Data transmission method and device based on anti-packet loss strategy, equipment and storage medium

Abstract

The embodiment of the application provides a data transmission method and device based on an anti-packet loss strategy, equipment and a storage medium. The technical scheme provided by the embodiment of the application predicts the packet loss rate of the subsequent to-be-sent data packet, accurately predicts the first service quality score of an automatic repeat request strategy, the second service quality score of a forward error correction strategy and the third service quality score of a hybrid automatic repeat request strategy according to the predicted packet loss rate, determines the anti-packet loss strategy corresponding to the minimum service quality score as the target anti-packet loss strategy, and performs data transmission based on the target anti-packet loss strategy, accurately predicts the service quality of different anti-packet loss strategies based on the predicted packet loss rate, and determines the target anti-packet loss strategy used for data transmission according to the service quality, thereby effectively improving the anti-packet loss effect of data transmission.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of data transmission, and in particular to a data transmission method and device based on an anti-packet loss strategy, an equipment and a storage medium. BACKGROUND

[0002] In real-time audio and video communication, it is necessary to ensure that high-priority or critical task data streams (such as video conferencing and real-time voice calls) can obtain the required bandwidth, reduce delay, jitter and packet loss rate, etc. in complex network environments, so as to ensure smooth and reliable user experience. Therefore, anti-packet loss processing needs to be performed on data transmission.

[0003] Traditional anti-packet loss methods mainly include automatic repeat request (ARQ), forward error correction (FEC), hybrid automatic repeat request (HARQ), etc. Among them, the automatic repeat request relying on the feedback (ACK / NACK) of the receiving party will introduce a large transmission delay when the round-trip time (RTT) is large, and the anti-packet loss effect of data transmission is poor. The forward error correction applies a fault-tolerant coding algorithm to the source data packet to generate a redundant data packet, which can recover the lost source data packet with relatively low delay when data packet loss occurs. The key of forward error correction lies in predicting the packet loss rate in advance to determine whether a redundant packet needs to be encoded in advance and the size of the redundancy. The method for predicting the packet loss rate in the related art usually uses the average packet loss rate or the maximum packet loss rate in a past time window as an index to replace the future packet loss rate. This packet loss rate prediction has the problems of information lag and low accuracy, and the anti-packet loss effect of data transmission is poor. SUMMARY

[0004] Embodiments of the present application provide a data transmission method and device based on an anti-packet loss strategy, an equipment and a storage medium to solve the technical problem of poor anti-packet loss effect of data transmission in the related art. The packet loss rate of subsequent to-be-sent data packets in data transmission can be accurately predicted, and the target anti-packet loss strategy can be accurately determined according to the packet loss rate, thereby effectively improving the anti-packet loss effect of data transmission.

[0005] In a first aspect, embodiments of the present application provide a data transmission method based on an anti-packet loss strategy, comprising:

[0006] determining the reception success information of a first preset number of sent data packets in a data transmission link, and determining a link predicted packet loss rate of a second preset number of subsequent to-be-sent data packets according to the reception success information;

[0007] determine a first predicted retransmission packet number and a first predicted introduced latency for data transmission based on an automatic repeat request strategy, a first predicted redundancy and a second predicted introduced latency for data transmission based on a forward error correction strategy, and a second predicted retransmission packet number, a second predicted redundancy and a third predicted introduced latency for data transmission based on a hybrid automatic repeat request strategy according to the link predicted packet loss rate;

[0008] determine a first quality of service score for the automatic repeat request strategy according to the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced latency, a second quality of service score for the forward error correction strategy according to the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced latency, and a third quality of service score for the hybrid automatic repeat request strategy according to the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced latency;

[0009] determine the target anti-packet loss strategy corresponding to the minimum quality of service score among the first quality of service score, the second quality of service score and the third quality of service score, and perform data transmission based on the target anti-packet loss strategy, the anti-packet loss strategy including the automatic repeat request strategy, the forward error correction strategy and the hybrid automatic repeat request strategy.

[0010] In a second aspect, an embodiment of the present application provides a data transmission device based on an anti-packet loss strategy, including a packet loss prediction module, a prediction analysis module, a quality determination module and a data transmission module, wherein:

[0011] The packet loss prediction module is configured to determine the reception success information of a first preset number of transmitted data packets in a data transmission link, and determine a link predicted packet loss rate of a second preset number of to-be-transmitted data packets according to the reception success information.

[0012] The prediction analysis module is configured to determine a first predicted retransmission packet number and a first predicted introduced latency for data transmission based on an automatic repeat request strategy, a first predicted redundancy and a second predicted introduced latency for data transmission based on a forward error correction strategy, and a second predicted retransmission packet number, a second predicted redundancy and a third predicted introduced latency for data transmission based on a hybrid automatic repeat request strategy according to the link predicted packet loss rate.

[0013] The quality determining module is configured to determine a first service quality score of the automatic repeat request strategy according to the link predicted packet loss rate, the first predicted number of retransmitted packets, and the first predicted introduced delay, determine a second service quality score of the forward error correction strategy according to the link predicted packet loss rate, the first predicted redundancy, and the second predicted introduced delay, and determine a third service quality score of the hybrid automatic repeat request strategy according to the link predicted packet loss rate, the second predicted number of retransmitted packets, the second predicted redundancy, and the third predicted introduced delay.

[0014] The data transmission module is configured to determine an anti-packet loss strategy corresponding to a minimum service quality score among the first service quality score, the second service quality score, and the third service quality score as a target anti-packet loss strategy, and perform data transmission based on the target anti-packet loss strategy, where the anti-packet loss strategy includes the automatic repeat request strategy, the forward error correction strategy, and the hybrid automatic repeat request strategy.

[0015] In a third aspect, an embodiment of the present application provides a data transmission device based on an anti-packet loss strategy, including a memory and one or more processors.

[0016] The memory is configured to store one or more programs.

[0017] When the one or more programs are executed by the one or more processors, the one or more processors implement the data transmission method based on the anti-packet loss strategy as described in the first aspect.

[0018] In a fourth aspect, an embodiment of the present application provides a non-volatile storage medium storing computer-executable instructions for performing the data transmission method based on the anti-packet loss strategy as described in the first aspect when executed by a computer processor.

[0019] In a fifth aspect, an embodiment of the present application provides a computer program product including a computer program stored in a computer-readable storage medium, and at least one processor of a device reads and executes the computer program from the computer-readable storage medium, so that the device performs the data transmission method based on the anti-packet loss strategy as described in the first aspect.

[0020] This application embodiment determines the predicted packet loss rate based on the successful reception information on the data transmission link, and determines the first predicted retransmission packet count and first predicted introduced delay of the automatic repeat request strategy, the first predicted redundancy and second predicted introduced delay of the forward error correction strategy, and the second predicted retransmission packet count, second predicted redundancy and third predicted introduced delay of the hybrid automatic repeat request strategy based on the predicted packet loss rate, the first predicted retransmission packet count and the first predicted introduced delay. It also determines the first quality of service score of the automatic repeat request strategy based on the predicted packet loss rate, the first predicted retransmission packet count and the first predicted introduced delay, and finally determines the second predicted introduced delay based on the predicted packet loss rate, the first predicted redundancy and the second predicted introduced delay. The delay determines the second Quality of Service (QoS) score for the forward error correction strategy, and the third QoS score, based on the link predicted packet loss rate, the second predicted number of retransmitted packets, the second predicted redundancy, and the third predicted delay, determines the hybrid automatic repeat request strategy. The anti-packet loss strategy corresponding to the minimum QoS score can be determined as the target anti-packet loss strategy, and data transmission can be performed based on the target anti-packet loss strategy. The packet loss rate of subsequent data packets to be sent can be accurately predicted in advance, and the QoS of different anti-packet loss strategies can be accurately predicted based on the predicted packet loss rate. The target anti-packet loss strategy used for data transmission can be determined based on the QoS, effectively improving the anti-packet loss effect of data transmission. Attached Figure Description

[0021] Figure 1 This is a flowchart of a data transmission method based on an anti-packet loss strategy provided in an embodiment of this application;

[0022] Figure 2 This is a flowchart of another data transmission method based on an anti-packet loss strategy provided in an embodiment of this application;

[0023] Figure 3 This is a schematic diagram of the structure of a data transmission device based on an anti-packet loss strategy provided in an embodiment of this application;

[0024] Figure 4 This is a schematic diagram of the structure of a data transmission device based on an anti-packet loss strategy provided in an embodiment of this application. Detailed Implementation

[0025] In order to make the purposes, technical solutions and advantages of the present application clearer, the specific embodiments of the present application are further described in detail below in combination with the drawings. It can be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application. In addition, it should be noted that, for the convenience of description, only the parts related to the present application are shown in the drawings, but not all the contents. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be implemented in parallel, concurrently or simultaneously. In addition, the order of the operations can be rearranged. The above processes can be terminated when the operations are completed, but can also have additional steps not included in the drawings. The above processes can correspond to methods, functions, procedures, subroutines, subprograms, etc.

[0026] The data transmission method based on the anti-packet loss strategy provided by the present application can be applied to real-time audio and video communication scenarios (such as video conferencing, real-time voice calls, etc.), aiming to accurately predict the packet loss rate of the subsequent to-be-sent data packets in advance, accurately predict the service quality of different anti-packet loss strategies based on the predicted packet loss rate, and determine the target anti-packet loss strategy used for data transmission according to the service quality, thereby effectively improving the anti-packet loss effect of data transmission.

[0027] In the anti-packet loss scheme of the related art, the anti-packet loss strategy mainly includes automatic repeat request, forward error correction, hybrid automatic repeat request, etc. Among them, the automatic repeat request strategy relying on the feedback (ACK / NACK) of the receiving party will introduce a larger transmission delay when the round-trip delay is large, and the quality of service is low in some application scenarios sensitive to delay. The proactive retransmission technology not relying on the feedback of the receiving party usually sends multiple copies of some important packets or all packets to achieve the purpose of anti-packet loss and delay reduction, but this easily causes the problem of bandwidth waste (especially in bandwidth-limited scenarios). The forward error correction strategy applies a fault-tolerant coding algorithm to the source data packet to generate a redundant data packet, which can recover the lost source data packet with relatively low delay when a data packet is lost. The key of the forward error correction strategy lies in predicting the packet loss rate in advance to determine whether to encode the redundant packet in advance and the size of the redundancy. The method for predicting the packet loss rate in the related art usually uses the average packet loss rate or the maximum packet loss rate in a past time window as an index to replace the future packet loss rate. This prediction method needs to feedback to the sending end after the receiving end detects the occurrence of packet loss, which will cause a lag in information. That is, the sending end cannot know the occurrence of packet loss in the network in the first time, and the packet loss rate changes over time. If the predicted packet loss rate is too small, the redundancy is not enough, and the recovery effect is poor. If the predicted packet loss rate is too large, the redundancy is too high, which wastes traffic and bandwidth and even aggravates congestion. The accuracy of the packet loss rate prediction will directly affect the quality of service of the forward error correction strategy, and the anti-packet loss effect of data transmission is poor. Based on this, the embodiment of the present application provides a data transmission method based on an anti-packet loss strategy to solve the technical problem of poor anti-packet loss effect of existing data transmission.

[0028] Figure 1 A flowchart of the data transmission method based on the anti-packet loss strategy provided by the embodiment of the present application is given. The data transmission method based on the anti-packet loss strategy provided by the embodiment of the present application can be executed by a data transmission device based on the anti-packet loss strategy. The data transmission device based on the anti-packet loss strategy can be realized by hardware and / or software and integrated in a data transmission device based on the anti-packet loss strategy.

[0029] The data transmission method based on the anti-packet loss strategy executed by the data transmission device based on the anti-packet loss strategy is described below. Referring to Figure 1 , the data transmission method based on the anti-packet loss strategy comprises:

[0030] S110: determining the reception success information of a first preset number of sent data packets in a data transmission link, and determining the link predicted packet loss rate of a subsequent second preset number of to-be-sent data packets according to the reception success information.

[0031] For example, during the process of transmitting data in the data transmission link, the reception of each transmitted data packet is recorded, i.e., whether the transmitted data packet is successfully received by the receiving end, and the corresponding reception success information of each transmitted data packet is generated to record whether the corresponding transmitted data packet is successfully received by the receiving end. The reception success information of the latest recorded first preset number (the value range of the first preset number can be 2-20, for example, the first preset number is set to 5, 10, etc.) of transmitted data packets can be determined.

[0032] In one embodiment, when the current to-be-transmitted data packet needs to be transmitted, the link prediction packet loss rate of the subsequent second preset number (the second preset number can be consistent with the first preset number, and the value range of the second preset number can be 2-20, for example, the second preset number is set to 5, 10, etc.) of to-be-transmitted data packets is predicted according to the reception success information of the first preset number of transmitted data packets (the first preset number of transmitted data packets before the current to-be-transmitted data packet) determined above. The link prediction packet loss rate is the predicted packet loss rate of the second preset number of to-be-transmitted data packets in the data transmission link. Optionally, the link prediction packet loss rate can be determined according to the second preset number of data packet prediction packet loss rates, for example, the average value, weighted average value, median, etc. of the second preset number of data packet prediction packet loss rates is taken as the link prediction packet loss rate, and the data packet prediction packet loss rate is the predicted packet loss rate corresponding to the to-be-transmitted data packet. The current to-be-transmitted data packet can be the first to-be-transmitted data packet in the second preset number of to-be-transmitted data packets.

[0033] Optionally, the prediction of the data packet prediction packet loss rate and / or the link prediction packet loss rate can be based on the change trend of the data packet loss rate and / or the link loss rate reflected by the reception success information of the first preset number of transmitted data packets, for example, based on the fitting curve (for example, the fitting curve obtained based on the least square method) of the data packet loss rate and / or the link loss rate reflected by the reception success information of the first preset number of transmitted data packets, and based on the fitting curve to determine the data packet prediction packet loss rate and / or the link prediction packet loss rate of the subsequent second preset number of to-be-transmitted data packets. The data packet prediction packet loss rate and / or the link prediction packet loss rate can also be predicted by a trained neural network model, for example, the reception success information of the first preset number of transmitted data packets is sent to the neural network model, and the data packet prediction packet loss rate and / or the link prediction packet loss rate of the second preset number of to-be-transmitted data packets is output by the neural network model.

[0034] S120: determining, according to the link predicted packet loss rate, a first predicted number of retransmission packets and a first predicted introduced delay for data transmission based on an automatic repeat request strategy, a first predicted redundancy and a second predicted introduced delay for data transmission based on a forward error correction strategy, and a second predicted number of retransmission packets, a second predicted redundancy and a third predicted introduced delay for data transmission based on a hybrid automatic repeat request strategy.

[0035] For example, after determining the link predicted packet loss rate, a first predicted number of retransmission packets and a first predicted introduced delay for data transmission based on an automatic repeat request strategy at the link predicted packet loss rate, a first predicted redundancy and a second predicted introduced delay for data transmission based on a forward error correction strategy at the link predicted packet loss rate, and a second predicted number of retransmission packets, a second predicted redundancy and a third predicted introduced delay for data transmission based on a hybrid automatic repeat request strategy at the link predicted packet loss rate can be determined according to the link predicted packet loss rate.

[0036] The predicted number of retransmission packets (including the first predicted number of retransmission packets and the second predicted number of retransmission packets) can be understood as the estimated number of retransmission packets needed for transmitting the current data packet based on an automatic repeat request strategy. The predicted redundancy (including the first predicted redundancy and the second predicted redundancy) can be understood as the estimated redundancy of the number of redundancy packets in the transmitted data packets needed for transmitting the current data packet based on a forward error correction strategy (i.e. the ratio of the number of redundancy packets to the sum of the number of source data packets and the number of redundancy packets). The predicted introduced delay (including the first to third predicted introduced delays) can be understood as the additional data transmission delay caused by the corresponding anti-packet loss strategy (automatic repeat request strategy, forward error correction strategy or hybrid automatic repeat request strategy).

[0037] S130: determining, according to the link predicted packet loss rate, the first predicted number of retransmission packets and the first predicted introduced delay, a first quality of service score for an automatic repeat request strategy, according to the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced delay, a second quality of service score for a forward error correction strategy, and according to the link predicted packet loss rate, the second predicted number of retransmission packets, the second predicted redundancy and the third predicted introduced delay, a third quality of service score for a hybrid automatic repeat request strategy.

[0038] For example, after determining the first predicted retransmission packet number and the first predicted introduced latency corresponding to the automatic repeat request strategy, a first service quality score of the automatic repeat request strategy can be calculated according to the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced latency. The first service quality score is positively correlated with the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced latency, and the smaller the first service quality score is, the better the corresponding service quality is. For example, corresponding weights can be set for the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced latency, and a weighted sum of the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced latency based on the weights can be obtained to obtain the first service quality score.

[0039] In one embodiment, after determining the first predicted redundancy and the second predicted introduced latency corresponding to the forward error correction strategy, a second service quality score of the forward error correction strategy can be calculated according to the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced latency. The second service quality score is positively correlated with the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced latency, and the smaller the second service quality score is, the better the corresponding service quality is. For example, corresponding weights can be set for the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced latency, and a weighted sum of the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced latency based on the weights can be obtained to obtain the second service quality score.

[0040] In one embodiment, after determining the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced latency corresponding to the hybrid automatic repeat request strategy, a third service quality score of the hybrid automatic repeat request strategy can be calculated according to the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced latency. The third service quality score is positively correlated with the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced latency, and the smaller the third service quality score is, the better the corresponding service quality is. For example, corresponding weights can be set for the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced latency, and a weighted sum of the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced latency based on the weights can be obtained to obtain the third service quality score.

[0041] In one embodiment, when calculating the service quality score, the round-trip delay of the data transmission link (which can be fed back by the receiving end), the round-trip delay jitter (the jitter of the round-trip delay fed back by the receiving end), the predicted bandwidth detected by the data transmission link, the bandwidth utilization, etc. can also be determined, and the round-trip delay, the round-trip delay jitter, the predicted bandwidth, the bandwidth utilization, etc. are combined into the calculation of the service quality score. For example, the round-trip delay, the round-trip delay jitter, the inverse of the predicted bandwidth, the bandwidth utilization, etc. are set with corresponding weights, and are weighted and summed with the predicted packet loss rate, the predicted number of retransmitted packets, the predicted redundancy, the predicted introduced delay, etc. to obtain the corresponding service quality score.

[0042] S140: Determine the anti-packet loss strategy corresponding to the minimum service quality score among the first service quality score, the second service quality score and the third service quality score as the target anti-packet loss strategy, and perform data transmission based on the target anti-packet loss strategy. The anti-packet loss strategy includes an automatic repeat request strategy, a forward error correction strategy and a hybrid automatic repeat request strategy.

[0043] For example, after determining the first service quality score of the automatic repeat request strategy, the second service quality score of the forward error correction strategy and the third service quality score of the hybrid automatic repeat request strategy, the minimum service quality score among the first service quality score, the second service quality score and the third service quality score is determined, and the anti-packet loss strategy corresponding to the minimum service quality score is determined as the target anti-packet loss strategy, and the current data packet to be sent is transmitted based on the target anti-packet loss strategy.

[0044] Among them, the anti-packet loss strategy includes an automatic repeat request strategy, a forward error correction strategy and a hybrid automatic repeat request strategy, that is, the service quality score of the automatic repeat request strategy, the forward error correction strategy and the hybrid automatic repeat request strategy is the smallest as the target anti-packet loss strategy, so as to obtain better service quality.

[0045] According to the above, by determining the link predicted packet loss rate according to the reception success information of the data transmission link, and determining the first predicted retransmission packet number and the first predicted introduced delay of the automatic retransmission request strategy, the first predicted redundancy and the second predicted introduced delay of the forward error correction strategy, and the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced delay of the hybrid automatic retransmission request strategy according to the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced delay are used to determine the first service quality score of the automatic retransmission request strategy, the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced delay are used to determine the second service quality score of the forward error correction strategy, and the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced delay are used to determine the third service quality score of the hybrid automatic retransmission request strategy, the anti-packet loss strategy corresponding to the minimum service quality score is determined as the target anti-packet loss strategy, and data transmission is performed based on the target anti-packet loss strategy, the packet loss rate of the subsequent to-be-sent data packet is accurately predicted in advance, the service quality of different anti-packet loss strategies is accurately predicted based on the predicted packet loss rate, and the target anti-packet loss strategy used for data transmission is determined according to the service quality, thereby effectively improving the anti-packet loss effect of data transmission.

[0046] On the basis of the above embodiment, Figure 2 A flowchart of another anti-packet loss strategy-based data transmission method provided by the embodiment of the application is given, and the anti-packet loss strategy-based data transmission method is a specific embodiment of the above anti-packet loss strategy-based data transmission method. Referring to Figure 2 The anti-packet loss strategy-based data transmission method comprises the following steps.

[0047] S210: Determine the reception success information of a first preset number of sent data packets in a data transmission link, and input the reception success information into a trained packet loss prediction model, and output the data packet predicted packet loss rate of a subsequent second preset number of to-be-sent data packets by the packet loss prediction model according to the reception success information.

[0048] S220: Determine the link predicted packet loss rate of the data transmission link according to the data packet predicted packet loss rate.

[0049] In an embodiment, the scheme can predict the packet loss rate of the data packet to be sent by training a completed packet loss prediction model. The packet loss prediction model provided by the scheme can be built and trained based on a neural network model, for example, a long short-term memory (LSTM) network can be used to build and train the packet loss prediction model. As a neural network specially designed for processing and predicting sequence data, the long short-term memory network can effectively capture the time dependence in the data. Since the network packet loss rate is usually a time series data that changes over time, the long short-term memory network can learn the historical trend of the packet loss rate, so as to consider these time sequence characteristics when predicting. Moreover, the change of the network packet loss rate can be affected by short-term fluctuations (such as instantaneous network congestion) or long-term trends (such as high load in a specific period of time). The long short-term memory network can model both short-term and long-term dependencies through its unique memory unit. Meanwhile, the change of the packet loss rate can be complex, and the long short-term memory network can handle these nonlinear characteristics of the change and adapt to the fluctuations of the data, thereby achieving accurate prediction of the network packet loss rate.

[0050] In an embodiment, when training the packet loss prediction model, the training data can be collected first, for example, historical packet receiving data in data transmission (such as offline weak network test data and online statistical data) is collected, the historical packet receiving data is uniformly processed into a packet receiving sequence of the source data packet at the receiving end, and the packet receiving sequence is sorted in chronological order. For example, 5 historical packet receiving data is grouped as a group, 0 represents that the corresponding source data packet of the sequence is not received, and 1 represents that the corresponding source data packet of the sequence is received, for example, “11100” represents that the first to third source data packets are received and the fourth to fifth source data packets are not received, a label is marked for each group of data, the label is the packet receiving condition of the next group of source data packets, for example, if the packet receiving condition of the next group of source data packets is “01100”, the label is “01100”, indicating that in the next group of source data packets, the second and third source data packets are received and the other source data packets are not received, and the above collected data is used as the training data. The packet loss prediction model is constructed based on the LSTM network, and the number of layers of the input layer, the output layer and the hidden layer of the LSTM network is configured, for example, the number of layers of the input layer can be configured to 5, the number of layers of the hidden layer can be configured to 5, and the number of layers of the output layer can be configured to 5. The training data collected above is used to train the packet loss prediction model, the training data can be divided into a training set, a validation set and a test set, the target of the packet loss prediction model is to predict the packet receiving record at the next moment, then the number of packet losses in the packet receiving record is counted, and then the packet loss rate of each data packet at the next moment can be calculated, the training set is used to train the packet loss prediction model, and the validation set is used to evaluate the performance of the model and prevent overfitting during the training of the packet loss prediction model. The performance of the packet loss prediction model is evaluated on the test set (for example, based on evaluation indexes such as mean square error (MSE) and mean absolute error (MAE)), the model parameters are adjusted according to the evaluation result, and the packet loss prediction model is optimized. The trained packet loss prediction model is used for real-time prediction of future packet loss rate online, the packet receiving record fed back by the receiving end is processed into training data and labeled, and input into the packet loss prediction model for real-time training and updating of the model, so that the packet loss prediction model can adapt to the real-time changes of the online network and improve the prediction accuracy. After the training of the packet loss prediction model is completed, the packet loss prediction model can be configured in the data transmission device based on the anti-packet loss strategy to predict the network packet loss rate.

[0051] For example, the receiving success information of the first preset number of sent data packets in the data transmission link is obtained, for example, the receiving success information of the first preset number of sent data packets is recorded in the form of a packet receiving sequence, and the receiving success information is input into the trained packet loss prediction model, and the packet loss prediction model outputs the data packet prediction packet loss rate of the subsequent second preset number of to-be-sent data packets according to the receiving success information.

[0052] After determining the packet prediction packet loss rate, the link prediction packet loss rate of the data transmission link can be determined according to the second preset number of packet prediction packet loss rates, for example, the average value of the second preset number of packet prediction packet loss rates is taken as the link prediction packet loss rate of the data transmission link.

[0053] For example, after inputting the first preset number of sent data packet reception success information into the packet loss prediction model, the data packet prediction packet loss rate of the 10 to-be-sent data packets output by the packet loss prediction model corresponds to the prediction sequence: 0.1, 0.01, 0.01, 0.02, 0.06, 0.8, 0.2, 0.3, 0.01, 0.49, and the data packet prediction packet loss rate of the current to-be-sent data packet is the first data of the prediction sequence, that is, 0.1, and the link prediction packet loss rate is: (0.1+0.01+0.01+0.02+0.06+0.8+0.2+0.3+0.01+0.49) / 10=0.2. The present scheme accurately predicts the data packet prediction packet loss rate of the subsequent second preset number of to-be-sent data packets according to the reception success information through the packet loss prediction model, and determines the link prediction packet loss rate of the data transmission link according to the data packet prediction packet loss rate, which can more accurately predict the service quality of different anti-packet loss strategies under the predicted network packet loss rate, and improve the anti-packet loss effect of data transmission.

[0054] S230: determining, according to the link prediction packet loss rate, a first predicted retransmission packet number and a first predicted introduced delay for data transmission based on an automatic repeat request strategy, a first predicted redundancy and a second predicted introduced delay for data transmission based on a forward error correction strategy, and a second predicted retransmission packet number, a second predicted redundancy, and a third predicted introduced delay for data transmission based on a hybrid automatic repeat request strategy.

[0055] In one possible embodiment, the automatic repeat request strategy provided by the present scheme can be a passive automatic repeat request strategy and / or an active automatic repeat request strategy, wherein the active automatic repeat request strategy refers to an automatic repeat request strategy in which the sending end retransmits the data packet only when receiving an explicit retransmission request (such as NACK or missing sequence number prompt) from the receiving end, and the active automatic repeat request strategy refers to an automatic repeat request strategy in which the sending end actively triggers the retransmission of the data packet when no acknowledgement (ACK) is received from the receiving end or an explicit denial (NACK) is received.

[0056] Correspondingly, the first predicted retransmission packet quantity can include a first predicted passive retransmission packet quantity corresponding to a passive automatic repeat request strategy and / or a first predicted active retransmission packet quantity corresponding to an active automatic repeat request strategy, and the type of the first predicted retransmission packet quantity can be determined according to the preset automatic repeat request strategy type. Correspondingly, the first service quality score includes a passive retransmission service quality score determined according to the link predicted packet loss rate, the first predicted passive retransmission packet quantity, and the first predicted introduced delay, and / or an active retransmission service quality score determined according to the link predicted packet loss rate, the first predicted active retransmission packet quantity, and the first predicted introduced delay. Subsequently, the target anti-packet loss strategy can be determined according to the anti-packet loss strategy corresponding to the minimum service quality score among the passive retransmission service quality score and / or the active retransmission service quality score, the second service quality score, and the third service quality score.

[0057] Based on this, the data transmission method based on the anti-packet loss strategy provided in the scheme can determine a first predicted retransmission packet quantity for data transmission based on an automatic repeat request strategy according to a link predicted packet loss rate. The first predicted retransmission packet quantity can be: calculating a first ratio of a natural logarithm of an expected packet loss rate and the link predicted packet loss rate, rounding up the first ratio, and determining the rounding up result as the first predicted passive retransmission packet quantity for data transmission based on a passive automatic repeat request strategy; and / or calculating a second ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate, calculating a first product of a difference between a first preset coefficient and a data packet predicted packet loss rate and a second preset coefficient, calculating an addition result of the second ratio and the first product, and rounding up the addition result, and determining the rounding up result as the first predicted active retransmission packet quantity for data transmission based on an active automatic repeat request strategy.

[0058] For example, for the passive automatic repeat request strategy, the first ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate is calculated, the first ratio is rounded up, and the rounding up result can be determined as the first predicted passive retransmission packet quantity for data transmission based on the passive automatic repeat request strategy. For example, the first predicted passive retransmission packet quantity can be determined based on the following formula:

[0059]

[0060] wherein n1 is the first predicted passive retransmission packet quantity, l_e is the expected packet loss rate, l_f is the link predicted packet loss rate, ln(·) is the natural logarithm, The rounding up is that if the recovered packet loss rate is lower than l_e, at least n1 retransmission packets need to be sent.

[0061] Wherein, for the passive automatic repeat request strategy, when receiving the NACK of the feedback of the meal break section, for the un-received data packet, firstly, the first predicted passive retransmission packet quantity is calculated according to the first ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate. It is determined whether the current data transmission link is in a congestion state. When the current network is in a non-congestion state, all of these retransmission packets can be sent at one time, and the introduced transmission delay (the first predicted introduced delay) is 0. When the current network is in a congestion state, the retransmission packets need to be sent after the congestion state is released, and the introduced transmission delay (the first predicted introduced delay) is the waiting time for waiting for the congestion state to be released. The scheme determines the first predicted passive retransmission packet quantity of the passive automatic repeat request strategy according to the first ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate, accurately predicts the retransmission packet quantity required to be retransmitted by the passive automatic repeat request strategy, more accurately predicts the service quality of the passive automatic repeat request strategy under the predicted network packet loss rate, and more accurately determines the target anti-packet loss strategy.

[0062] In one embodiment, for the active automatic repeat request strategy, the second ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate is calculated, the first product of the difference between the first preset coefficient and the data packet predicted packet loss rate and the second preset coefficient is calculated, the addition result of the second ratio and the first product is calculated, and the upward rounding of the addition result is performed. The upward rounding result can be determined as the first predicted active retransmission packet quantity for data transmission based on the active automatic repeat request strategy. For example, the first predicted active retransmission packet quantity of a to-be-sent data packet can be determined based on the following formula:

[0063] n2=[p*(ln(l_e) / ln(l_f))+(a1-p)*a2]

[0064] Wherein, n2 is the first predicted active retransmission packet quantity, p is the data packet predicted packet loss rate of the current to-be-sent data packet, a1 is the first preset coefficient, for example, a1=1, a2 is the second preset coefficient, for example, a2=1, l_e is the expected packet loss rate, l_f is the link predicted packet loss rate, ln(·) is the natural logarithm, is the upward rounding, that is, if the recovered packet loss rate is lower than l_e, at least n2 retransmission packets need to be sent.

[0065] For the proactive automatic repeat request strategy, the first predicted proactive retransmission packet number of each to-be-sent data packet that needs to be retransmitted is calculated respectively according to the data packet prediction packet loss rate of each to-be-sent data packet. It is determined whether the current data transmission link is in a congestion state. When the current network is in a non-congestion state, all of these retransmission packets can be sent at one time, and the transmission delay introduced (the first predicted introduced delay) is 0. When the current network is in a congestion state, the retransmission packets are sent after the congestion state is released, and the transmission delay introduced (the first predicted introduced delay) is the waiting time for the congestion state to be released. The first predicted proactive retransmission packet number is accurately predicted according to the second ratio of the natural logarithm of the expected packet loss rate and the link prediction packet loss rate, and the first product of the difference between the first preset coefficient and the data packet prediction packet loss rate and the second preset coefficient, which can more accurately predict the service quality of the proactive automatic repeat request strategy under the predicted network packet loss rate, and more accurately determine the target anti-packet loss strategy.

[0066] In one possible embodiment, the data transmission method based on the anti-packet loss strategy provided by the present scheme determines the first predicted redundancy and the second predicted introduced delay of data transmission based on the forward error correction strategy according to the link prediction packet loss rate, comprising:

[0067] S231: determining the acceptable introduced delay according to the data round-trip delay of the data transmission link, and determining the number of redundant packet groups of the forward error correction strategy according to the acceptable introduced delay and the redundant packet duration.

[0068] S232: determining the first predicted redundancy of data transmission based on the forward error correction strategy according to the number of redundant packet groups and the link prediction packet loss rate.

[0069] S233: determining the second predicted introduced delay of data transmission based on the forward error correction strategy according to the number of redundant packet groups and the redundant packet duration.

[0070] For the forward error correction strategy, the data round-trip delay of the data transmission link is determined, and the acceptable introduced delay is determined according to the data round-trip delay. Different acceptable introduced delays can be configured for different data round-trip delays, or the acceptable introduced delay can be calculated according to the data round-trip delay, for example, the product of the data round-trip delay and a preset proportion (for example, 0.8) is taken as the acceptable introduced delay. When the product of the data round-trip delay and the preset proportion is less than the minimum acceptable introduced delay, the minimum acceptable introduced delay can be taken as the acceptable introduced delay.

[0071] In one embodiment, after determining the acceptable introduction delay, the number of redundant packet groups of the forward error correction strategy can be determined according to the acceptable introduction delay and the redundant packet duration, for example, the acceptable introduction delay is divided by the redundant packet duration (i.e. the duration of a single data packet), and the result of the division is rounded down to obtain the number of redundant packet groups of the forward error correction strategy.

[0072] In one embodiment, after determining the number of redundant packet groups of the forward error correction strategy, the first predicted redundancy for data transmission based on the forward error correction strategy can be determined according to the number of redundant packet groups and the predicted packet loss rate of the link. For example, a third preset coefficient (e.g. 0.5) can be added to the product of the number of redundant packet groups and the predicted packet loss rate of the link, the result of the addition is divided by the difference between 1 and the predicted packet loss rate of the link, and the result of the division is rounded up to obtain the number of redundant packets that need to be encoded, and the ratio of the number of redundant packets that need to be encoded to the number of redundant packet groups can be taken as the first predicted redundancy.

[0073] For example, it is required that:

[0074] (N+K)*(1-l_f)≥N+a3

[0075] Wherein, N is the number of redundant packet groups, K is the number of redundant packets that need to be encoded, l_f is the predicted packet loss rate of the link, and a3 is the third preset coefficient.

[0076] Based on this, the number of redundant packets that need to be encoded can be determined based on the following formula:

[0077] K=(a3+N*l_f) / (1-l_f)

[0078] Correspondingly, the first predicted redundancy is K / N.

[0079] In one embodiment, after determining the number of redundant packet groups of the forward error correction strategy, the second predicted introduction delay for data transmission based on the forward error correction strategy can be determined according to the number of redundant packet groups and the redundant packet duration, for example, the product of the number of redundant packet groups and the redundant packet duration is taken as the second predicted introduction delay. The scheme determines the acceptable introduction delay according to the data round-trip time of the data transmission link, determines the number of redundant packet groups of the forward error correction strategy according to the acceptable introduction delay and the redundant packet duration, accurately predicts the first predicted redundancy according to the number of redundant packet groups and the predicted packet loss rate of the link, and accurately predicts the second predicted introduction delay according to the number of redundant packet groups and the redundant packet duration, which can more accurately predict the service quality of the forward error correction strategy under the predicted network packet loss rate, and more accurately determine the target anti-packet loss strategy.

[0080] In one possible implementation, the anti-packet loss strategy based data transmission method provided in the present solution determines a second predicted retransmission packet number, a second predicted redundancy and a third predicted introduced delay for data transmission based on a hybrid automatic repeat request strategy according to a link predicted packet loss rate, comprising:

[0081] S234: determining a first weight coefficient of an automatic repeat request and a second weight coefficient of a forward error correction in the hybrid automatic repeat request strategy in the case of data transmission based on the hybrid automatic repeat request strategy.

[0082] S235: determining a second predicted retransmission packet number and a fourth predicted introduced delay of the automatic repeat request according to the first weight coefficient and the link predicted packet loss rate.

[0083] S236: determining a second predicted redundancy and a fifth predicted introduced delay of the forward error correction according to the second weight coefficient and the link predicted packet loss rate.

[0084] S237: determining the third predicted introduced delay according to the fourth predicted introduced delay and the fifth predicted introduced delay.

[0085] In one implementation, the hybrid automatic repeat request strategy can be based on a hybrid manner of the automatic repeat request and the forward error correction for retransmission of data packets and transmission of redundant packets, and the corresponding weight coefficients of the automatic repeat request and the forward error correction under the hybrid automatic repeat request strategy can be preset. For example, it is determined that the first weight coefficient of the automatic repeat request and the second weight coefficient of the forward error correction in the hybrid automatic repeat request strategy are determined when data transmission is based on the hybrid automatic repeat request strategy. The sum of the first weight coefficient and the second weight coefficient is 1. Optionally, when the hybrid automatic repeat request strategy is only anti-packet loss processing through the automatic repeat request, the first weight coefficient is 1 and the second weight coefficient is 0, and when the hybrid automatic repeat request strategy is only anti-packet loss processing through the forward error correction, the first weight coefficient is 0 and the second weight coefficient is 1.

[0086] Optionally, the first weight coefficient of automatic repeat request and the second weight coefficient of forward error correction can be determined according to the current data round-trip time and the link predicted packet loss rate. For example, assuming that the second weight coefficient is a, the second weight coefficient is b, a+b=1, a≤1, b≤1, in the network environment of low data round-trip time+low link predicted packet loss rate, the first weight coefficient can be set to be higher than the second weight coefficient, in the network environment of high data round-trip time+low link predicted packet loss rate, the second weight coefficient can be set to be higher than the first weight coefficient, in the network environment of high data round-trip time+high link predicted packet loss rate, the first weight coefficient can be set to be consistent or close to the second weight coefficient, or the second weight coefficient can be set to be slightly higher than the first weight coefficient, such as the second weight coefficient is 0.6, the first weight coefficient is 0.4, in the network environment of low data round-trip time+high link predicted packet loss rate, the first weight coefficient can be set to be higher than the second weight coefficient, such as the second weight coefficient is 0.3, the first weight coefficient is 0.7.

[0087] In one embodiment, the second predicted retransmission packet quantity of automatic repeat request and the fourth predicted introduced delay are determined according to the first weight coefficient and the link predicted packet loss rate determined above. For example, based on the ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate, the ratio is rounded up, and the rounded-up result is determined as the first predicted passive retransmission packet quantity of the passive automatic repeat request strategy; and / or the ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate is calculated, the product of the difference between the first preset coefficient and the data packet predicted packet loss rate and the second preset coefficient is calculated, the addition result of the ratio and the product is calculated, and the addition result is rounded up, and the rounded-up result is determined as the first predicted active retransmission packet quantity of the active automatic repeat request strategy. The second predicted retransmission packet quantity is obtained by multiplying the first weight coefficient and the first predicted passive retransmission packet quantity and / or the first predicted active retransmission packet quantity. Wherein, when the current network is in a non-congestion state, the fourth predicted introduced delay is 0, and when the current network is in a congestion state, the fourth predicted introduced delay is the waiting time for waiting for the congestion state to be eliminated.

[0088] In one embodiment, the second predicted redundancy of forward error correction and the fifth predicted introduced delay are determined according to the second weight coefficient and the link predicted packet loss rate determined above. For example, the acceptable introduced delay is determined according to the data round-trip time of the data transmission link, the number of redundant packet groups of the forward error correction strategy is determined according to the acceptable introduced delay and the redundant packet duration, the first predicted redundancy of data transmission based on the forward error correction strategy is determined according to the number of redundant packet groups and the link predicted packet loss rate, the product of the second weight coefficient and the first predicted redundancy is taken as the second predicted redundancy, and the product of the number of redundant packet groups and the redundant packet duration is taken as the fifth predicted introduced delay. In one embodiment, the fourth predicted introduced delay and the fifth predicted introduced delay are added to obtain the third predicted introduced delay.

[0089] The scheme can more accurately predict the service quality of the hybrid automatic repeat request strategy under the predicted network packet loss rate, and more accurately determine the target anti-packet loss strategy, by determining the first weight coefficient of the automatic repeat request and the second weight coefficient of the forward error correction, determining the second predicted retransmission packet number and the fourth predicted introduced delay of the automatic repeat request according to the first weight coefficient and the link predicted packet loss rate, determining the second predicted redundancy and the fifth predicted introduced delay of the forward error correction according to the second weight coefficient and the link predicted packet loss rate, and determining the third predicted introduced delay according to the fourth predicted introduced delay and the fifth predicted introduced delay.

[0090] S240: determining a first service quality score of the automatic repeat request strategy according to the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced delay, determining a second service quality score of the forward error correction strategy according to the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced delay, and determining a third service quality score of the hybrid automatic repeat request strategy according to the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced delay.

[0091] In one possible embodiment, a delay cost score can be determined according to the predicted introduced delay, the round-trip delay jitter, the network available bandwidth of the data transmission link and the data sending waiting time length, a traffic cost score can be determined according to the predicted retransmission packet number and the original packet number, a quality index score can be determined according to the link predicted packet loss rate and the bandwidth utilization of the data transmission link, and a weighted sum result can be obtained by performing weighted sum processing on the delay cost score, the traffic cost score and the quality index score according to a preset delay cost weight, a preset traffic cost weight and a preset quality index weight, and the weighted sum result is determined as the service quality score of the corresponding anti-packet loss strategy. For example, the service quality score can be calculated based on the following formula:

[0092] QoS=w1*f_delay+w2*f_cost+w3*f_other

[0093] f_delay=delay+k*jitter+j*1 / bw+wait

[0094] f_cost=a*T+b*R

[0095] f_other=c*P_loss-d*U_bw

[0096] Wherein, QoS is a service quality score, w1 is a preset delay cost weight, w2 is a preset traffic cost weight, w3 is a preset quality index weight, f_delay is a delay cost score, f_cost is a traffic cost score, f_other is a quality index score, delay is a predicted introduced delay, jitter is a round-trip delay jitter, bw is a network available bandwidth of a data transmission link, wait is a data sending waiting time length, T is an original packet quantity, R is a predicted retransmission packet quantity, a quantity of redundant packets sent based on a predicted redundancy, or a sum of the predicted retransmission packet quantity and the quantity of redundant packets sent based on the predicted redundancy, P_loss is a link predicted packet loss rate, U_bw is a bandwidth utilization rate of the data transmission link, k is a round-trip delay jitter weight, j is a network available bandwidth weight, a is an original packet quantity weight, b is a strategy weight, c is a link predicted packet loss rate weight, and d is a bandwidth utilization rate weight. The preset delay cost weight, the preset traffic cost weight, and the preset quality index weight can be adjusted in real time according to a network state or application demand, for example, the preset quality index weight is increased in a high packet loss rate environment, and the preset delay cost weight is increased in a high delay environment. The model can also be trained using historical data, and the appropriate weights are trained based on the historical data, and then the preset delay cost weight, the preset traffic cost weight, and the preset quality index weight are dynamically optimized online according to the predicted network situation.

[0097] Correspondingly, the anti-packet loss strategy-based data transmission method provided in the scheme can determine a first service quality score of an automatic repeat request strategy according to a link predicted packet loss rate, a first predicted retransmission packet quantity, and a first predicted introduced delay, which can be: determining a first delay cost score according to the first predicted introduced delay, a round-trip delay jitter, a network available bandwidth of a data transmission link, and a data sending waiting time length, determining a first traffic cost score according to the first predicted retransmission packet quantity and an original packet quantity, and determining a first quality index score according to the link predicted packet loss rate and a bandwidth utilization rate of the data transmission link; performing weighted summation processing on the first delay cost score, the first traffic cost score, and the first quality index score according to a preset delay cost weight, a preset traffic cost weight, and a preset quality index weight to obtain a first weighted summation result, and determining the first weighted summation result as the first service quality score of the automatic repeat request strategy.

[0098] The first predicted introduced delay, the product of the round-trip delay jitter and the round-trip delay jitter weight, the inverse of the network available bandwidth of the data transmission link and the network available bandwidth weight, and the data sending waiting time are added to obtain a first delay cost score. The product of the first predicted retransmission packet number and the strategy weight, and the product of the original packet number and the original packet number weight are added to obtain a first traffic cost score. The product of the link predicted packet loss rate and the link predicted packet loss rate weight, and the product of the bandwidth utilization of the data transmission link and the bandwidth utilization weight are added to obtain a first quality indicator score.

[0099] In one embodiment, the product of the first delay cost score and the preset delay cost weight, the product of the first traffic cost score and the preset traffic cost weight, and the product of the first quality indicator score and the preset quality indicator weight are added to obtain a first weighted sum result, and the first weighted sum result is taken as the first quality of service score of the automatic repeat request strategy. The preset delay cost weight, the preset traffic cost weight, and the preset quality indicator weight are used to weight and sum the first delay cost score, the first traffic cost score, and the first quality indicator score to obtain the first quality of service score, so that the quality of service score of the automatic repeat request strategy under the predicted network packet loss rate can be more accurately predicted, and the target anti-packet loss strategy can be more accurately determined.

[0100] The method for data transmission based on the anti-packet loss strategy provided by the scheme is characterized in that a second quality of service score of a forward error correction strategy is determined according to a link predicted packet loss rate, a first predicted redundancy, and a second predicted introduced delay, and the method comprises the following steps: a second delay cost score is determined according to the second predicted introduced delay, a round-trip delay jitter, a network available bandwidth of a data transmission link, and a data sending waiting time; a second traffic cost score is determined according to the first predicted redundancy and an original packet number; and a second quality indicator score is determined according to the link predicted packet loss rate and a bandwidth utilization of the data transmission link. The second delay cost score, the second traffic cost score, and the second quality indicator score are subjected to weighted sum processing according to a preset delay cost weight, a preset traffic cost weight, and a preset quality indicator weight to obtain a second weighted sum result, and the second weighted sum result is determined as the second quality of service score of the forward error correction strategy.

[0101] The second predicted introduced delay, the product of the round-trip delay jitter and the round-trip delay jitter weight, the inverse of the network available bandwidth of the data transmission link and the network available bandwidth weight, and the data sending waiting time are added to obtain a second delay cost score. The product of the number of redundant packets sent based on the first predicted redundancy and the strategy weight, and the product of the original packet number and the original packet number weight are added to obtain a second traffic cost score. The product of the link predicted packet loss rate and the link predicted packet loss rate weight, and the product of the bandwidth utilization of the data transmission link and the bandwidth utilization weight are added to obtain a second quality indicator score.

[0102] In one embodiment, the product of the second latency cost score and the preset latency cost weight, the product of the second traffic cost score and the preset traffic cost weight, and the product of the second quality indicator score and the preset quality indicator weight are added to obtain a first weighted sum result, and the first weighted sum result is taken as the second service quality score of the automatic repeat request strategy. The second service quality score is obtained by weighting and summing the second latency cost score, the second traffic cost score and the second quality indicator score through the preset latency cost weight, the preset traffic cost weight and the preset quality indicator weight, which can more accurately predict the service quality score of the forward error correction strategy under the predicted network packet loss rate, and more accurately determine the target anti-packet loss strategy.

[0103] The anti-packet loss strategy-based data transmission method provided by the scheme is characterized in that determining the third service quality score of the hybrid automatic repeat request strategy according to the link predicted packet loss rate, the second predicted retransmission packet quantity, the second predicted redundancy and the third predicted introduced latency comprises: determining a third latency cost score according to the third predicted introduced latency, the round-trip delay jitter, the network available bandwidth of the data transmission link and the data sending waiting time length, determining a third traffic cost score according to the second predicted retransmission packet quantity, the second predicted redundancy and the original packet quantity, and determining a third quality indicator score according to the link predicted packet loss rate and the bandwidth utilization of the data transmission link; weighting and summing the third latency cost score, the third traffic cost score and the third quality indicator score according to a preset latency cost weight, a preset traffic cost weight and a preset quality indicator weight to obtain a third weighted sum result, and determining the third weighted sum result as the third service quality score of the hybrid automatic repeat request strategy.

[0104] For example, the third predicted introduced latency, the product of the round-trip delay jitter and the round-trip delay jitter weight, the inverse of the network available bandwidth of the data transmission link and the network available bandwidth weight, and the data sending waiting time length are added to obtain the third latency cost score. The product of the addition result of the number of redundant packets sent based on the second predicted redundancy and the second predicted retransmission packet quantity and the strategy weight, and the product of the original packet quantity and the original packet quantity weight are added to obtain the third traffic cost score. The product of the link predicted packet loss rate and the link predicted packet loss rate weight, and the product of the bandwidth utilization of the data transmission link and the bandwidth utilization weight are added to obtain the third quality indicator score.

[0105] In one embodiment, the product of the third delay cost score and the preset delay cost weight, the product of the third traffic cost score and the preset traffic cost weight, and the product of the third quality index score and the preset quality index weight are added to obtain a first weighted sum result, and the first weighted sum result is taken as the third service quality score of the automatic repeat request strategy. The third service quality score is obtained by weighting and summing the third delay cost score, the third traffic cost score and the third quality index score through the preset delay cost weight, the preset traffic cost weight and the preset quality index weight, which can more accurately predict the service quality score of the automatic repeat request strategy under the predicted network packet loss rate, and more accurately determine the target anti-packet loss strategy.

[0106] S250: determining the anti-packet loss strategy corresponding to the minimum service quality score among the first service quality score, the second service quality score and the third service quality score as the target anti-packet loss strategy, and performing data transmission based on the target anti-packet loss strategy, wherein the anti-packet loss strategy includes the automatic repeat request strategy, the forward error correction strategy and the hybrid automatic repeat request strategy.

[0107] According to the above, the link predicted packet loss rate is determined according to the reception success information of the data transmission link, the first predicted retransmission packet number and the first predicted introduced delay of the automatic repeat request strategy, the first predicted redundancy and the second predicted introduced delay of the forward error correction strategy, and the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced delay of the hybrid automatic repeat request strategy are determined according to the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced delay, the first service quality score of the automatic repeat request strategy is determined according to the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced delay, the second service quality score of the forward error correction strategy is determined according to the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced delay, and the third service quality score of the hybrid automatic repeat request strategy is determined according to the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced delay, the anti-packet loss strategy corresponding to the minimum service quality score can be determined as the target anti-packet loss strategy, and data transmission is performed based on the target anti-packet loss strategy, the packet loss rate of the subsequent to-be-sent data packet is accurately predicted in advance, the service quality of different anti-packet loss strategies is accurately predicted based on the predicted packet loss rate, and the target anti-packet loss strategy used for data transmission is determined according to the service quality, thereby effectively improving the anti-packet loss effect of data transmission. At the same time, the packet loss prediction model accurately predicts the data packet predicted packet loss rate of the subsequent second preset number of to-be-sent data packets according to the reception success information, and the link predicted packet loss rate of the data transmission link is determined according to the data packet predicted packet loss rate, which can more accurately predict the service quality of different anti-packet loss strategies under the predicted network packet loss rate, and improve the anti-packet loss effect of data transmission.

[0108] Figure 3is a structural schematic diagram of a data transmission device based on an anti-packet loss strategy provided by an embodiment of the present application. Referring to Figure 3 The data transmission device based on the anti-packet loss strategy includes a packet loss prediction module 31, a prediction analysis module 32, a quality determination module 33, and a data transmission module 34.

[0109] The packet loss prediction module 31 is configured to determine the reception success information of a first preset number of transmitted data packets in a data transmission link, and determine the link predicted packet loss rate of a second preset number of to-be-transmitted data packets according to the reception success information. The prediction analysis module 32 is configured to determine, according to the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced delay for data transmission based on an automatic repeat request strategy, the first predicted redundancy and the second predicted introduced delay for data transmission based on a forward error correction strategy, and the second predicted retransmission packet number, the second predicted redundancy, and the third predicted introduced delay for data transmission based on a hybrid automatic repeat request strategy. The quality determination module 33 is configured to determine, according to the link predicted packet loss rate, the first predicted retransmission packet number, and the first predicted introduced delay, the first service quality score of the automatic repeat request strategy, determine, according to the link predicted packet loss rate, the first predicted redundancy, and the second predicted introduced delay, the second service quality score of the forward error correction strategy, and determine, according to the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy, and the third predicted introduced delay, the third service quality score of the hybrid automatic repeat request strategy. The data transmission module 34 is configured to determine, as a target anti-packet loss strategy, the anti-packet loss strategy corresponding to the minimum service quality score among the first service quality score, the second service quality score, and the third service quality score, and perform data transmission based on the target anti-packet loss strategy. The anti-packet loss strategy includes the automatic repeat request strategy, the forward error correction strategy, and the hybrid automatic repeat request strategy.

[0110] According to the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced delay, a first service quality score of the automatic repeat request strategy is determined, according to the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced delay, a second service quality score of the forward error correction strategy is determined, and according to the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced delay, a third service quality score of the hybrid automatic repeat request strategy is determined, the anti-packet loss strategy corresponding to the minimum service quality score is determined as the target anti-packet loss strategy, and data transmission is performed based on the target anti-packet loss strategy, the packet loss rate of the subsequent to-be-sent data packet is accurately predicted in advance, the service quality of different anti-packet loss strategies is accurately predicted based on the predicted packet loss rate, and the target anti-packet loss strategy used for data transmission is determined according to the service quality, thereby effectively improving the anti-packet loss effect of data transmission.

[0111] In one possible embodiment, the packet loss prediction module 31 determines the link predicted packet loss rate of the subsequent second preset number of to-be-sent data packets according to the reception success information, and is configured to:

[0112] The reception success information is input into the trained packet loss prediction model, and the packet loss prediction model outputs the data packet predicted packet loss rate of the subsequent second preset number of to-be-sent data packets according to the reception success information;

[0113] The link predicted packet loss rate of the data transmission link is determined according to the data packet predicted packet loss rate.

[0114] In one possible embodiment, the prediction analysis module 32 determines the first predicted retransmission packet number of data transmission based on the automatic repeat request strategy according to the link predicted packet loss rate, and is configured to:

[0115] The first ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate is calculated, the first ratio is rounded up, and the rounded-up result is determined as the first predicted passive retransmission packet number of data transmission based on the passive automatic repeat request strategy; and / or

[0116] The second ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate is calculated, the first product of the difference between the first preset coefficient and the data packet predicted packet loss rate and the second preset coefficient is calculated, the addition result of the second ratio and the first product is calculated, and the addition result is rounded up, and the rounded-up result is determined as the first predicted active retransmission packet number of data transmission based on the active automatic repeat request strategy.

[0117] In one possible embodiment, the prediction analysis module 32 determines, according to the link predicted packet loss rate, a first predicted redundancy and a second predicted introduced latency for data transmission based on the forward error correction strategy, configured to:

[0118] determine an acceptable introduced latency according to a data round-trip time of the data transmission link, and determine a number of redundant packet groups of the forward error correction strategy according to the acceptable introduced latency and a redundant packet duration;

[0119] determine a first predicted redundancy for data transmission based on the forward error correction strategy according to the number of redundant packet groups and the link predicted packet loss rate;

[0120] determine a second predicted introduced latency for data transmission based on the forward error correction strategy according to the number of redundant packet groups and the redundant packet duration.

[0121] In one possible embodiment, the prediction analysis module 32 determines, according to the link predicted packet loss rate, a second predicted number of retransmission packets, a second predicted redundancy and a third predicted introduced latency for data transmission based on the hybrid automatic repeat request strategy, configured to:

[0122] determine a first weight coefficient of the automatic repeat request and a second weight coefficient of the forward error correction in the hybrid automatic repeat request strategy under the condition of data transmission based on the hybrid automatic repeat request strategy;

[0123] determine a second predicted number of retransmission packets and a fourth predicted introduced latency of the automatic repeat request according to the first weight coefficient and the link predicted packet loss rate;

[0124] determine a second predicted redundancy and a fifth predicted introduced latency of the forward error correction according to the second weight coefficient and the link predicted packet loss rate;

[0125] determine the third predicted introduced latency according to the fourth predicted introduced latency and the fifth predicted introduced latency.

[0126] In one possible embodiment, the quality determination module 33 determines, according to the link predicted packet loss rate, the first predicted number of retransmission packets and the first predicted introduced latency, a first quality of service score of the automatic repeat request strategy, configured to:

[0127] determine a first latency cost score according to the first predicted introduced latency, a round-trip time jitter, a network available bandwidth of the data transmission link and a data transmission waiting time, determine a first traffic cost score according to the first predicted number of retransmission packets and an original packet number, and determine a first quality index score according to the link predicted packet loss rate and a bandwidth utilization rate of the data transmission link;

[0128] The first weighted sum result is determined as the first service quality score of the automatic repeat request strategy according to preset delay cost weight, preset traffic cost weight and preset quality index weight.

[0129] In one possible embodiment, the quality determination module 33 is configured to determine the second service quality score of the forward error correction strategy according to the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced delay, and is configured to:

[0130] The second delay cost score is determined according to the second predicted introduced delay, the round-trip delay jitter, the network available bandwidth of the data transmission link and the data sending waiting time length, the second traffic cost score is determined according to the first predicted redundancy and the original packet quantity, and the second quality index score is determined according to the link predicted packet loss rate and the bandwidth utilization of the data transmission link;

[0131] The second weighted sum result is determined as the second service quality score of the forward error correction strategy according to preset delay cost weight, preset traffic cost weight and preset quality index weight.

[0132] In one possible embodiment, the quality determination module 33 is configured to determine the third service quality score of the hybrid automatic repeat request strategy according to the link predicted packet loss rate, the second predicted retransmission packet quantity, the second predicted redundancy and the third predicted introduced delay, and is configured to:

[0133] The third delay cost score is determined according to the third predicted introduced delay, the round-trip delay jitter, the network available bandwidth of the data transmission link and the data sending waiting time length, the third traffic cost score is determined according to the second predicted retransmission packet quantity, the second predicted redundancy and the original packet quantity, and the third quality index score is determined according to the link predicted packet loss rate and the bandwidth utilization of the data transmission link;

[0134] The third weighted sum result is determined as the third service quality score of the hybrid automatic repeat request strategy according to preset delay cost weight, preset traffic cost weight and preset quality index weight.

[0135] It is worth noting that in the above embodiment of the data transmission device based on the anti-packet loss strategy, each unit and module included is only divided according to the functional logic, but is not limited to the above division, as long as the corresponding function can be realized; in addition, the specific name of each functional unit is only for easy mutual differentiation, and does not limit the protection scope of the embodiments of the present application.

[0136] The embodiments of the present application also provide a data transmission device based on an anti-packet loss strategy, which can integrate the data transmission device based on the anti-packet loss strategy provided by the embodiments of the present application. Figure 4 is a structural schematic diagram of a data transmission device based on an anti-packet loss strategy provided by the embodiments of the present application. Referring to Figure 4 , the data transmission device based on the anti-packet loss strategy includes an input device 43, an output device 44, a memory 42, and one or more processors 41; the memory 42 is used to store one or more programs; when the one or more programs are executed by the one or more processors 41, the one or more processors 41 implement the data transmission method based on the anti-packet loss strategy provided by the above embodiments. The data transmission device, device and computer provided above can be used to execute the data transmission method based on the anti-packet loss strategy provided by any of the above embodiments, and have the corresponding functions and advantages.

[0137] The embodiments of the present application also provide a non-volatile storage medium storing computer executable instructions, which are used to execute the data transmission method based on the anti-packet loss strategy provided by the above embodiments when executed by a computer processor. Of course, the non-volatile storage medium storing computer executable instructions provided by the embodiments of the present application is not limited to the data transmission method based on the anti-packet loss strategy provided above, but can also execute the related operations in the data transmission method based on the anti-packet loss strategy provided by any of the embodiments of the present application. The data transmission device, device and storage medium provided in the above embodiments can execute the data transmission method based on the anti-packet loss strategy provided by any of the embodiments of the present application, and the technical details not described in the above embodiments can be referred to the data transmission method based on the anti-packet loss strategy provided by any of the embodiments of the present application.

[0138] On the basis of the above embodiments, the embodiments of the present application also provide a computer program product, the technical solutions of the present application essentially or say the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product, and the computer program product is stored in a storage medium and includes a plurality of instructions to make a computer device, a mobile terminal or a processor therein execute all or part of the steps of the data transmission method based on the anti-packet loss strategy provided by the embodiments of the present application.

Claims

1. A data transmission method based on an anti-packet loss strategy, characterized in that, The method comprises the following steps: determining the receiving success information of a first preset number of sent data packets in a data transmission link, and determining the link predicted packet loss rate of a second preset number of subsequent to-be-sent data packets according to the receiving success information; determining the first predicted retransmission packet number and the first predicted introduced delay for data transmission based on the automatic retransmission request strategy, the first predicted redundancy and the second predicted introduced delay for data transmission based on the forward error correction strategy, and the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced delay for data transmission based on the hybrid automatic retransmission request strategy according to the link predicted packet loss rate; determining the first service quality score of the automatic retransmission request strategy according to the link predicted packet loss rate, the first predicted retransmission packet number and the first predicted introduced delay, determining the second service quality score of the forward error correction strategy according to the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced delay, and determining the third service quality score of the hybrid automatic retransmission request strategy according to the link predicted packet loss rate, the second predicted retransmission packet number, the second predicted redundancy and the third predicted introduced delay; determining the anti-packet loss strategy corresponding to the minimum service quality score among the first service quality score, the second service quality score and the third service quality score as the target anti-packet loss strategy, and performing data transmission based on the target anti-packet loss strategy, wherein the anti-packet loss strategy comprises the automatic retransmission request strategy, the forward error correction strategy and the hybrid automatic retransmission request strategy.

2. The data transmission method based on the anti-packet loss strategy according to claim 1, characterized in that, The method comprises the following steps: inputting the receiving success information into a trained packet loss prediction model, and outputting the data packet predicted packet loss rate of a second preset number of subsequent to-be-sent data packets from the packet loss prediction model according to the receiving success information; determining the link predicted packet loss rate of the data transmission link according to the data packet predicted packet loss rate.

3. The data transmission method based on the anti-packet loss strategy according to claim 1, characterized in that, The method comprises the following steps: calculating the first ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate, taking the first ratio as an integer, and determining the first predicted passive retransmission packet number for data transmission based on the passive automatic retransmission request strategy as the integer; and / or calculating the second ratio of the natural logarithm of the expected packet loss rate and the link predicted packet loss rate, calculating the first product of the difference between the first preset coefficient and the data packet predicted packet loss rate and the second preset coefficient, calculating the addition result of the second ratio and the first product, and taking the addition result as an integer, and determining the first predicted active retransmission packet number for data transmission based on the active automatic retransmission request strategy as the integer.

4. The data transmission method based on the anti-packet loss strategy according to claim 1, characterized in that, The method comprises the following steps: determining an acceptable introduced latency according to a data round-trip time of the data transmission link, and determining a number of redundant packet groups of a forward error correction strategy according to the acceptable introduced latency and a redundant packet duration; determining a first predicted redundancy for data transmission based on the forward error correction strategy according to the number of redundant packet groups and the predicted packet loss rate of the link; determining a second predicted introduced latency for data transmission based on the forward error correction strategy according to the number of redundant packet groups and the redundant packet duration.

5. The data transmission method based on the anti-packet loss strategy according to claim 1, characterized in that, determining a second predicted number of retransmission packets, a second predicted redundancy and a third predicted introduced latency for data transmission based on the hybrid automatic repeat request strategy according to the predicted packet loss rate of the link, including: determining a first weight coefficient of an automatic repeat request and a second weight coefficient of a forward error correction in the hybrid automatic repeat request strategy in a case of data transmission based on the hybrid automatic repeat request strategy; determining a second predicted number of retransmission packets and a fourth predicted introduced latency of the automatic repeat request according to the first weight coefficient and the predicted packet loss rate of the link; determining a second predicted redundancy and a fifth predicted introduced latency of the forward error correction according to the second weight coefficient and the predicted packet loss rate of the link; determining the third predicted introduced latency according to the fourth predicted introduced latency and the fifth predicted introduced latency.

6. The data transmission method based on the anti-packet loss strategy according to claim 1, characterized in that, The determining the first quality of service score of the automatic repeat request strategy according to the predicted packet loss rate of the link, the first predicted number of retransmission packets and the first predicted introduced latency, includes: determining a first delay cost score according to the first predicted introduced latency, a round-trip time jitter, a network available bandwidth of the data transmission link and a data sending waiting time, determining a first traffic cost score according to the first predicted number of retransmission packets and an original packet number, and determining a first quality indicator score according to the predicted packet loss rate of the link and a bandwidth utilization of the data transmission link; performing weighted summation processing on the first delay cost score, the first traffic cost score and the first quality indicator score according to a preset delay cost weight, a preset traffic cost weight and a preset quality indicator weight to obtain a first weighted summation result, and determining the first weighted summation result as the first quality of service score of the automatic repeat request strategy.

7. The data transmission method based on the anti-packet loss strategy according to claim 1, characterized in that, The determining the second quality of service score of the forward error correction strategy according to the predicted packet loss rate of the link, the first predicted redundancy and the second predicted introduced latency, includes: determining a second delay cost score according to the second predicted introduced latency, a round-trip time jitter, a network available bandwidth of the data transmission link and a data sending waiting time, determining a second traffic cost score according to the first predicted redundancy and an original packet number, and determining a second quality indicator score according to the predicted packet loss rate of the link and a bandwidth utilization of the data transmission link; performing weighted summation processing on the second delay cost score, the second traffic cost score and the second quality indicator score according to a preset delay cost weight, a preset traffic cost weight and a preset quality indicator weight to obtain a second weighted summation result, and determining the second weighted summation result as the second quality of service score of the forward error correction strategy.

8. The data transmission method based on the anti-packet loss strategy according to claim 1, characterized in that, The third service quality score of the hybrid automatic repeat request strategy is determined according to the third predicted introduced delay, round-trip delay jitter, network available bandwidth of the data transmission link and data transmission waiting time length, the third predicted retransmission packet quantity, the second predicted redundancy and the original packet quantity, and the link predicted packet loss rate and bandwidth utilization of the data transmission link. The third delay cost score is determined according to the third predicted introduced delay, round-trip delay jitter, network available bandwidth of the data transmission link and data transmission waiting time length, the third traffic cost score is determined according to the second predicted retransmission packet quantity, the second predicted redundancy and the original packet quantity, and the third quality index score is determined according to the link predicted packet loss rate and bandwidth utilization of the data transmission link. The third delay cost score, the third traffic cost score and the third quality index score are weighted and summed according to a preset delay cost weight, a preset traffic cost weight and a preset quality index weight to obtain a third weighted sum result, and the third weighted sum result is determined as the third service quality score of the hybrid automatic repeat request strategy.

9. A data transmission apparatus based on an anti-packet loss strategy, characterized by, It comprises a packet loss prediction module, a prediction analysis module, a quality determination module and a data transmission module, wherein: The packet loss prediction module is configured to determine the reception success information of a first preset quantity of transmitted data packets in the data transmission link, and to determine the link predicted packet loss rate of a second preset quantity of subsequent to-be-transmitted data packets according to the reception success information. The prediction analysis module is configured to determine the first predicted retransmission packet quantity and the first predicted introduced delay for data transmission based on the automatic repeat request strategy, the first predicted redundancy and the second predicted introduced delay for data transmission based on the forward error correction strategy, and the second predicted retransmission packet quantity, the second predicted redundancy and the third predicted introduced delay for data transmission based on the hybrid automatic repeat request strategy according to the link predicted packet loss rate. The quality determination module is configured to determine the first service quality score of the automatic repeat request strategy according to the link predicted packet loss rate, the first predicted retransmission packet quantity and the first predicted introduced delay, to determine the second service quality score of the forward error correction strategy according to the link predicted packet loss rate, the first predicted redundancy and the second predicted introduced delay, and to determine the third service quality score of the hybrid automatic repeat request strategy according to the link predicted packet loss rate, the second predicted retransmission packet quantity, the second predicted redundancy and the third predicted introduced delay. The data transmission module is configured to determine the target anti-packet loss strategy corresponding to the minimum service quality score among the first service quality score, the second service quality score and the third service quality score, and to perform data transmission based on the target anti-packet loss strategy, wherein the anti-packet loss strategy comprises the automatic repeat request strategy, the forward error correction strategy and the hybrid automatic repeat request strategy.

10. A data transmission apparatus based on an anti-packet-loss strategy, characterized by It comprises: a memory and one or more processors; The memory is used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the anti-packet loss strategy-based data transmission method according to any one of claims 1-8.

11. A non-volatile storage medium for storing computer-executable instructions, characterized in that, The computer executable instructions, when executed by a computer processor, perform the data transmission method based on the anti-packet loss strategy as claimed in any one of claims 1-8.

12. A computer program product comprising a computer program, characterized in that, The computer program, when executed by a processor, implements the data transmission method based on the anti-packet loss strategy as claimed in any one of claims 1-8.

Images