A Wireless VR Video Transmission Method Assisted by Mobile Edge Computing

Abstract

The present invention discloses a wireless VR video transmission method assisted by mobile edge computing, including: 1. Establishing a VR video wireless transmission system assisted by an edge computing server; 2. Collecting system information; 3. Loading a VR video coding parameter optimization algorithm; 4. Loading a decision algorithm for rendering offloading to determine the nodes for field of view rendering; 5. Loading a MIMO precoding matrix design algorithm; 6. Loading a computing resource allocation algorithm to determine the computing resource allocation between VR users and the edge computing server. 7. Iteratively loading steps 3 to 6 until the weighted sum of the total energy consumption and distortion of the VR video wireless transmission system converges. The present invention can alleviate the rendering burden of VR users, and the multiple-input multiple-output MIMO transmission technology can improve the transmission rate, reduce latency, and enhance the VR user experience.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communication, and particularly relates to a wireless VR video transmission method assisted by mobile edge computing. Background Art

[0002] Due to the ability to provide immersive experiences, wireless virtual reality (VR) is becoming an important use case for 6G networks and has great prospects in industries such as education, healthcare, and gaming. The realization of VR experiences depends on constructing 3D-effect videos at the user display end. During the VR experience, due to the mobility of the head-mounted display (HMD), it is necessary to track the VR user's viewport and frequently update the field of view (FOV), which poses huge challenges to wireless transmission and computing. First, VR videos have the characteristic of ultra-high resolution, and even after compression, the data volume is huge, imposing a heavy burden on the wireless link. Second, when the VR video frame rate is lower than 60 frames per second, stuttering occurs and the user experience deteriorates, which places stringent requirements on computing and transmission latency. In addition, FOV rendering requires per-pixel position conversion, and this computationally intensive operation requires powerful computing capabilities and consumes a lot of energy. However, due to limitations in computing power and battery capacity, VR users are not capable of ultra-high-definition and long-duration VR experiences.

[0003] To address the above challenges, by deploying a mobile edge computing (MEC) server near VR users, caching part of the FOV in advance, and offloading part of the rendering tasks to the edge computing server, the FOV transmission latency can be greatly reduced and the computing pressure on VR users can be alleviated. In addition, the multiple-input multiple-output (MIMO) transmission technology is adopted to improve the VR video transmission rate, thereby meeting the ultra-low latency requirements.

[0004] Although mobile edge computing and MIMO transmission technologies can alleviate the computing pressure on VR users and reduce the VR video transmission latency, the rendered VR videos will bring greater latency and higher energy consumption to wireless transmission, and the transmit precoding design of MIMO transmission technology is complex. In addition, the quality of experience of VR users is affected not only by latency but also by the VR video coding quality. Higher VR video quality corresponds to a larger data volume, resulting in greater transmission latency and energy consumption. Therefore, it is necessary to comprehensively consider the computing resources of VR users and edge computing servers, the wireless transmission environment, and user QoE requirements, carefully design the rendering offloading and resource allocation schemes, and balance the smoothness, clarity, and experience duration of the user's VR experience. Summary of the Invention

[0005] The object of the present invention is to provide a wireless VR video transmission method assisted by mobile edge computing. By selecting appropriate VR video coding parameters, offloading some rendering calculation tasks to the edge computing server, and improving the data throughput of VR videos through multiple-input multiple-output (MIMO), the latency of VR video transmission and the energy consumption of VR users are reduced, thereby providing a smoother, clearer, and longer VR experience for VR users.

[0006] To solve the above technical problems, the specific technical solution of the present invention is as follows:

[0007] A wireless VR video transmission method assisted by mobile edge computing includes the following steps:

[0008] Step S1: Establish a VR video transmission system assisted by an edge computing server; including a VR video provider, a wireless access point, an edge computing server, and a VR user. The VR user is configured with N r antennas, and the wireless access point is configured with N t antennas; the VR user and the edge computing server are configured with computing units to process rendering tasks; the VR user communicates wirelessly with the wireless access point, the wireless access point is connected to the VR video provider by wire, the edge computing server is embedded in the wireless access point, and the VR user offloads the rendering calculation tasks to the edge computing server or completes the rendering tasks locally; the wireless access point serves as a control center to collect system information of the VR video transmission system, make decisions, and distribute control instructions;

[0009] Step S2: The VR user sends a VR video demand instruction to the wireless access point. The edge computing server first extracts the 2D field of view (FOV) from the 2D video according to the VR user's gaze point and VR user posture information, and then determines whether to render the 2D FOV as a 3D FOV at the edge computing server or the VR user according to the system information collected by the control center, and finally plays it on the VR user's head-mounted display;

[0010] Step S3: Design a VR video coding parameter optimization algorithm according to the system information collected by the control center in Step S2 to determine the VR video coding parameters;

[0011] Step S4: Design a rendering offloading algorithm according to the system information collected by the control center in Step S2 and the VR video coding parameters obtained in Step S3 to determine the rendering offloading strategy;

[0012] Step S5: Design a MIMO transmission precoding matrix optimization algorithm according to the system information collected by the control center in Step S2 and the VR video coding parameters and rendering offloading strategy determined in Steps S3 and S4 to determine the precoding matrix;

[0013] Step S6: Design a computing resource allocation algorithm for the edge computing server and the VR user based on the VR video encoding parameters, rendering offloading strategy, and precoding matrix determined in Steps S3 - S5, and determine the computing resource allocation.

[0014] Step S7: Iteratively load Steps S3 to S6 until the weighted sum of the total energy consumption and distortion of the VR video transmission system converges, and finally determine the VR video encoding parameters, rendering offloading decision, precoding matrix, and computing resource allocation.

[0015] Further, in Step S2, the system information collected by the control center includes the duration T of a playback window p , the number N of foveated video slices v , the number N of non - foveated video slices s , the number N of pixel points in a single video slice p , the channel state information between the VR user and the wireless access point, the distortion threshold D th , the total computing resource f of the edge computing server tot，A , the total computing resource f of the VR user tot，L .

[0016] Further, the steps of the VR video encoding parameter optimization algorithm in Step S3 include:

[0017] Step S301: Input the system information in Step S2, and initialize the rendering offloading decision variables, MIMO precoding matrix, and computing resource allocations for the edge computing server and the VR user that satisfy the system operation constraints.

[0018] Step S302: Solve the foveated video slice bitrate optimization problem using the interior - point method to obtain the optimal bitrate of the foveated video slice. The specific optimization problem is as follows:

[0019]

[0020] s.t.γ v ∈[γ l , γ h , (2)

[0021] D≤D th , (3)

[0022]

[0023]

[0024] where γ v =γ(q v ) represents the bitrate of the foveated video slice, and the specific expression is

[0025]

[0026] where γ(·) represents a function, q v represents the encoding parameter of the fixation video slice, δ1 and σ1 are positive parameters related to the video content characteristics, and v1 and v2 refer to the parameters related to the encoding structure; γ h =δ1 exp(σ1(v1 - Q l )) / v2) and γ l =δ1 exp(σ1(v1 - Q h )) / v2), Q l represents the minimum encoding parameter of the fixation video slice, Q h represents the maximum encoding parameter of the fixation video slice; w1, w2, and w3 represent intermediate variables, w1 = wβ A x r , w2 = wβ A (1 - x r ), w3 = wβ L x r , where w represents the weight coefficient for balancing energy consumption and distortion, β A represents the energy consumption weight coefficient of the wireless access point, β L represents the energy consumption weight coefficient of the local VR user; x r ∈{0, 1} represents the rendering decision variable, x r = 1 indicates that the rendering task is executed locally, x r = 0 indicates that it is executed on the edge computing server; D represents the distortion of the FOV, and its expression is as follows:

[0027]

[0028] where ξ∈(0.5, 1) represents the weight of the distortion, represents the distortion size of the fixation video slice, δ2 and σ2 are positive parameters related to the video content characteristics, represents the distortion size of the non - fixation video slice, q s refers to the encoding parameter of the non - fixation video slice; represents the 2D FOV data size extracted per unit window, represents the 2D FOV data size stored at the wireless access point, represents the 3D FOV data size rendered by the edge computing server per unit window, represents the 3D FOV data size stored at the wireless access point, represents the maximum storage space of the edge computing server; refers to the energy consumed by the wireless access point to transmit the 2D FOV, refers to the energy consumed by the edge computing server to render the 2D FOV, Indicates the energy consumed for transmitting the 3D FOV, refers to the energy consumed by the local VR user for rendering the 2D FOV, and the expressions of are as follows:

[0029]

[0030]

[0031]

[0032]

[0033] where the ratio κ represents the degree of increase in transmission overhead, P refers to the MIMO precoding matrix, μ L represents the computing power related to the hardware structure of the local VR user, μ A represents the computing power related to the hardware of the edge computing server, η L refers to the number of computing cycles required for the local VR user to compute and process 1 bit of data, η A refers to the number of computing cycles required for the edge computing server to process 1 bit of data, f L represents the computing resources allocated to the local VR user, f A represents the computing resources allocated to the edge computing server; B F represents the data size of the 2D FOV, and its expression is as follows:

[0034] B F = N p (γ(q v )N v +γ(q s )N s ), (12)

[0035] where B o = αB F refers to the data volume size obtained from the 3D FOV. The constant parameter α ≥ 2 indicates that it contains at least the playback information of two eyes; R represents the wireless transmission rate, and its expression is as follows:

[0036]

[0037] where H refers to the channel matrix between the wireless access point and the VR user, W d refers to the channel bandwidth, σ 2 refers to the variance of the noise on a single antenna port, represents the N r th order identity matrix, P H represents the conjugate transpose of the precoding matrix P, HH Denotes the conjugate transpose of the channel matrix H;

[0038] Step S303: Iterate step S302 until the objective function of the formula converges to obtain the optimal bit rate γ v .

[0039] Step S304: Calculate the optimal encoding parameters of the fixation video segment according to the following to obtain the optimal

[0040]

[0041] where The function is the ceiling function.

[0042] Furthermore, the optimization algorithm steps for rendering offloading in step S4 include:

[0043] Step S401: Input the system information in step S2, and initialize the encoding parameters of the fixation video segment, the MIMO transmission precoding matrix, the computing resource sizes allocated to the edge computing server and the local VR user that satisfy the system operation constraint conditions;

[0044] Step S402: Solve the rendering decision optimization problem using the interior point method. The specific optimization problem is as follows:

[0045]

[0046] s.t. x r ∈[0, 1], (16)

[0047] x r (x r -1) = 0, (17)

[0048]

[0049] where, Denotes the size of the rendered FOV data stored in the local VR user, Denotes the data volume of the 3D FOV rendered per unit window by the local VR user; v L Denotes the playback rate of the VR user, b th Denotes the cache threshold of the rendered data. When the rendered data is lower than this threshold, the video playback will pause;

[0050] Step S403: Iterate step 402 until the objective function of the formula converges to obtain the relaxed optimal decision variable

[0051] Step S404: Calculate the optimal rendering decision variable according to the following formula

[0052]

[0053] where ρ represents the decision threshold constant.

[0054] Further, the steps of the MIMO precoding matrix optimization algorithm in step S5 include:

[0055] Step S501: Input the system information in step S2, and initialize the encoding parameters of the fixation video slice, the rendering offloading decision variable, the computing resource sizes allocated to the edge computing server and the local VR user that satisfy the system operation constraint conditions;

[0056] Step S502: Calculate the first-order Taylor expansion of R The expression is as follows:

[0057]

[0058] where P (n) refers to the precoding matrix obtained by the nth iterative calculation, represents the first derivative of R.

[0059] Step S503: Update the variable z, and the update expression is as follows:

[0060]

[0061] Step S504: Solve the MIMO precoding matrix optimization problem using the interior point method. The specific optimization problem is as follows:

[0062]

[0063]

[0064]

[0065]

[0066]

[0067] where represents the maximum storage capacity of the local VR user, and P max represents the maximum transmission power of the wireless access point.

[0068] Step S505: Iterate steps S502 to S504 until the objective function of the formula converges to obtain the optimal MIMO transmit precoding matrix P * .

[0069] Further, the computing resource allocation method for the edge computing server and the local VR user in step S6 is expressed as:

[0070] The optimal solution for the computing resource allocation of the edge computing server is expressed as:

[0071]

[0072] The optimal solution for the computing resource allocation of local VR users is expressed as:

[0073]

[0074] Among them, represents the minimum value of the computing resource allocation of the edge computing server, represents the maximum value of the computing resource allocation of the edge computing server, represents the minimum value of the local computing resource allocation of VR users, represents the maximum value of the local computing resource allocation of VR users. The specific expressions are as follows:

[0075]

[0076]

[0077]

[0078]

[0079] A wireless VR video transmission method assisted by mobile edge computing according to the present invention has the following advantages:

[0080] The present invention considers a wireless VR video transmission network assisted by mobile edge computing, where the wireless access point (AP) and VR users are equipped with multiple antennas. By designing appropriate transmit precoding, the throughput of wireless VR video transmission is improved, and the VR video transmission delay is reduced. By jointly optimizing the rendering offloading and resource allocation schemes, according to the requirements of VR video transmission delay and distortion, the rendering offloading and resource allocation schemes are continuously iteratively optimized to achieve a good compromise among energy consumption, distortion, and delay. In addition, considering adaptively adjusting the encoding parameter (QP) of video slices to compress the FOV with acceptable distortion and balance the distortion, transmission delay, and energy consumption of VR videos. The method proposed by the present invention can reduce the rendering calculation delay and relieve the rendering calculation pressure and energy consumption of VR users. Description of the Drawings

[0081] Figure 1 is an application scenario diagram of a wireless VR video transmission method assisted by a mobile edge computing node provided in an embodiment of the present invention;

[0082] Figure 2It is a flowchart showing a method for wireless VR video transmission assisted by a mobile edge computing node provided in this embodiment of the present invention. Detailed implementation manners

[0083] To better understand the purpose, structure and function of the present invention, the following further describes in detail a method for wireless VR video transmission assisted by a mobile edge computing node of the present invention with reference to the accompanying drawings.

[0084] See Figure 1 - Figure 2 , this embodiment provides a method for wireless VR video transmission assisted by a mobile edge computing node. Taking a user experiencing VR video as an example to introduce this method, specifically as Figure 1 shown, one VR user watches a VR video, and the video rendering is completed on the user side or the edge computing server. The edge computing server is connected to the wireless access point by a wired manner. To meet the requirements of the intensive computing and latency-sensitive tasks of VR video rendering, the wireless access point, as the control center, needs to determine the VR video coding parameters, rendering offloading strategy, MIMO precoding matrix design and computing resource allocation according to the channel conditions of the VR video transmission system, the computing resources of the VR user and the edge computing server, the communication resources of the wireless access point, and the QoE requirements of the VR user.

[0085] Figure 2 It is a flowchart of the implementation process of the present invention, and the specific implementation steps are as follows:

[0086] Step S1: Establish a VR user video transmission system assisted by an edge computing server. This system includes one VR video provider, one wireless access point, one edge computing server and one VR user. Among them, the VR user is configured with N r antennas, and the wireless access point is configured with N t antennas; the VR user and the edge computing server are configured with computing units to process rendering tasks; the VR user communicates wirelessly with the wireless access point, the wireless access point is connected to the VR video provider by wire, the edge computing server is embedded in the wireless access point, and the VR user can offload the rendering calculation task to the edge computing server; the wireless access point, as the control center, completes data collection and information distribution.

[0087] Step S2: The VR user sends a VR video demand instruction to the wireless access point. The edge computing server first extracts the 2D field of view (FOV) from the 2D video according to the VR user's gaze point and user pose information, and then determines whether the 2D FOV is rendered as a 3D FOV at the edge computing server or the VR user according to the system information of the VR video transmission system collected by the control center. Finally, it is played on the VR user's head-mounted display. Among them, the system information collected by the control center includes the duration T of a playback window p, the number N of fixation point video clips v , the number N of non-fixation point video clips s , the number N of pixel points of a single video clip P , the channel state information between the VR user and the wireless access point, the distortion threshold D th , the total computing resource f of the edge computing server tot，A , the total computing resource f of the VR user tot，L ;

[0088] Step S3: According to the system information collected by the control center in Step S2, design a VR video coding parameter optimization algorithm to determine the VR video coding parameters.

[0089] Specifically, in this embodiment, Step 3 specifically includes:

[0090] Step S301: Input the system information in Step S2, and initialize the rendering offloading decision variable, the MIMO precoding matrix, and the computing resource allocation of the edge computing server and the VR user that satisfy the system operation constraints;

[0091] Step S302: Use the interior point method to solve the fixation point video clip bitrate optimization problem to obtain the optimal bitrate of the fixation point video clip. The specific optimization problem is as follows:

[0092]

[0093] s.t. γ v ∈[γ l , γ h , (2)

[0094] D ≤ D th , (3)

[0095]

[0096]

[0097] where γ v = γ(q v ) represents the bitrate of the fixation point video clip, and the specific expression is

[0098]

[0099] where γ(·) represents a function, q v represents the coding parameters of the fixation point video clip, δ1 and σ1 are positive parameters related to the video content characteristics, and v1 and v2 refer to parameters related to the coding structure; γ h = δ1 exp(σ1(v1 - Q l ) / v2) and γ l= δ1 exp(σ1(v1 - Q h )) / v2), Q l represents the minimum coding parameter of the fixation video slice, Q h represents the maximum coding parameter of the fixation video slice; w1, w2, and w3 represent intermediate variables, w1 = wβ A x r , w2 = wβ A (1 - x r ), wx = wβ L x r , where w represents the weight coefficient for balancing energy consumption and distortion, β A represents the energy consumption weight coefficient of the wireless access point, β L represents the energy consumption weight coefficient of the local VR user; x r ∈{0, 1} represents the rendering decision variable, x r = 1 indicates that the rendering task is executed locally, x r = 0 indicates that it is executed on the edge computing server; D represents the distortion of the FOV, and its expression is as follows:

[0100]

[0101] where ξ ∈ (0.5, 1) represents the weight of distortion, represents the distortion size of the fixation video slice, δ2 and σ2 are positive parameters related to the video content characteristics, represents the distortion size of the non - fixation video slice, q s refers to the coding parameter of the non - fixation video slice; represents the size of the 2D FOV data extracted per unit window, represents the size of the 2D FOV data stored at the wireless access point, represents the size of the 3D FOV data rendered by the edge computing server per unit window, represents the size of the 3D FOV data stored at the wireless access point, represents the maximum storage space of the edge computing server; refers to the energy consumed by the wireless access point to transmit 2D FOV, refers to the energy consumed by the edge computing server to render 2D FOV, represents the energy consumed to transmit 3D FOV, refers to the energy consumed by the local VR user to render 2D FOV, and The expressions of are as follows:

[0102]

[0103]

[0104]

[0105]

[0106] Among them, the ratio κ represents the degree of increase in transmission overhead, P refers to the MIMO precoding matrix, and μ L represents the computing power related to the hardware structure of the local VR user, and μ A represents the computing power related to the hardware of the edge computing server, and η L refers to the number of computing cycles required for the local VR user to process 1 bit of data, and η A refers to the number of computing cycles required for the edge computing server to process 1 bit of data, and f L represents the computing resources allocated to the local VR user, and f A represents the computing resources allocated to the edge computing server; B F represents the data size of 2DFOV, and its expression is as follows:

[0107] B F = N p (γ(q v )N v +γ(q s )N s ), (12)

[0108] Among them, B o = αB F refers to the data volume size obtained by 3D FOV. The constant parameter α ≥ 2 indicates that it contains at least the playback information of two eyes; R represents the wireless transmission rate, and its expression is as follows:

[0109]

[0110] Among them, H refers to the channel matrix between the wireless access point and the VR user, and W d refers to the channel bandwidth, σ 2 refers to the variance of the noise on a single antenna port, represents the N r th order identity matrix, and P H represents the conjugate transpose of the precoding matrix P, and H H represents the conjugate transpose of the channel matrix H;

[0111] Step S303: Iterate step S302 until the objective function of the formula converges to obtain the optimal bit rate γ v .

[0112] Step S304: Calculate the optimal coding parameters of the fixation point video slice according to the following to obtain the optimal

[0113]

[0114] wherein the function is the ceiling function.

[0115] Step S4: Design a rendering offloading algorithm and determine a rendering offloading strategy according to the system information collected by the control center in Step S2 and the VR video encoding parameters obtained in Step S3.

[0116] Specifically, in this embodiment, Step 4 specifically includes:

[0117] Step S401: Input the system information in Step S2, and initialize the encoded parameters of the foveated video slices that meet the system operation constraint conditions, the MIMO transmission precoding matrix, the computing resources allocated to the edge computing server and the local VR user.

[0118] Step S402: Solve the rendering decision optimization problem by using the interior point method. The specific optimization problem is as follows:

[0119]

[0120] s.t. x r ∈ [0, 1], (16)

[0121] x r (x r - 1) = 0, (17)

[0122]

[0123] wherein represents the amount of rendered FOV data stored in the local VR user, represents the amount of 3D FOV data rendered per unit window of the local VR user; v L represents the playback rate of the VR user, b th represents the cache threshold of the rendered data. When the rendered data is lower than this threshold, the video playback will pause;

[0124] Step S403: Iterate Step 402 until the objective function of the formula converges to obtain the relaxed optimal decision variable

[0125] Step S404: Calculate the optimal rendering decision variable according to the following formula

[0126]

[0127] where ρ represents the decision threshold constant.

[0128] Step S5: Design a MIMO transmission precoding matrix optimization algorithm and determine the precoding matrix according to the system information collected by the control center in step S2, and the VR video encoding parameters and rendering offloading strategies determined in steps S3 and S4.

[0129] Specifically, in this embodiment, step 5 specifically includes:

[0130] Step S501: Input the system information in step S2, and initialize the encoding parameters of the fixation point video slice, the rendering offloading decision variable, and the computing resource sizes allocated to the edge computing server and the local VR user that satisfy the system operation constraint conditions;

[0131] Step S502: Calculate the first-order Taylor expansion of R Its expression is as follows:

[0132]

[0133] where P (n) refers to the precoding matrix obtained by the nth iterative calculation, represents the first derivative of R.

[0134] Step S503: Update the variable z, and the update expression is as follows:

[0135]

[0136] Step S504: Solve the MIMO precoding matrix optimization problem using the interior point method. The specific optimization problem is as follows:

[0137]

[0138]

[0139]

[0140]

[0141]

[0142] where represents the maximum storage capacity of the local VR user, and P max represents the maximum transmission power of the wireless access point.

[0143] Step S505: Iterate steps S502 to S504 until the objective function of the formula converges to obtain the optimal MIMO transmission precoding matrix P * .

[0144] Step S6: Design a computing resource allocation algorithm for the edge computing server and VR users based on the VR video encoding parameters, rendering offloading strategy, and precoding matrix determined in steps S3 - S5, and determine the allocation of computing resources.

[0145] Specifically, in this embodiment, step 6 specifically includes:

[0146] The optimal solution for the computing resource allocation of the edge computing server is expressed as:

[0147]

[0148] The optimal solution for the computing resource allocation of the local VR user is expressed as:

[0149]

[0150] Among them, represents the minimum value of the computing resource allocation of the edge computing server, represents the maximum value of the computing resource allocation of the edge computing server, represents the minimum value of the computing resource allocation of the VR user locally, represents the maximum value of the computing resource allocation of the VR user locally, and the specific expressions are as follows:

[0151]

[0152]

[0153]

[0154]

[0155] Step S7: Iteratively load steps S3 to S6 until the weighted sum of the total system energy consumption and distortion converges, and finally determine the VR video encoding parameters, rendering offloading decision, MIMO transmission precoding, and computing resource allocation.

[0156] The parts not detailed in the present invention are all well-known technologies to those skilled in the art.

[0157] It can be understood that the present invention is described through some embodiments. Those skilled in the art know that, without departing from the spirit and scope of the present invention, various changes or equivalent replacements can be made to these features and embodiments. Additionally, under the teaching of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application belong to the scope protected by the present invention.

Claims

1. A method for wireless VR video transmission assisted by mobile edge computing, characterized in that, Including the following steps: Step S1, establish a VR video transmission system assisted by an edge computing server; it includes a VR video provider, a wireless access point, an edge computing server, and VR users, where the VR users are configured with N r antennas, and the wireless access point is configured with N t antennas; the VR users and the edge computing server are configured with computing units to process rendering tasks; the VR users communicate wirelessly with the wireless access point, the wireless access point is wired to the VR video provider, the edge computing server is embedded in the wireless access point, and the VR users unload the rendering calculation tasks to the edge computing server or complete the rendering tasks locally; the wireless access point serves as a control center to collect the system information of the VR video transmission system, make decisions, and distribute control instructions; Step S2: The VR user sends a VR video demand instruction to the wireless access point. The edge computing server first extracts the 2D field of view (FOV) from the 2D video according to the VR user's fixation point and VR user pose information, and then determines whether the 2D FOV is rendered as a 3D FOV at the edge computing server or the VR user according to the system information of the VR video transmission system collected by the control center. Finally, it is played on the VR user's head-mounted display; Step S3: Design a VR video coding parameter optimization algorithm according to the system information collected by the control center in step S2, and determine the VR video coding parameters; Step S4: Design a rendering offloading algorithm according to the system information collected by the control center in step S2 and the VR video coding parameters obtained in step S3, and determine the rendering offloading strategy; Step S5: Design a MIMO transmission precoding matrix optimization algorithm according to the system information collected by the control center in step S2, the VR video coding parameters determined in step S3, and the rendering offloading strategy determined in step S4, and determine the precoding matrix; Step S6: Design a computing resource allocation algorithm for the edge computing server and the VR user according to the VR video coding parameters, rendering offloading strategy, and precoding matrix determined in steps S3 - S5, and determine the computing resource allocation; Step S7: Iteratively load steps S3 to S6 until the weighted sum of the total energy consumption and distortion of the VR video transmission system converges, and finally determine the VR video coding parameters, rendering offloading decision, precoding matrix, and computing resource allocation.

2. The method for wireless VR video transmission assisted by mobile edge computing according to claim 1, wherein In the step S2, the system information collected by the control center includes the duration T of a playback window p , the number N of fixation video clips v , the number N of non-fixation video clips s , the number N of pixel points of a single video clip p , the channel state information between the VR user and the wireless access point, the distortion threshold D th , the total computing resources f of the edge computing server tot，A , the total computing resources f of the VR user tot，L .

3. The method for wireless VR video transmission assisted by mobile edge computing according to claim 2, wherein The steps of the VR video coding parameter optimization algorithm in step S3 include: Step S301: Input the system information in step S2, and initialize the rendering offloading decision variable, MIMO precoding matrix, and computing resource allocation for the edge computing server and the VR user that satisfy the system operation constraints; Step S302: Use the interior point method to solve the fixation point video slice bitrate optimization problem to obtain the optimal bitrate of the fixation point video slice. The specific optimization problem is as follows: s.t.γ v ∈[γ l ,γ h , (2) D≤D th , (3) where γ v = γ(q v ) represents the bit rate of the fixation video slice, and the specific expression is where γ(.) represents a function, q v represents the encoding parameter of the fixation video slice, δ1 and σ1 are positive parameters related to the video content characteristics, and v1 and v2 denote parameters related to the encoding structure; γ h = δ1exp(σ1(v1 - Q l ) / v2) and γ l = δ1exp(σ1(v1 - Q h ) / v2), Q l represents the minimum encoding parameter of the fixation video slice, and Q h represents the maximum encoding parameter of the fixation video slice; w1, w2, and w3 denote intermediate variables, w1 = wβ A x r , w2 = wβ A (1 - x r ), w a = wβ L x r , where w represents the weight coefficient for balancing energy consumption and distortion, β A represents the energy consumption weight coefficient of the wireless access point, and β L represents the energy consumption weight coefficient of the local VR user; x r ∈{0, 1} represents the rendering decision variable, x r = 1 indicates that the rendering task is executed locally, and x r = 0 indicates that it is executed on the edge computing server; D represents the distortion of the FOV, and its expression is as follows: Among them, ξ∈(0.5, 1) represents the weight of distortion, represents the distortion size of the foveated video slice, and δ2 and σ2 are positive parameters related to the video content characteristics. represents the distortion size of the non-foveated video slice, q s refers to the encoding parameter of the non-foveated video slice; represents the size of the 2D FOV data extracted from the unit window, represents the size of the 2D FOV data stored at the wireless access point, represents the size of the 3D FOV data rendered by the edge computing server in the unit window, represents the size of the 3D FOV data stored at the wireless access point, represents the maximum storage space of the edge computing server; refers to the energy consumed by the wireless access point to transmit the 2D FOV, refers to the energy consumed by the edge computing server to render the 2D FOV, represents the energy consumed to transmit the 3D FOV, refers to the energy consumed by the local VR user to render the 2D FOV, and The expressions of are as follows: Among them, the ratio κ represents the degree of increase in transmission overhead, P refers to the MIMO precoding matrix, and μ L represents the computing power related to the hardware structure of the local VR user, μ4 represents the computing power related to the hardware of the edge computing server, and η L refers to the computing cycles required for the local VR user to compute and process 1 bit of data, and η A refers to the computing cycles required for the edge computing server to process 1 bit of data, and f L represents the computing resources allocated to the local VR user, and f A represents the computing resources allocated to the edge computing server; B F represents the data size of the 2D FOV, and its expression is as follows: B F = N p (γ(q v )N v + γ(q s )N s ), (12) Among them, B o = αB F refers to the amount of data obtained from the 3D FOV. The constant parameter α ≥ 2 indicates that it contains at least the playback information of two eyes; R represents the wireless transmission rate, and its expression is as follows: where H represents the channel matrix between the wireless access point and the VR user, W d represents the channel bandwidth, σ 2 represents the variance of the noise on a single antenna port, denotes the N r th order identity matrix, P H represents the conjugate transpose of the precoding matrix P, H H represents the conjugate transpose of the channel matrix H; Step S303: Iterate step S302 until the objective function of the formula converges to obtain the optimal bit rate γ v ; Step S304: Calculate the optimal encoding parameters for the fixation video clip according to the following to obtain the optimal where The function is the ceiling function.

4. The method for wireless VR video transmission assisted by mobile edge computing according to claim 3, wherein The steps of the rendering offloading optimization algorithm in step S4 include: Step S401: Input the system information in step S2, and initialize the coding parameters of the fixation point video slice, MIMO transmission precoding matrix, and the size of the computing resources allocated to the edge computing server and the local VR user that satisfy the system operation constraints; Step S402: Use the interior point method to solve the rendering decision optimization problem. The specific optimization problem is as follows: s.t.x r ∈[0, 1], (16) x r (x r - 1) = 0, (17) Among them, represents the size of the rendered FOV data stored in the local VR user, represents the data volume of the 3D FOV rendered by the local VR user per unit window; v L represents the playback rate of the VR user, b th represents the cache threshold of the rendered data. When the rendered data is lower than this threshold, video playback will pause; Step S403: Iterate step 402 until the objective function of the formula converges to obtain the relaxed optimal decision variable Step S404: Calculate the optimal rendering decision variable according to the following formula where ρ represents the decision threshold constant.

5. The method for wireless VR video transmission assisted by mobile edge computing according to claim 4, wherein, The steps of the MIMO precoding matrix optimization algorithm in step S5 include: Step S501: Input the system information in step S2, and initialize the coding parameters of the fixation point video slice, rendering offloading decision variable, and the size of the computing resources allocated to the edge computing server and the local VR user that satisfy the system operation constraints; Step S502: Calculate the first-order Taylor expansion of R The expression is as follows: Among them, P (n) refers to the precoding matrix obtained by the nth iterative calculation, represents the first derivative of R; Step S503: Update the variable z, and the update expression is as follows: Step S504: Use the interior point method to solve the MIMO precoding matrix optimization problem. The specific optimization problem is as follows: Among them represents the maximum storage capacity of the local VR user, P max represents the maximum transmission power of the wireless access point; Step S505: Iterate from step S502 to step S504 until the objective function of the formula converges, and obtain the optimal MIMO transmit precoding matrix P * .

6. The mobile edge computing-assisted wireless VR video transmission method according to claim 5, wherein The method for allocating computing resources for the edge computing server and the local VR user in step S6 is expressed as: The optimal solution for the computing resource allocation of the edge computing server is expressed as: The optimal solution for the computing resource allocation of local VR users is expressed as: Among them, represents the minimum value of the computing resource allocation of the edge computing server, represents the maximum value of the computing resource allocation of the edge computing server, represents the minimum value of the local computing resource allocation of VR users, represents the maximum value of the local computing resource allocation of VR users, and the specific expressions are as follows:

Images