A method for optimizing video bit rate allocation

Abstract

The present invention provides a method for optimizing video bit rate allocation, and belongs to the field of wireless communication technology. The method of the present invention comprises: taking the distance s between the user and the base station as a hidden variable, the actual network speed Y as an observed variable, collecting user samples to establish a hidden Markov model HMM; obtaining the current network speed of the target user, and using HMM to predict the distance s; expressing the network speed distribution as a Poisson distribution with θ as a hyperparameter, θ being a function of the distance s, calculating the hyperparameter θ according to the predicted distance s, and determining the network speed distribution of the target user; according to the obtained network speed distribution, selecting the video bit rate allocation scheme with the best experience quality from the set of video bit rate allocation schemes corresponding to the network state. The method of the present invention realizes intelligent adjustment of bit rate allocation according to the actual situation of the user network, thereby optimizing the video bit rate allocation and improving the user experience quality.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless communications, and in particular to a method for optimizing video bit rate allocation. Background Art

[0002] The widespread popularity of the Internet and the widespread dissemination of short videos have led to a significant increase in video transmission traffic. Currently, video traffic accounts for the majority of network traffic. Therefore, optimizing video bitrate allocation is of great significance for saving network resource consumption and improving user experience.

[0003] Video bitrate control usually consists of a target bitrate allocation algorithm, a rate-distortion RD model, a quantization parameter QP calculation algorithm, an adaptive quantization algorithm based on human visual characteristics, and a buffer overflow control algorithm. The target bitrate allocation algorithm can be subdivided into GOP (Group of pictures) level bitrate allocation, frame level bitrate allocation, and basic unit level bitrate allocation according to the granularity of bitrate allocation. The basic principle of allocation is to allocate more bits to frames or intra-frame areas with high texture complexity to ensure uniform distribution of coding distortion.

[0004] The video bitrate control scheme is mainly selected by weighing the five factors of visual quality stability, instant output bitrate, controllable output video file size, encoding speed, and power consumption. In different network environments and different devices, the user experience can achieve good video viewing quality. The bitrate control models include CQP, CRF, CBR, VBR, ABR, etc. CQP is a fixed QP, that is, each frame of the image is encoded according to a specific QP. This encoding does not consider the complexity of the encoded content, so the video quality is not high and is basically not used. CRF is a constant bitrate factor, which achieves this goal by reducing the quality of frames that consume bitrate but are difficult to detect with the naked eye (high-speed motion or rich texture) and increasing the bitrate of static frames. This mode prioritizes video quality, so the bitrate fluctuates greatly. It is suitable for occasions with requirements for video quality. CBR is a constant bitrate, that is, the bitrate remains basically constant over a period of time. The CBR bitrate priority model causes unstable video quality and is generally not used. VBR is a variable bitrate. Simple scenes are assigned a relatively large QP, with a low compression rate and high quality; complex scenes are assigned a smaller QP. The VBR model is used to obtain basically stable visual quality, because the visual characteristics of the human eye are insensitive to complex scenes. The disadvantage is that the output bit rate is uncontrollable. VBR has two control modes: quality priority mode and 2PASS secondary encoding mode. The quality priority mode does not consider the size of the output video file, and allocates the bit rate completely according to the complexity of the video content, so that the video playback quality is the best. The secondary encoding method 2PASS detects the simple and complex parts of the video content in the first encoding, and determines the ratio of simple and complex. The second encoding will keep the average bit rate of the video unchanged, and allocate more bits to complex places and fewer bits to simple places. This encoding effect is very good, but the speed is relatively slow. ABR is a constant average target bit rate. Simple scenes are allocated with lower bits, and complex scenes are allocated with enough bits, so that the limited number of bits can be reasonably allocated in different scenes, which is similar to VBR. At the same time, within a certain period of time, the average bit rate is close to the set target bit rate, so that the size of the output file can be controlled, which is similar to CBR. ABR can be considered as a compromise between CBR and VBR, and is more commonly used in live broadcast and low-latency systems. Summary of the invention

[0005] The purpose of the present invention is to propose a method for optimizing video bit rate allocation, by predicting the distance between the user and the base station, inferring the hyperparameters of the network speed distribution, performing QoE optimization on the distribution to determine the bit rate allocation; compared with the above-mentioned prior art, the bit rate allocation can be intelligently adjusted to make the bit rate allocation process more optimal.

[0006] Specifically, a method for optimizing video bit rate allocation of the present invention comprises the following steps:

[0007] Step 1: Detect a certain number of user samples, including the distance between the user and the base station and the real-time network speed. Take the distance s between the user and the base station as the hidden variable and the actual network speed Y as the observed variable to establish a hidden Markov model.

[0008] Step 2: Detect the real-time network speed of the target user, and predict the distance s between the target user and the base station based on the hidden Markov model established by the sample;

[0009] Step 3: Express the network speed distribution as a Poisson distribution with θ as a hyperparameter, and express the hyperparameter θ as a function of the distance s. Calculate the hyperparameter θ according to the predicted distance s between the target user and the base station to obtain the network speed distribution of the target user expressed by the Poisson distribution.

[0010] Step 4: According to the obtained network speed distribution, select a video bitrate allocation scheme that optimizes the quality of experience (QoE) from a set of video bitrate allocation schemes corresponding to the network status; wherein the corresponding set of video bitrate allocation schemes is pre-set according to different network statuses.

[0011] The advantages and positive effects of the present invention are: the method of the present invention obtains the network speed distribution by predicting the distance between the user and the base station, selects the QoE-optimized solution from the set of video coding allocation solutions suitable for the network status, thereby optimizing the video bit rate allocation and improving the user experience quality. Compared with the prior art, the present invention can adaptively select the optimal video bit rate allocation solution according to the actual situation of the user network, realize intelligent adjustment of bit rate allocation, and improve the user experience quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A flow chart of an implementation of the method for optimizing video bit rate allocation of the present invention;

[0013] Figure 2 Schematic diagram of network speed distribution at different locations of a base station in an embodiment of the present invention. DETAILED DESCRIPTION

[0014] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0015] The method for optimizing video bit rate allocation implemented by the embodiment of the present invention pre-installs a distance prediction module, a network speed distribution determination module, and a video bit rate allocation scheme selection module on the user equipment, such as Figure 1 As shown, the following is explained in three steps.

[0016] Step S1: Collect a certain number of user samples in advance to obtain trajectory samples of user network speed changing with distance. The samples include the distance between the user and the base station and the real-time network speed. Then, take the distance s between the user and the base station as the hidden variable and the actual network speed Y as the observed variable to establish an HMM model and store it in the distance prediction module.

[0017] Hidden Markov Model (HMM) is a statistical model used to describe a Markov process with hidden unknown parameters. The state is not directly visible, but some variables affected by the state are visible. Each state has a certain probability distribution on the possible output results, so the output results can reveal some information about the state sequence. The present invention uses the relationship between the network speed Y and the distance s of each sample to establish a hidden Markov model, where the distance s is the hidden variable and the network speed Y is the observed variable.

[0018] The embodiment of the present invention collects a certain number of user samples in advance, obtains trajectory samples of network speed changing with distance, and calculates the initial state probability, state transition probability and output observation probability involved in the hidden Markov model according to the measurement samples.

[0019] Step S2: Detect the real-time network speed of the target user and input it into the distance prediction module. The distance prediction module predicts the distance s between the user and the base station according to the established hidden Markov model.

[0020] The distance prediction module provided in the embodiment of the present invention loads the calculated hidden Markov model, and then in actual testing, the distance s between the user and the base station is predicted by the stored hidden Markov model according to the currently measured network speed.

[0021] Step S3, assuming that the network speed distribution is known, it is a Poisson distribution p(Y|θ) with θ as a hyperparameter. The hyperparameter θ is a function of the distance s and can be expressed as θ=f(s). The hyperparameter θ of the network speed distribution p(Y|θ) is inferred based on the distance s obtained in step S1, thereby determining the network speed distribution of the target user.

[0022] Poisson distribution is suitable for describing the probability distribution of the number of random events that occur per unit time. For example, the number of service requests received by a certain service facility within a certain period of time, the number of calls received by a telephone exchange, the number of passengers waiting at a bus station, the number of machine failures, the number of natural disasters, the number of mutations in DNA sequences, the number of decays of radioactive nuclei, the distribution of the number of photons of a laser, etc. In an ideal network environment, the probability of n data packets appearing in a time period obeys Poisson distribution. In the method of the present invention, the network speed distribution is represented as a Poisson distribution with θ as a hyperparameter, i.e., Y~P(θ). When in use, the current network speed can be measured, and the distance s can be predicted according to step S1, and the hyperparameter θ can be further inferred.

[0023] like Figure 2As shown in the figure, the network speed distribution under the ideal condition of different distances s from the base station is shown in the curve in the figure. The network speed distribution of each location conforms to the Poisson distribution about the hyperparameter θ. The farther away from the base station, the smaller the mean network speed, the larger the variance, the worse the network condition, and a more conservative encoding method is needed.

[0024] The embodiment of the present invention obtains the network speed distribution at different distances in advance, as well as the trajectory samples of the network speed changing with the distance, and then fits and calculates the function f(s) of the hyperparameter θ with respect to the distance s, and stores it in advance in the network speed distribution determination module. In actual measurement, the network speed distribution determination module calculates the hyperparameter θ according to the distance output by the distance prediction module, and determines the network speed distribution represented by the Poisson distribution. The network speed distribution function pre-stored in the network speed distribution determination module in the embodiment of the present invention is the Poisson distribution, and network speed distribution functions represented by other forms or models may also be stored according to actual conditions.

[0025] Step S4: Perform QoE (Quality of Experience) optimization on the network speed distribution obtained in step S3, thereby optimizing video bit rate allocation.

[0026] After obtaining the network speed distribution, the network condition of the current user location can be obtained. In the embodiment of the present invention, different video bit rate allocation scheme sets are set in advance for network conditions with different network speeds, such as setting corresponding video bit rate allocation scheme sets A, B, and C for three situations of good, average, and poor network conditions, respectively. For example, the schemes set in set A include CRF, VBR, ABR, etc. According to the obtained network speed distribution, the video bit rate allocation scheme with the maximum QoE is taken at the user location as the optimal scheme.

[0027] As can be seen from the above, the method of the present invention makes the video bit rate allocation more suitable for the actual situation of the current user network, and the user experience quality is improved along with the measured network speed.

[0028] Except for the technical features described in the specification, all of them are known technologies to those skilled in the art. The present invention omits the description of known components and known technologies to avoid redundancy and unnecessary limitation of the present invention. The implementation methods described in the above embodiments do not represent all implementation methods consistent with the present application. On the basis of the technical solution of the present invention, various modifications or variations that can be made by those skilled in the art without creative labor are still within the protection scope of the present invention.

Claims

1. A method for optimizing video bit rate allocation, characterized in that: The steps include: Step 1: Collect trajectory samples of network speed changing with distance in advance. The samples include the distance between the user and the base station and the real-time network speed. Then, take the distance s between the user and the base station as the hidden variable and the actual network speed Y as the observed variable to establish a hidden Markov model. Step 2: Detect the real-time network speed of the target user and predict the distance s between the target user and the base station based on the established hidden Markov model; Step 3: Express the network speed distribution as a Poisson distribution with θ as a hyperparameter, and express the hyperparameter θ as a function of the distance s. Calculate the hyperparameter θ according to the predicted distance s between the target user and the base station to obtain the network speed distribution of the target user expressed by the Poisson distribution. Step 4: According to the obtained network speed distribution, select a video bitrate allocation scheme that optimizes the quality of experience (QoE) from a set of video bitrate allocation schemes corresponding to the network status; wherein the corresponding set of video bitrate allocation schemes is pre-set according to different network statuses.

2. The method according to claim 1, characterized in that In the step three, a hyperparameter θ=f(s) is set, the network speed distribution at different distances and the trajectory samples of the network speed changing with the distance are obtained in advance, and the function f is fitted and calculated.

Images