request stage: a transmitting end A sends a communication request, and a receiving end B carries parameters to give a response; s2, in a test stage, the sending end A sends the test packets in groups until all the test packets are sent; s3, in the negotiation stage, the receiving end B obtains communication parameters through algorithm processing and informs the sending end A of the communication parameters; and S4, in a communication stage, the sending end A sends the image data to the receiving end B. According to the invention, the performance and robustness of image transmission are optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of image transmission, and in particular to a point-to-point pre-negotiation adaptive image transmission system and method. Background Art

[0002] In the field of point-to-point image transmission, existing industrial camera transmission protocols such as GigEVision, CoaXPress and CameraLink, although they perform well in specific application scenarios, generally have problems such as fixed transmission parameters, lack of adaptability, fixed packet header format and insufficient cache congestion control, resulting in large differences in transmission performance in different network environments. In addition, a large number of retransmissions are easily generated when the cache is congested, affecting transmission efficiency and reliability. Summary of the invention

[0003] Based on the technical problems existing in the background technology, the present invention proposes a point-to-point pre-negotiation adaptive image transmission system and method, which optimizes the performance and robustness of image transmission.

[0004] The point-to-point pre-negotiation adaptive image transmission method proposed by the present invention has the following steps: S1: Request phase, the sender A sends a communication request, and the receiver B responds with parameters; S2: Testing phase, sender A sends test packets in groups until all packets are sent; S3: During the negotiation phase, the receiving end B obtains the communication parameters through algorithm processing and informs the sending end A; S4: Communication phase, the sender A sends image data to the receiver B.

[0005] Preferably, the method steps of the algorithm processing in S3 are as follows: S31: Abstracting comprehensive performance indicators Y 0 Model and Equivalence Dimensions Overall Index Y 1 Model; S32: Calculate the total index based on the weight obtained by entropy weight method Y 1 ; S33: Least squares fitting comprehensive performance index Y 0 ; S34: Gradient ascent method to obtain comprehensive performance indicators Y 0 The optimal value is used and communication is performed with the current MTU value and IFG value.

[0006] Preferably, the comprehensive performance index in S31 is Y 0 The model is:

[0007]

[0008]

[0009]

[0010] In the formula, is the transmission efficiency weight; is the delay weight; is the processing capacity weight; For transmission efficiency; is the total amount of data that needs to be transmitted; is the total amount of data transmitted; is the total number of Ethernet frames; It is the maximum transmission unit of the application layer data packet; It is the sum of Ethernet protocol overhead outside the application layer; is the size of the application layer data packet header; is the total delay; is the total transmission time; is the total processing time; It is the total Ethernet frame interval IFG time; Transmission time for each Ethernet frame; Processing time for each Ethernet frame; is the minimum time interval between Ethernet frames; The number of Ethernet frames that can be processed per second.

[0011] Preferably, the total index of equivalent dimension in S31 Y 1 The model is:

[0012] In the formula, is the throughput; is the packet loss rate; is the weight of throughput; is the weight of the packet loss rate.

[0013] Preferably, in S32, the total index is calculated based on the weight obtained by the entropy weight method Y 1 The method steps are as follows: S321: Taking the initial byte interframe interval IFG as the initial value, the bytes are increased gradually, and several value tests are performed to obtain the throughput and packet loss rate; S322: For n samples and m indicators, let Represents the value of the jth indicator of the i-th sample and normalizes the value; S323: Calculate the weight of the j-th indicator sample value in the indicator in turn for the normalized data; S324: Calculate the entropy value of the j-th indicator ;

[0014] In the formula, ,satisfy ; S325: Calculate the weight of each indicator through the entropy value of each indicator to determine the total indicator Y 1 expression.

[0015]

[0016] Preferably, the comprehensive performance index of the fitting in S33 is Y 0 The method steps are as follows: S331: Based on the overall index Y 1 The expression is calculated under a certain sample Y 1 The value of Y 1 Assign to Y 0 ; S332: Comprehensive performance indicators Y 0 get , , and Y 0 The mapping relationship between them is used to construct the data matrix from the statistical samples and the target vector Y 0 ; S333: When When the specific value of the weight is obtained, the comprehensive performance index is achieved Y 0 The fitting of

[0017] Preferably, the comprehensive performance index in S34 is Y 0 The optimal value solution is: select the initial MTU value and initial IFG time , set the learning rate Perform a gradient ascent update to maximize Y 0 :

[0018]

[0019]

[0020]

[0021] Repeat the above steps until Y 0 When convergence or the maximum number of iterations is reached, communication is performed based on the MTU and IFG values ​​at that time.

[0022] Preferably, the communication stage in S4 adopts sliding blocking window, adaptive variable frame storage and cumulative response.

[0023] Preferably, the workflow of the sliding blocking window is: The sender A continuously sends data packets according to the negotiated frame interval IFG, and the receiver B returns a response signal each time it receives a data packet. The response signal contains the secondary sequence number of the next data packet to be received. The sender A receives the expected response signal from the receiver B and continuously pushes the sliding blocking window forward according to the sequence number, thereby allowing the sender A to send new data packets; When the sender A receives two identical expected response packets, it enters the retransmission logic of data packet reading back or reassembly. At this time, the sliding blocking window stops sliding and performs the blocking function. The sequential sending logic of the data packet automatically enters PAUSE until the retransmission is completed, and then continues to send sequentially.

[0024] Preferably, the workflow of cumulative response is as follows: Divide a frame of image into several data payload packets, first transmit the header packet of this frame of image, then transmit several data payload packets in sequence, and finally transmit the tail packet. Each packet has its own secondary sequence number. The first-level sequence number indicates the current data block to which it belongs, and the second-level sequence number indicates the number of the data packet to which it belongs. The data payload packet is collected by FPGA from the front-end image sensor and stored in DDRSDRAM. The header packet and tail packet are generated by FPGA according to the current transmission data block. The sender A receives the response signal from the receiver B. The response signal contains the secondary sequence number that the receiver B expects to receive. If the sender A receives two identical expected response packets in succession, it reads back the data of the specific address from DDRSDRAM and re-packs it for retransmission. If the expected response packet is the expected retransmission header packet or tail packet, it is re-packed and retransmitted.

[0025] Beneficial technical effects of the present invention: (1) The present invention can automatically adjust the MTU and IFG values ​​according to the network conditions of the communication and the response capability of the terminal device. Under high-latency or unstable network conditions, the protocol adopts a more conservative data transmission strategy; while in a high-speed and stable network environment, a more radical optimization measure is adopted. The adaptability and robustness of the image transmission protocol are improved.

[0026] (2) The present invention can also avoid unnecessary retransmission caused by misjudgment when the response packet is lost, and perform selective retransmission to improve transmission efficiency when packet loss occurs. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 A flowchart of the point-to-point pre-negotiation adaptive image transmission method proposed by the present invention; Figure 2 The application layer protocol based on the five-layer communication model proposed by the present invention; Figure 3 The FPGA proposed by the present invention realizes the data flow of the Ethernet protocol stack; Figure 4 This is a diagram of the application layer data packet structure proposed by the present invention. DETAILED DESCRIPTION

[0028] The present invention will be further explained below in conjunction with specific embodiments.

[0029] Reference Figure 2-4The image transmission protocol of the present invention supports the communication parties to conduct pre-negotiation before formal transmission to determine the best transmission parameters (such as the application layer maximum transmission unit MTU and frame interval IFG). This process takes into account the actual situation of the network and the capabilities of the terminal equipment, thereby achieving more flexible and adaptable communication. In addition, the header information of the application layer data packet can be dynamically adjusted according to different network conditions and transmission requirements. For example, when the MTU is small, a complex verification mechanism may not be required; when the MTU is large, a more powerful verification function is added, such as 16-bit or 32-bit CRC verification, to ensure the accuracy and reliability of the data. The protocol also uses a sliding blocking window to control the data flow, and adjusts the sending rate through the response signal fed back by the receiving end, which not only effectively avoids network congestion, but also ensures that the data packets arrive in order, reducing the possibility of erroneous retransmission. Multiple DDRSDRAMs are used to store image data, and efficient data transmission is achieved through window blocking and automatic PAUSE mechanisms. When data packet loss is detected, it can be quickly retransmitted while maintaining the integrity and real-time nature of the image data. In addition, the adaptive variable frame storage mechanism allows a frame of image to be stored as a data block in a DDRSDRAM. The FPGA, as the camera end, i.e., the transmitter A, has multiple DDRSDRAMs, which send image data to the host, i.e., the receiver B. The transmitter A will receive the response signal from the receiver B, and continuously push the sliding blocking window forward according to the sequence number, thereby allowing the transmitter to send new data packets. When the transmitter A receives two identical expected response packets, the sliding blocking window stops sliding, and the sending logic automatically PAUSEs, and enters the retransmission logic of data packet readback or reorganization, avoiding the huge bandwidth waste caused by the cumulative packet loss caused by the continuous output of data streams, and improving the bandwidth utilization. The protocol uses the entropy weight method, the least squares method, and the gradient descent method to design the test value strategy, calculate the weights, and select the optimization parameters, ensuring the optimal performance of the protocol under various network conditions.

[0030] Specifically, refer to Figure 1 The point-to-point pre-negotiation adaptive image transmission method proposed by the present invention has the following steps: S1: Request phase, the sender A sends a communication request, and the receiver B responds with parameters; S2: Testing phase, sender A sends test packets in groups until all packets are sent; S3: During the negotiation phase, the receiving end B obtains the communication parameters through algorithm processing and informs the sending end A; S4: Communication phase, the sender A sends image data to the receiver B.

[0031] The steps of algorithm processing in S3 are as follows: Abstract comprehensive performance indicators Y 0 Model and Equivalence Dimensions Overall IndexY 1 Model Comprehensive performance indicators Y 0 The model is as follows:

[0032]

[0033]

[0034]

[0035] In the formula, is the transmission efficiency weight; is the delay weight; is the processing capacity weight; For transmission efficiency; is the total amount of data that needs to be transmitted; is the total amount of data transmitted; is the total number of Ethernet frames; It is the maximum transmission unit of the application layer data packet; It is the sum of Ethernet protocol overhead outside the application layer; is the size of the application layer data packet header; is the total delay; is the total transmission time; is the total processing time; It is the total Ethernet frame interval IFG time; Transmission time for each Ethernet frame; Processing time for each Ethernet frame; is the minimum time interval between Ethernet frames; The number of Ethernet frames that can be processed per second.

[0036] Total index of equivalent dimension Y 1 The model is as follows:

[0037] In the formula, is the throughput; is the packet loss rate; is the weight of throughput; is the weight of the packet loss rate.

[0038] The total index is calculated based on the weight obtained by entropy weight method Y 1 In this embodiment, the test value strategy (MTU and IFG) is: set the initial value of the MTU (Maximum Transmission Unit) test of the application layer data packet to the maximum multiple of 128 bytes contained by half of the maximum value of the receiving end buffer, with 128 bytes as the step size, and the MTU gradually decreases to a minimum of 128 bytes. (The maximum buffer area of ​​the receiving end is 4096 bytes under normal circumstances, ensuring that the receiving end B can cache at least two data packets, avoiding waste of the receiving buffer area and maximizing the use of the receiving area buffer area). At the beginning, the initial value of the MTU of the application layer maximum transmission unit is calculated by the receiving end B as the maximum multiple of 128 bytes contained by half of the maximum value of the receiving end buffer area, and sends it to the sending end A, which then conducts subsequent tests.

[0039] Selecting an integer multiple of 128 bytes as the packet size can make full use of memory alignment and cache optimization, reduce the number of memory accesses, improve memory access speed, and make better use of cache lines to reduce cache miss rates. DMA controllers usually transmit in fixed-size data blocks. 128 bytes is a common DMA transmission block size, which can optimize DMA transmission, reduce the number of transmissions, improve transmission efficiency, and reduce CPU burden. Simplify the header overhead management and segmentation reorganization process in the network protocol stack, reduce processing complexity and time; the maximum value that can be stored in the register of the hardware circuit is an exponential power of 2, better utilize the characteristics of the network processor and FPGA hardware accelerator, improve data processing speed and overall performance, thereby significantly reducing data processing delays and improving bandwidth utilization of data transmission.

[0040] The Ethernet standard IEEE802.3 stipulates that the minimum frame interval is 12 bytes (96 bit time) to ensure that network devices have enough time to process the previous frame and prepare to receive the next frame. Modern network devices usually take this standard into consideration when designing, and their processing time and buffer management are optimized to operate efficiently within the range of 12 bytes to 20 bytes. Multiple studies and actual tests have shown that the impact of frame intervals in this range on network throughput and latency is significant. A smaller frame interval can improve throughput, but a too small frame interval will increase the probability of collision; a larger frame interval will increase latency, but can reduce device processing pressure and packet loss rate. 20 bytes, as a relatively large value, can provide enough data points to evaluate the impact of different frame intervals on network performance, while avoiding waste of resources caused by excessive frame intervals. Therefore, it is scientific and reasonable to choose a range of 12 bytes to 20 bytes for testing. Take the frame interval IFG of 12 bytes as the initial value, and increase it by 1 byte in steps up to 20 bytes. Each group of (M, g) values ​​is tested 2 16 times, in order to obtain the throughput and packet loss rate. For n samples and m indicators, let Represents the value of the jth indicator of the ith sample.

[0041] Since the dimensions of various indicators are not consistent, the range transformation method is used for normalization before calculation to make heterogeneous indicators homogeneous. Differentiation is used for positive indicators such as throughput T and negative indicators such as packet loss rate L. For convenience, the normalized data is still recorded as .

[0042] Positive indicators:

[0043] Negative indicators:

[0044] After normalizing the test data, the weight of the j-th indicator sample value in the indicator is calculated in turn:

[0045] Calculate the entropy value of the j-th indicator, where ,satisfy :

[0046] The weight of each indicator is calculated by the information entropy of each indicator:

[0047] This can be obtained Specific formula.

[0048] Least squares method to fit comprehensive performance index Y 0 within a certain range Y 0 and Y 1 can be considered equivalent. According to the test results, Calculate under a certain sample The value of Assign to . Based on comprehensive performance indicators , can be obtained , , and The mapping relationship between them. Construct the data matrix from the statistical samples and the target vector .

[0049] when When , the specific value of the weight is obtained. The expression for the error sum of squares is minimize.

[0050] Gradient ascent method to obtain comprehensive performance indicators Y 0 The optimal value of At this point the objective function has a specific form:

[0051] Calculate the objective function with respect to and The gradient is:

[0052] in:

[0053]

[0054] Select the initial MTU value and initial IFG time ; Set the learning rate , perform gradient ascent update to maximize :

[0055]

[0056] Repeat the above steps until When convergence or the maximum number of iterations is reached, communication is performed based on the MTU and IFG values ​​at that time.

[0057] The image transmission protocol of the present invention can automatically adjust the values ​​of its MTU and IFG according to the network conditions of the communication and the response capability of the terminal device. Under high-latency or unstable network conditions, the protocol adopts a more conservative data transmission strategy; while in a high-speed and stable network environment, a more radical optimization measure is adopted. The adaptability and robustness of the image transmission protocol are improved.

[0058] In addition, the communication stage in S4 of the present invention adopts sliding blocking window, adaptive variable frame storage and cumulative response.

[0059] For the sliding blocking window, the window size is notified by the receiving end B during the negotiation phase. The window size is equal to the size of the buffer area of ​​the receiving end B. The sending end A can continuously send data packets within the maintained window capacity according to the frame interval IFG negotiated in the previous stage. The receiving end B returns a response signal for each data packet received. The response signal contains the secondary sequence number of the next data packet to be received. The sending end A will accept the expected response signal from the receiving end B and continuously push the sliding blocking window forward according to the sequence number, thereby allowing the sending end to send new data packets. When the sending end A receives two identical expected response packets and enters the retransmission logic of data packet readback or reassembly, the sliding blocking window stops sliding and performs the blocking function. The sequential transmission logic of the data packet automatically enters PAUSE until the retransmission ends, and then continues to send sequentially. The blocking window limits the sending end A from uncontrolled burst transmission of data when the receiving end B is congested, avoiding the cumulative congestion of the flow channel.

[0060] For adaptive variable frame storage, a frame of image is stored as a data block in a DDRSDRAM. FPGA, as the camera end, has multiple DDRSDRAMs, and sends image data to the host, namely, the receiving end B. The sending end A will receive the response signal from the receiving end B, and continuously push the sliding blocking window to slide forward according to the sequence number, so as to allow the sending end to send new data packets. When the sending end A receives two identical expected response packets, it enters the retransmission logic of data packet readback or reorganization. At this time, the sliding blocking window stops sliding and performs the blocking function. The sequential transmission logic of the data packet automatically enters PAUSE until the retransmission is completed, and the sequential transmission continues.

[0061] The sender A transmits an image to the receiver B. The variable frame storage is realized by window blocking (the response with a serial number pushes the window forward) plus automatic PAUSE, and the missing data is read back and retransmitted. The sender A receives the header packet ACK of the next frame of the image data block of a certain frame of the image data block of the receiver B, which means that the image data block of the receiver B has been completely received. The DDRSDRAM storing this data block can be overwritten by the new image sensor input data. During the retransmission logic on the camera side, the sequential transmission logic is suspended, and the data to be sent from the upstream image sensor is stored in the overwriteable DDRSDRAM. The onboard multi-chip DDRSDRAM adopts a sequential overwriting cycle read and write strategy to avoid image frame loss in the case of network transmission congestion, smooth the image data flow from the front-end acquisition sensor to the receiving host, and ensure the accuracy and integrity of the data collected by the camera.

[0062] For cumulative response, a frame of image is stored as a data block in a DDRSDRAM. To transmit this complete data block, it is divided into several data payload packets. When transmitting a data block, the header packet of the data block is transmitted first, and then several data payload packets are transmitted in sequence, and finally the tail packet is transmitted. These three packets have their own secondary sequence numbers. The first-level sequence number indicates the current data block to which it belongs, and the second-level sequence number indicates the number of the data packet to which the current data packet belongs.

[0063] The data payload packet is collected by FPGA from the front-end image sensor and stored in DDRSDRAM. The header packet and tail packet are generated by FPGA according to the current transmission data block. The sender A will receive the response signal from the receiver B. The response signal contains the secondary sequence number that the receiver B expects to receive. Once the sender A receives two identical expected response packets in succession, it reads back the data of the specific address from DDRSDRAM and re-packs it for retransmission. If the expected response packet is the expected retransmission header packet or tail packet, it will be re-packed and retransmitted.

[0064] The cumulative response can successfully receive the data packet at the receiving end B. In the event of packet loss, selective retransmission can be performed based on logical judgment, and only the lost data packet is transmitted to reduce unnecessary retransmission. In the event of accidental loss of the returned response packet, the probability of erroneous retransmission is greatly reduced, avoiding unnecessary waste of network resources and improving bandwidth utilization. The receiving end B maintains an expected response sequence number register ERSNREGISTER (Expect response serialnumberregister). Whenever the receiving end B receives a data packet, the register ERSN enters the logic of determining whether to self-increment. The receiving end B returns a response signal based on the register ERSN, which contains the secondary sequence number of the next received data packet.

[0065] For the receiving end B, only when the data packets received by the receiving end B are sequential, such as receiving data packets with sequence numbers T, T+1, and T+2 in sequence, the expected response sequence number maintained by the register ERSN is incremented to T+3 in sequence, and the expected sequence numbers contained in the returned response packets are T+1, T+2, and T+3 in sequence. If data packets with sequence numbers T, T+2, and T+3 are received in sequence, the data packet with sequence number T+1 is lost. The judgment process of the receiving end is analyzed as follows: the expected response sequence number maintained by the register ERSN is incremented to T+1 when the data packet with sequence number T is received, and the response packet with expected sequence number T+1 is returned. Then the receiving end B receives the data packet with sequence number T+2, which is inconsistent with the expected data packet. The register ERSN does not increment, and the expected response packet with expected sequence number T+1 is sent again. At the same time, the receiving end B stores the T+2 data packet. During the transmission of the entire data block, the sending end can send the lost packet. At this point, receiving end A will receive two identical response packets, and determine that the data packet with sequence number T+1 is lost, so it organizes retransmission.

[0066] For sender A, when a response packet returned by receiver B to sender A is accidentally lost, the expected sequence number contained in the received response packet is discontinuous, and only the maximum sequence number needs to be paid attention to. The mechanism of judging packet loss by two identical response packets perfectly avoids the situation where a certain response packet is lost.

Claims

1. A point-to-point pre-negotiation adaptive image transmission method, characterized in that: The steps are as follows: S1: Request phase, the sender A sends a communication request, and the receiver B responds with parameters; S2: Testing phase, sender A sends test packets in groups until all packets are sent; S3: During the negotiation phase, the receiving end B obtains the communication parameters through algorithm processing and informs the sending end A; S4: Communication phase, the sender A sends image data to the receiver B.

2. The point-to-point pre-negotiation adaptive image transmission method according to claim 1, characterized in that: The steps of algorithm processing in S3 are as follows: S31: Abstract the comprehensive performance index Y0 model and the equivalent dimension total index Y1 model; S32: Calculate the total index Y1 based on the weight obtained by the entropy weight method; S33: Least squares method fitting comprehensive performance index Y0; S34: The gradient ascent method is used to obtain the optimal value of the comprehensive performance indicator Y0, and communication is performed using the MTU value and IFG value at this time.

3. The point-to-point pre-negotiation adaptive image transmission method according to claim 2, characterized in that: The comprehensive performance index Y0 model in S31 is: ； ； ； ； In the formula, is the transmission efficiency weight; is the delay weight; is the processing capacity weight; For transmission efficiency; is the total amount of data that needs to be transmitted; is the total amount of data transmitted; is the total number of Ethernet frames; It is the maximum transmission unit of the application layer data packet; It is the sum of Ethernet protocol overhead outside the application layer; is the size of the application layer data packet header; is the total delay; is the total transmission time; is the total processing time; It is the total Ethernet frame interval IFG time; Transmission time for each Ethernet frame; Processing time for each Ethernet frame; is the minimum time interval between Ethernet frames; The number of Ethernet frames that can be processed per second.

4. The point-to-point pre-negotiation adaptive image transmission method according to claim 2, characterized in that: The total index Y1 model of the equivalent dimension in S31 is: ； In the formula, is the throughput; is the packet loss rate; is the weight of throughput; is the weight of the packet loss rate.

5. The point-to-point pre-negotiation adaptive image transmission method according to claim 4, characterized in that: The method steps for calculating the total index Y1 based on the entropy weight method in S32 are as follows: S321: Taking the initial byte interframe interval IFG as the initial value, the bytes are increased gradually, and several value tests are performed to obtain the throughput and packet loss rate; S322: For n samples and m indicators, let Represents the value of the jth indicator of the i-th sample and normalizes the value; S323: Calculate the weight of the j-th indicator sample value in the indicator in turn for the normalized data; S324: Calculate the entropy value of the j-th indicator ; ； In the formula, ,satisfy ; S325: Calculate the weight of each indicator through the entropy value of each indicator, so as to determine the expression of the total indicator Y1; 。 6. The point-to-point pre-negotiation adaptive image transmission method according to claim 5, characterized in that: The method steps for fitting the comprehensive performance index Y0 in S33 are as follows: S331: Calculate the value of Y1 under a certain sample according to the expression of the total index Y1, and assign Y1 to Y0; S332: obtained from the comprehensive performance index Y0 , , The mapping relationship between Y0 and Y1 is to construct a data matrix based on the statistical samples. and target vector Y0; S333: When When , the specific value of the weight is solved to achieve the fitting of the comprehensive performance index Y0.

7. The point-to-point pre-negotiation adaptive image transmission method according to claim 3, characterized in that: The method for solving the optimal value of the comprehensive performance indicator Y0 in S34 is: select the initial MTU value and initial IFG time , set the learning rate Perform a gradient ascent update to maximize Y0: ； ； ； ； Repeat the above steps until Y0 converges or reaches the maximum number of iterations, and then communicate with the MTU value and IFG value at this time.

8. The point-to-point pre-negotiation adaptive image transmission method according to claim 1, characterized in that: The communication phase in S4 uses sliding blocking window, adaptive variable frame storage and cumulative response.

9. The point-to-point pre-negotiation adaptive image transmission method according to claim 8, characterized in that: The workflow of the sliding blocking window is: The sender A continuously sends data packets according to the negotiated frame interval IFG, and the receiver B returns a response signal each time it receives a data packet. The response signal contains the secondary sequence number of the next data packet to be received. The sender A receives the expected response signal from the receiver B and continuously pushes the sliding blocking window forward according to the sequence number, thereby allowing the sender A to send new data packets; When the sender A receives two identical expected response packets, it enters the retransmission logic of data packet reading back or reassembly. At this time, the sliding blocking window stops sliding and performs the blocking function. The sequential sending logic of the data packet automatically enters PAUSE until the retransmission is completed, and then continues to send sequentially.

10. The point-to-point pre-negotiation adaptive image transmission method according to claim 8, characterized in that: The workflow of cumulative response is as follows: Divide a frame of image into several data payload packets, first transmit the header packet of this frame of image, then transmit several data payload packets in sequence, and finally transmit the tail packet. Each packet has its own secondary sequence number. The first-level sequence number indicates the current data block to which it belongs, and the second-level sequence number indicates the number of the data packet to which it belongs. The data payload packet is collected by FPGA from the front-end image sensor and stored in DDR SDRAM. The header packet and tail packet are generated by FPGA according to the current transmission data block. The sending end A receives the response signal from the receiving end B. The response signal contains the secondary sequence number that the receiving end B expects to receive. If the sending end A receives two identical expected response packets in succession, it reads back the data of the specific address from the DDR SDRAM and re-packs it for retransmission. If the expected response packet is the expected retransmission header packet or tail packet, it is re-packed and retransmitted.

Images