Adaptive adjustment streaming media transmission method based on link perception

Abstract

The invention discloses a self-adaptive adjustment streaming media transmission method based on link perception, and belongs to the technical field of satellite communication networks. Comprising the following steps: acquiring historical link state information of a satellite network, and performing preprocessing; extracting link state characteristics of the satellite network according to the preprocessed historical link state information; constructing and training a neural network prediction model to obtain a trained neural network prediction model; collecting current link state information of the satellite network in real time, inputting the current link state information into the trained neural network prediction model, and obtaining a prediction result of a future link state; and dynamically adjusting a coding strategy and a coding parameter in the FEC according to the prediction result. According to the method, the state parameters of the satellite network are collected in real time, the neural network is used for predicting the satellite link, the FEC parameters are optimized in advance, and the accuracy and the real-time performance of link state prediction can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of satellite communication networks, and particularly to an adaptive adjustment method for streaming media transmission based on link awareness. Background Art

[0002] With the development of satellite communication technology, the quality requirements for data transmission are getting higher and higher. In practical applications, due to factors such as changes in atmospheric conditions and solar interference, satellite communication links often encounter problems such as bandwidth fluctuations, increased latency, and increased packet loss rate, which directly affect the quality and stability of communication.

[0003] Traditional network optimization methods mainly rely on static or simple dynamic parameter adjustment, unable to capture and process complex link state changes in real time, resulting in non-real-time link state results, unable to timely reflect the current actual link conditions, lacking an adaptive adjustment mechanism, unable to automatically adjust system configuration according to changes in link state. Traditional satellite link state prediction systems usually use fixed FEC (Forward Error Correction) parameters and are unable to dynamically optimize redundancy according to the actual link state, resulting in affected transmission efficiency and reliability in the case of high packet loss rate or low bandwidth. Summary of the Invention

[0004] Aiming at the defects of the prior art, the present invention collects the state parameters of the satellite network in real time and uses a neural network to predict the satellite link for optimizing FEC parameters, which can significantly improve the accuracy and real-time performance of link state prediction.

[0005] To achieve the above object, the present invention provides an adaptive adjustment method for streaming media transmission based on link awareness, including the following steps:

[0006] (1) Obtain the historical link state information of the satellite network and perform preprocessing;

[0007] (2) Extract the link state features of the satellite network according to the preprocessed historical link state information;

[0008] (3) Construct a neural network prediction model, and use the link state features in step (2) to train the neural network prediction model to obtain a trained neural network prediction model;

[0009] (4) Real-time collect the current link state information of the satellite network, input it into the trained neural network prediction model, and obtain the prediction result of the future link state;

[0010] (5) Dynamically adjust the encoding strategy and encoding parameters in FEC according to the prediction result.

[0011] Further, step (1) is specifically as follows:

[0012] (1.1) Obtain the historical link state information of the satellite network at preset time intervals;

[0013] (1.2) Perform preprocessing on the historical link state information, including data cleaning and data conversion.

[0014] Further, in step (2), link state features are extracted from the preprocessed historical link state information according to the Pearson correlation coefficient;

[0015]

[0016] where: rx represents the Pearson correlation coefficient between the historical link state variables x and y of the preprocessed satellite network, x i represents the x value of the i-th observation point, y i represents the y value of the i-th observation point, respectively represent the means of x and y, and n represents the number of samples.

[0017] Further, the neural network prediction model uses an LSTM neural network.

[0018] Further, the historical link state information of the satellite network includes delay, bandwidth, and packet loss rate, and the future delay, bandwidth, and packet loss rate of the satellite network are obtained through the neural prediction network.

[0019] Further, step (5) is specifically as follows: Adjust the coding strategy and coding parameters in FEC according to the predicted future packet loss rate of the satellite network:

[0020] k / n = 1 - (lossrate + 2%)

[0021] where: n is the codeword length in Reed - Solomon coding; k is the information bit length in Reed - Solomon coding; lossrate is the predicted future packet loss rate of the satellite network.

[0022] The present invention also provides an adaptive adjustment streaming media transmission system based on link perception, including:

[0023] A data acquisition module, configured to obtain the historical link state information of the satellite network and perform preprocessing;

[0024] A feature extraction module, configured to extract the link state features of the satellite network according to the preprocessed historical link state information;

[0025] A modeling and training module, which is used to build a neural network prediction model and train the neural network prediction model with the link state features obtained by the feature extraction module to obtain a trained neural network prediction model;

[0026] A prediction module, which is used to collect the current link state information of the satellite network in real time, input the trained neural network prediction model, and obtain the prediction result of the future link state;

[0027] An adjustment module, which is used to dynamically adjust the encoding strategy and encoding parameters in the FEC according to the prediction result.

[0028] Advantages of the present invention:

[0029] 1. The present invention collects the state parameters of the satellite network in real time, uses a neural network to predict the satellite link, and optimizes the FEC parameters in advance, which can significantly improve the accuracy and real-time performance of link state prediction.

[0030] 2. The present invention automatically adjusts the system configuration according to the change of the link state, including the data collection frequency, the prediction parameters of the neural network model, and the adjustment strategy of the FEC encoding. This adaptive adjustment ability enables the system to flexibly respond to different network environments and conditions and optimize the transmission effect.

[0031] 3. According to the prediction result of the neural network prediction model, the present invention dynamically adjusts the Reed-Solomon encoding parameters in advance, can dynamically optimize the redundancy according to the actual link state, thereby improving the transmission efficiency and reliability. Especially when a high packet loss rate or low bandwidth is predicted, it can automatically increase or decrease the redundancy to meet different transmission requirements. By predicting the link condition in advance and pre-adjusting the parameters, the reduction of transmission quality caused by delay is avoided, and the response speed of the system is improved.

[0032] 4. By monitoring and adjusting the transmission effect in real time, the present invention can perform load balancing among multiple links, avoid overloading of a single link, and improve the overall transmission efficiency. By combining the neural network and FEC technologies, a complete adaptive network transmission optimization system is formed, which has good scalability and self-adjustment ability. Description of the Drawings

[0033] Figure 1 It is a schematic flowchart of the method for adaptively adjusting streaming media transmission based on link perception according to an embodiment of the present invention.

[0034] Figure 2 It is a schematic structural diagram of the neural network prediction model (LSTM) according to an embodiment of the present invention.

[0035] Figure 3 It is a schematic structural diagram of the system for adaptively adjusting streaming media transmission based on link perception according to an embodiment of the present invention. Specific Embodiments

[0036] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0037] An embodiment of the present invention provides an adaptive adjustment streaming media transmission method based on link awareness, including the following steps:

[0038] S101. Obtain the historical link state information of the satellite network and perform preprocessing.

[0039] (1) Obtain the historical link state information of the satellite network at preset time intervals.

[0040] The historical link state information includes delay, bandwidth, and packet loss rate.

[0041] (2) Perform preprocessing on the historical link state information, including data cleaning and data conversion.

[0042] Data cleaning: Process missing data, remove outliers, and correct incorrect data, including filling missing values, deleting duplicate records, etc.

[0043] Data conversion: Convert the original data into a form suitable for model training, such as performing standardization or normalization processing, so that data of different magnitudes can be compared on the same scale.

[0044] The preprocessed data is divided into a training set and a test set.

[0045] S102. Extract the link state features of the satellite network according to the preprocessed historical link state information.

[0046] Extract link state features from the preprocessed historical link state information according to the Pearson correlation coefficient.

[0047]

[0048] Where: rx represents the Pearson correlation coefficient between the historical link state variables x and y of the preprocessed satellite network, x i represents the x value of the i-th observation point, y i represents the y value of the i-th observation point, represent the means of x and y respectively, and n represents the number of samples.

[0049] S103. Construct a neural network prediction model, and use the link state features in step S102 to train the neural network prediction model to obtain a trained neural network prediction model.

[0050] Such as Figure 2As shown in the figure, the embodiment of the present invention uses an LSTM neural network as a neural network prediction model, trains the LSTM neural network using the link state features obtained in step S102, and iteratively trains the LSTM neural network according to the loss function until the number of iterations meets the preset number to obtain a trained LSTM neural network.

[0051] S104. Collect the current link state information of the satellite network in real time, input it into the trained neural network prediction model, and obtain the prediction result of the future link state.

[0052] Collect the current link state information (delay, bandwidth, and packet loss rate) of the satellite network in real time, and input it into the trained LSTM neural network for predicting the future link state (delay, bandwidth, and packet loss rate).

[0053] S105. Dynamically adjust the coding strategy and coding parameters in the FEC according to the prediction result.

[0054] The core parameters of FEC coding (such as forward error correction coding) are n and k in Reed - Solomon coding. Among them: n is the codeword length (the total length of the encoded data); k is the information bit length (the length of the original data). The redundancy is determined by n - k, indicating the size of the error - correction ability. k / n represents the code rate, which is the proportion of the source data. The higher the code rate (the closer to 1), the less redundant information, the higher the coding efficiency, but the weaker the error - correction ability. The lower the code rate (the closer to 0), the more redundant information, the stronger the error - correction ability, but the lower the coding efficiency.

[0055] n = 2m - 1

[0056] k = n – 2t

[0057] Among them: m is the number of bits per symbol, usually an integer (8 or 16), depending on the limitations of hardware and software implementation capabilities. m = 8; at this time, the symbol field size is 256 (2^8), which is often used in storage such as optical discs (CD, DVD), and this symbol size is also used in many data communication systems. m = 16, at this time, the symbol field size is 65536 (2^16), which is suitable for high - capacity storage or communication systems that require higher error - correction capabilities.). t is the number of errors that can be corrected (i.e., the number of symbol errors that each codeword can correct). For a codeword, at most t symbol errors can be corrected, which is set according to the actual application scenario. The relationship between k and n can be expressed as:

[0058] k / n = 1 - (lossrate + 2%)

[0059] Among them: lossrate is the predicted packet loss rate at that time.

[0060] Therefore, according to the predicted packet loss rate, the encoding strategy and encoding parameters in FEC can be dynamically adjusted, and the selected Reed-Solomon encoding parameters are updated to the FEC encoder.

[0061] As Figure 3 shown, an embodiment of the present invention further provides an adaptive adjustment streaming media transmission system based on link awareness, including:

[0062] A data acquisition module 310, configured to obtain historical link state information of a satellite network and perform preprocessing.

[0063] A feature extraction module 320, configured to extract link state features of the satellite network according to the preprocessed historical link state information.

[0064] A modeling and training module 330, configured to build a neural network prediction model, and use the link state features obtained by the feature extraction module to train the neural network prediction model to obtain a trained neural network prediction model.

[0065] A prediction module 340, configured to collect current link state information of the satellite network in real time, input it into the trained neural network prediction model, and obtain a prediction result of future link states.

[0066] An adjustment module 350, configured to dynamically adjust the encoding strategy and encoding parameters in FEC according to the prediction result.

[0067] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent replacements, and improvements made within the principle and spirit of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for adaptively adjusting streaming media transmission based on link perception, characterized in that: The steps include: (1) Obtain historical link status information of the satellite network and perform preprocessing; (2) extracting link status features of the satellite network based on the preprocessed historical link status information; (3) constructing a neural network prediction model, and using the link state features in step (2) to train the neural network prediction model to obtain a trained neural network prediction model; (4) Collect the current link status information of the satellite network in real time, input it into the trained neural network prediction model, and obtain the prediction results of the future link status; (5) Dynamically adjust the coding strategy and coding parameters in FEC according to the prediction results.

2. The link-aware adaptive streaming media transmission method according to claim 1, characterized in that: The step (1) is specifically: (1.1) Obtaining historical link status information of the satellite network at preset time intervals; (1.2) Preprocessing the historical link status information, including data cleaning and data conversion.

3. The link-aware adaptive streaming media transmission method according to claim 1 is characterized in that: The step (2) extracts link state features from the pre-processed historical link state information according to the Pearson correlation coefficient; Where: rx represents the Pearson correlation coefficient between the link state variables x and y of the satellite network after preprocessing, x i represents the x value of the i-th observation point, y i represents the y value of the i-th observation point, and They represent the means of x and y respectively, and n represents the number of samples.

4. The method for adaptively adjusting streaming media transmission based on link perception according to claim 1, characterized in that: The neural network prediction model adopts LSTM neural network.

5. The link-aware adaptive streaming media transmission method according to claim 1, characterized in that: The historical link status information of the satellite network includes delay, bandwidth and packet loss rate, and the future delay, bandwidth and packet loss rate of the satellite network are obtained through the neural prediction network.

6. The link-aware adaptive streaming media transmission method according to claim 5, characterized in that: The step (5) specifically includes: adjusting the coding strategy and coding parameters in FEC according to the predicted future packet loss rate of the satellite network: k / n=1-(loss rate+2%) Where: n is the codeword length in Reed-Solomon coding; k is the information bit length in Reed-Solomon coding; lossrate is the predicted future packet loss rate of the satellite network.

7. A link-aware adaptive streaming media transmission system, characterized in that: include: Data acquisition module, used to obtain historical link status information of satellite network and perform preprocessing; A feature extraction module, used to extract link status features of the satellite network based on the pre-processed historical link status information; A modeling and training module, used to construct a neural network prediction model, and train the neural network prediction model using the link state features obtained by the feature extraction module to obtain a trained neural network prediction model; The prediction module is used to collect the current link status information of the satellite network in real time, input the trained neural network prediction model, and obtain the prediction results of the future link status; The adjustment module is used to dynamically adjust the coding strategy and coding parameters in FEC according to the prediction result.

Images