Resource allocation method and system for VR cloud game in meta universe in edge environment

Abstract

The invention discloses a resource allocation method and system for a VR cloud game in a meta universe in an edge environment. The method comprises the following steps: constructing a task execution scene of a VR cloud game in a meta universe in an edge environment, and detailedly describing a task process of the VR cloud game; determining resources needing to be allocated by the VR cloud game scene in the meta universe in the edge environment; determining a heterogeneous user set, wherein the resolutions of the user VR devices in the scene are different; constructing a user experience quality function in the VR cloud game scene in the meta universe in the edge environment; an optimization problem is constructed, constraint conditions are determined, and the average user experience quality is maximized under the condition of limited resources; the number of decision variables in the optimization problem is equal to the number of resources, and due to mutual coupling of the decision variables, an alternating optimization method is adopted for iterative optimization, and an optimal resource allocation scheme is solved; in the alternative optimization process, it is proved that the optimization problem of each decision variable is a convex optimization problem, constraint conditions are inequality constraints, and a KKT condition is adopted for solving. Limited edge server resources are distributed in the static edge environment, so that the average experience quality of users in the system reaches the maximum value.

Description

Technical Field

[0001] The present invention relates to the field of edge computing, and in particular to a resource allocation method and system for VR cloud games in a metaverse in an edge environment. Background Art

[0002] Edge computing is a technology trend that has attracted much attention in recent years. Edge computing pushes computing resources and data processing capabilities from traditional cloud data centers to the edge of the network to meet real-time and low-latency requirements. This distributed computing architecture provides faster and more efficient computing resources for various application scenarios, especially in scenarios that require instant decision-making and interaction, such as autonomous driving, smart IoT, and virtual reality. The metaverse requires a large amount of computing and storage resources, and edge computing can provide the metaverse with computing resources closer to users to reduce latency and improve user experience. In addition, user-generated content and interaction in the metaverse also require powerful computing support, which is exactly where edge computing excels. By combining the metaverse with edge computing, we can achieve a more interactive and immediate metaverse experience and expand its application areas, such as remote collaboration, virtual tourism, and telemedicine.

[0003] Cloud gaming, sometimes also called on-demand gaming or game streaming, is an online game that runs video games on a remote server and streams the game's output (video, sound, etc.) directly to the user's device, or more commonly, plays the game remotely from the cloud. Head-mounted devices such as virtual reality (VR) glasses and extended reality (XR) glasses are the first application devices for people to contact the metaverse. Through these devices, people can experience an immersive and seamless metaverse experience. By wearing VR glasses and XR glasses, people can hold online meetings in the metaverse as if they were offline, and can also travel online to experience the virtual world created by digital twin technology, but these devices usually have limited computing resources and energy resources, and it is difficult to maintain a high load state for a long time. Immersive VR games are usually presented in the form of 360-degree panoramic images with high resolution. Usually, the user's head movement, gesture operations, etc. are collected in real time to track the picture. It is difficult to provide users with high-quality game services relying on the computing power of the head-mounted device itself, so it is considered to use edge servers to perform image rendering and then transmit the image to the user's device. Metaverse immersive games usually take up tens of GB of storage resources due to their rich graphics and numerous interactions. Secondly, if users want to experience Metaverse immersive games, they must download them in advance, which will cause a long wait and affect the user experience.

[0004] The Metaverse service provider stores game resources on the edge server in advance, and users obtain game qualifications by subscribing to the service. The user starts the game locally, the game process is executed on the edge server, and the edge transmits the game screen to the user through streaming media. It meets the user's image quality and latency requirements, and brings users an immersive and seamless experience. However, the computing resources and communication resources of the edge server are limited and cannot be allocated to connected users without limit, so my research is mainly to design a reasonable resource allocation algorithm so that the computing resources and communication resources of the edge server are reasonably allocated to ensure the maximization of user experience. Summary of the invention

[0005] Purpose of the invention: In view of the shortcomings of the prior art, the present invention provides a resource allocation method and system for VR cloud games in the metaverse in an edge environment. When the user set and user resolution are known, the GPU resources of the edge server and the bandwidth resources of the communication base station are reasonably allocated to maximize the user experience quality.

[0006] Technical solution: In order to achieve the above invention objectives, the technical solution of the present invention is as follows:

[0007] An edge cloud gaming system for VR cloud gaming in the metaverse in edge environments, including:

[0008] A VR cloud gaming system in the metaverse under the edge environment is constructed, and the task flow of VR cloud gaming is described in detail. The system includes an edge server, a communication base station, and multiple cloud gaming users. Cloud gaming users connect to the edge server through the base station and offload the running and rendering tasks of cloud gaming to the edge server. The edge server transmits the picture frames to the user and displays them on the user's device. The task flow of VR cloud gaming in the metaverse under the edge environment includes the following steps: (1) The user device collects the user's input instructions, encodes them and transmits them to the edge server. (2) The server decodes the operation flow and simulates the user operation. (3) The server generates the original picture frame. (4) The edge server encodes the picture and audio. (5) The edge server sends the encoded audio and video data packets to the user through the downlink channel. (6) The user decodes them on the device and displays the picture on the device.

[0009] A resource allocation method for VR cloud games in a metaverse in an edge environment, comprising the following steps:

[0010] According to the task execution scenario of VR cloud games in the metaverse under the edge environment, the optimization problem P1 is constructed: an edge server has limited GPU resources G and bandwidth resources B. At a certain moment, N users connect to this edge server and offload the game process to this server. The user set is represented as for The resolution of the user's device is expressed in sn The user's experience quality score Q n Defined as the frame rate f n and the delay t n The optimization problem P1 is how to allocate GPU resources and bandwidth resources to achieve the average quality of experience. Maximization, the optimization objective is expressed as:

[0011] This optimization problem contains two decision variables, namely GPU resource allocation and bandwidth resource allocation, and there is coupling. In addition, it is proved that the optimization problems of GPU resource allocation and bandwidth resource allocation are both convex optimization problems. Therefore, alternating iteration is used to solve the optimization problem. First, GPU resources are evenly allocated, and the bandwidth resource allocation is obtained by solving the optimization problem. Then, the GPU resource allocation is solved using the obtained bandwidth resource allocation. Iterate continuously until the stopping condition is met.

[0012] Preferably, for use TFLOPs means generating a frame with a resolution of s n The edge server allocates computing resources A to user n. nG , ideally the number of frames per second that can be generated for the user

[0013] Preferably, the user's game delay refers to the time from when the user performs any operation to when the corresponding game screen is displayed on the user's device, which includes five parts: the delay t of uploading the user's operation instruction to the edge server uplink ; The time delay for the edge server to generate a picture frame after receiving the instruction is t generate ; The delay of the edge server encoding the picture frame is t encode ; The delay of picture frame transmission to the user is t downlink ; The decoding delay of the user terminal device is t decode Taking user n as an example, the game delay is expressed as:

[0014] Furthermore, since the user command data packet is small and the uplink channel is relatively stable, the delay of user command upload is not considered for the time being. In addition, in the cloud gaming scenario, the computing power of the user device is used for picture decoding, and the workload is low, so the delay of user-side decoding is not considered. The frame generation time is equal to the frame interval time, v represents the GPU encoding rate, μ represents the encoding compression rate, and k represents the bit width of the pixel in the image. A nB represents the bandwidth resources allocated by the edge server to user n, r nrepresents the transmission rate of the edge server transmitting the picture to user n, the transmission power is p, the channel gain is g, and the noise power density is σ 2 , Resolution is s n The frame transmission delay of the picture is expressed as:

[0015] Preferably, the frame rate quality of experience function and the delay quality of experience function are fitted. f (f n )=a f log f n +b f Fit the frame rate experience quality function. Using the formula Q t (t n )=a t log t n +b t Fitting delay quality of experience function.

[0016] Preferably, the user experience quality function is expressed as: Q n =ω f *Q f (f n )+ω t *Q t (t n ), f n is the frame rate of the game rendered by the edge server for the user, t n is the user's game delay, weight coefficient ω f and ω t Used to adjust Q f (f n ) and Q t (t n ) as a percentage of user experience quality.

[0017] Preferably, the objective function of the optimization problem P1 contains two decision variables, and there is coupling between the two decision variables, which makes it difficult to solve directly. Therefore, an alternating iteration method is used to solve the optimization problem.

[0018] Furthermore, we first select GPU resource allocation A, which appears more frequently in the objective function and handles more complex nG . The average allocation strategy is adopted to evenly allocate the GPU resources of the edge server to the cloud game users. At this time, the optimization problem is only related to the bandwidth resource allocation variable A nBIt is proved that the optimization problem is a convex optimization problem, and the constraints are inequality constraints. The KKT condition is used to solve it, and the bandwidth resource allocation strategy is obtained. Then the bandwidth resource allocation strategy obtained by the solution is substituted back into the optimization problem. At this time, the decision variable of the optimization problem is only the GPU resource allocation A. nG , we continue to prove that the optimization problem is a convex optimization problem, and the constraints are inequality constraints. We use KKT conditions to solve it, and get the GPU resource allocation strategy. We regard a bandwidth resource allocation solution and a GPU resource allocation solution as a round of iteration, and repeat the iteration continuously. Introduce the decision factor ε, Defined as The experience quality score after one iteration is Stop iteration when .

[0019] The present invention also provides an edge VR cloud gaming system, which is composed of an edge server, a communication base station and VR users. The VR user is connected to the nearest edge server through the communication base station, and the edge server provides cloud gaming services for the user.

[0020] The present invention also provides a computer device, comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of resource allocation for VR cloud games in a metaverse in an edge environment as described above are implemented.

[0021] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of resource allocation for VR cloud games in a metaverse in an edge environment as described above.

[0022] Beneficial effects: In order to solve the problem of insufficient computing power of user VR devices in the metaverse scenario, an edge VR cloud gaming system for metaverse users in an edge environment is provided, which offloads the game process to the edge server for execution. In addition, in order to efficiently utilize the GPU resources of the edge server and the bandwidth resources of the communication base station so that the system can provide services to multiple users at the same time, and in order to maximize the service quality brought by limited resources, a resource allocation method for VR cloud games in the metaverse in an edge environment is proposed. For game users, when the device resolution is fixed, higher game frame rates and lower game delays will bring users a better gaming experience. Existing work has all taken resolution as the optimization goal, and no work has considered the frame rate factor, so it is not suitable for the cloud gaming scenario. The resource allocation algorithm of the present invention can allocate reasonable GPU resources and bandwidth resources to users in the edge VR cloud gaming scenario where the user resolution is known, so that the average user experience quality of the entire user cluster is maximized. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is an application scenario diagram of an edge VR cloud gaming system;

[0024] Figure 2 It is a flow chart of the resource allocation method for VR cloud games in the metaverse in edge environments;

[0025] Figure 3 This is the edge VR cloud game user frame rate experience quality curve;

[0026] Figure 4 This is the latency experience quality curve of edge VR cloud gaming users;

[0027] Figure 5 It is a flow chart of the resource allocation algorithm for cloud gaming users in the metaverse under a static edge environment. DETAILED DESCRIPTION

[0028] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0029] Figure 1The following is an application scenario diagram of an edge VR cloud gaming system. At a certain moment, there are N cloud gaming users connected to the edge server through the base station, and the running and rendering tasks of the cloud gaming are offloaded to the edge server. The edge server transmits the rendered picture frames to the user and displays them on the user's device. The task flow of VR cloud gaming in the metaverse under the edge environment includes the following steps: (1) The user device collects the user's input instructions, encodes them and transmits them to the edge server (2) The server decodes the operation flow and simulates the user operation (3) The server generates the original picture frame (4) The edge server encodes the picture and audio (5) The edge server sends the encoded audio and video data packets to the user through the downlink channel (6) The user decodes on the device and displays the picture on the device.

[0030] Reference Figure 2 The present invention provides a resource allocation method for VR cloud games in a metaverse in an edge environment, comprising the following steps:

[0031] Step S1, modeling the optimization problem, frame rate model, delay model and experience quality model is described as follows:

[0032] The edge server has limited GPU resources G, and the base station has limited bandwidth resources B. The user set is represented as for The resolution of the user's device is expressed in s n The GPU resources allocated by the edge server to the user are represented by A nG , the bandwidth resource allocated to the user is represented by A nB According to A nG and A nB Calculate the game frame rate f of user n n and the delay t n , and then calculate the user's experience quality score Q n The optimization problem is how to allocate GPU resources and bandwidth resources to achieve the average experience quality. maximize.

[0033]

[0034] Define this problem as P1.

[0035] for use TFLOPs means generating a frame with a resolution of s n The edge server allocates computing resources A to user n. nG , ideally the number of frames per second that can be generated for the user

[0036] The user's game delay refers to the time from when the user performs any operation to when the corresponding game screen is displayed on the user's device. It consists of five parts: the delay of uploading the user's operation instruction to the edge server t uplink ; The time delay for the edge server to generate a picture frame after receiving the instruction is t generate ; The delay of the edge server encoding the picture frame is t encode ; The delay of picture frame transmission to the user is t downlink ; The decoding delay of the user terminal device is t decode Taking user n as an example, the game delay is expressed as follows:

[0037]

[0038] Furthermore, since the user command data packet is small and the uplink channel is relatively stable, the delay of user command upload is not considered for the time being. In addition, in the cloud gaming scenario, the computing power of the user device is used for picture decoding, and the workload is low, so the delay of user-side decoding is not considered. The frame generation time is equal to the frame interval time, v represents the GPU encoding rate, μ represents the encoding compression rate, and k represents the bit width of the pixel in the image. A nB represents the bandwidth resources allocated by the edge server to user n, r n represents the transmission rate of the edge server transmitting the picture to user n, the transmission power is p, the channel gain is g, and the noise power density is σ 2 , Resolution is s n The frame transmission delay of the picture is expressed as:

[0039] So the user's game delay is expressed as:

[0040]

[0041] Figure 3 and Figure 4 It is the frame rate experience quality function and delay experience quality function of cloud game users. f (f n )=a f log f n +b f Fit the frame rate experience quality function of cloud game users, and the fitting result is a f =1.352, b f = -1.270. t (t n )=a t log t n +b tFitting the delay experience quality function of cloud game users, the fitting result is a t =-1.5289, b t =11.0185.

[0042] The user experience quality function is defined as follows:

[0043] Q n =ω f *Q f (f n )+ω t *Q t (t n )

[0044] f n is the frame rate of the game rendered by the edge server for the user, t n is the user's game delay, weight coefficient ω f and ω t Used to adjust Q f (f n ) and Q t (t n ) as a percentage of user experience quality.

[0045] Substitute the frame rate experience quality function and the delay experience quality function into the optimization problem P1. In order to simplify the objective function, we temporarily ignore the constant term and the average term, and obtain the simplified optimization problem P2 as follows:

[0046]

[0047] Where ω1, ω2 are both greater than 0, and ω1 = ω f *a f ,ω2=ω t *a t .

[0048] Step S2, solving bandwidth allocation based on known GPU resource allocation.

[0049] The decision variables of the optimization problem P2 are GPU resource allocation A nG and bandwidth resource allocation A nB , and only the delayed experience quality is related to A nB Therefore, we use the alternating optimization method to first fix the GPU resource allocation A G =[A 1G ...A nG ...A NG ] value, optimize bandwidth allocation A B =[A 1B ...A nB ...A NB], and then use bandwidth allocation to optimize GPU resource allocation.

[0050] First, we assume that the GPU resource allocation is known. At this point, the optimization problem P2 is only related to the bandwidth resource allocation A nB It is related. It can be written as optimization problem P3:

[0051]

[0052] We first transform the optimization problem P3 into the standard form of the optimization problem, invert the problem, rewrite the maximum value into the minimum value, and rewrite it into the optimization problem P4:

[0053]

[0054] The multivariate function F is a function of any variable A. iB -th order partial derivative of (1≤i≤N) The formula is as follows:

[0055]

[0056] The multivariate function F is a function of any variable A. iB (1≤i≤N) After taking the first-order partial derivative, for any other variable A jB (1≤j≤N and i≠j) find the second-order partial derivative are all 0. In addition, Then for variable A iB Find the second-order partial derivative as follows:

[0057]

[0058] Since all parameters are positive, Molecular Department So F for any variable A iB The second-order partial derivative of (1≤i≤N) is positive in its domain.

[0059] Therefore, the Hessian matrix of the objective function F can be obtained as follows:

[0060]

[0061] Since the Hessian matrix of the objective function F is a diagonal matrix, and its diagonal elements are positive numbers at any point on its domain, it can be proved that the matrix is ​​a positive definite matrix; and when the Hessian matrix of a multivariate function is a positive definite matrix, the corresponding multivariate function is a convex function.

[0062] In addition, the constraints of the optimization problem P4 are all linear, so its constraints constitute an affine function, and the affine function is a convex function. Therefore, the feasible domain set of the constraints of the optimization problem P4 is a convex set. Therefore, it can be proved that the optimization problem P4 is a convex optimization problem.

[0063] The convex optimization problem is solved by the KKT conditions (Karush-Kuhn-Tucker Conditions).

[0064] First, write the Lagrangian function of the optimization problem P4 as follows:

[0065]

[0066] The KKT conditions are:

[0067]

[0068] λ n ≥0, 0≤n≤N

[0069]

[0070] Next, the parameter values ​​in the KKT conditions are discussed by category and solved using the equation.

[0071] a) λ n ≠0, at this time Since we only consider providing cloud game rendering services to all users, that is, the bandwidth allocated to users is all positive, we discard this case.

[0072] b) λ n =0 and λ0=0. Since in the definition domain of the optimization problem P3, is always less than 0, so this situation does not hold.

[0073] c) λ n =0 and λ0≠0. At this time, according to the KKT condition, the equations are listed as follows:

[0074]

[0075] This system of equations has n+1 equations and n+1 unknowns. Theoretically, there is a unique solution. Next, solve this system of equations. Expand to get the following formula:

[0076]

[0077] Use substitution method to simplify the formula. And for any user n, C n , P n are all constants and are greater than 0. At this time, we can get the value of A nB The quadratic equation of one variable is as follows:

[0078]

[0079] According to the discriminant theorem, the equation has two unequal real roots. By discarding the negative real root, we get A nB The values ​​of are as follows:

[0080]

[0081] Substituting this value back into the first equation in the system of equations, we get the following equation. This equation contains only one unknown variable, λ0, but the left side of the equation is a polynomial summation and λ0 is in the radical form, which is difficult to solve using conventional methods. Therefore, we consider using numerical methods to solve the equation.

[0082]

[0083] After finding the value of λ0, the bandwidth allocation A can be calculated for each user. nB .

[0084] Step S3, solving GPU resource allocation when bandwidth allocation is known.

[0085] Allocate the bandwidth A B =[A 1B ...A nB ...A NB ] is substituted into the optimization problem P2. At this time, the optimization problem only contains A nG A decision variable, which is defined as the optimization problem P5:

[0086]

[0087] Similar to the proof of optimization problem P4, it can be proved that optimization problem P5 is also a convex optimization problem, and the constraints are inequality constraints, and can also be solved using KKT conditions.

[0088] The Lagrangian function of the optimization problem P5 is given below:

[0089]

[0090] The KKT conditions are:

[0091]

[0092] λ n ≥0, 0≤n≤N

[0093]

[0094] Next, the parameter values ​​in the KKT conditions are discussed by category and solved using the equation.

[0095] a) λ n ≠0, at this time Since we only consider providing cloud game rendering services to all users, that is, the GPU resources allocated to users are all positive values, we discard this case.

[0096] b) λ n =0 and λ0=0.

[0097]

[0098] Since in the definition domain of optimization problem P5, is always less than 0, so this situation does not hold.

[0099] c) λ n =0 and λ0≠0. At this time, according to the KKT condition, the equations are listed as follows:

[0100]

[0101] This system of equations has n+1 equations and n+1 unknowns. Theoretically, there is a unique solution. Next, solve this system of equations. The expanded formula is as follows:

[0102]

[0103] Arrange and convert, And in the case of fixed bandwidth resource allocation All have M n is a constant greater than 0, and we get nG The quadratic equation of

[0104]

[0105] According to the discriminant theorem, the equation has two unequal real roots. By discarding the negative real root, we get A nG The values ​​of are as follows:

[0106]

[0107] Substituting this value back into the first equation in the system of equations, we obtain the following equation, which contains only one unknown variable λ0. However, the equation is complex and difficult to solve by algebraic methods, so we also consider solving it by numerical methods.

[0108]

[0109] After finding the value of λ0, the bandwidth allocation A can be calculated for each user. nG .

[0110] Step S4, alternately iteratively optimize and solve the optimal resource allocation.

[0111] Figure 5 This is a flow chart of the resource allocation algorithm for cloud gaming users in the metaverse in a static edge environment. The specific steps are as follows:

[0112] The average quality of experience score of global users is defined as Will Defined as The experience quality score after one iteration. Introducing the decision factor ε, when The iteration stops when . The algorithm starts with initializing the relevant parameters. Input GPU available resources G, bandwidth available resources B, decision factor ε, influence factor ω f ,ω t .

[0113] First, an average allocation strategy is adopted, that is, Next, we use the known GPU resource allocation to solve the optimization problem P4 and get the bandwidth allocation. Then we use the bandwidth allocation to solve the optimization problem P5, complete a round of iteration, and get a new Then continue to iterate until The iteration is stopped when , and the resource allocation plan is obtained.

[0114] The present invention also provides an edge VR cloud gaming system, which is composed of an edge server, a communication base station and VR users. The VR user is connected to the nearest edge server through the communication base station, and the edge server provides cloud gaming services for the user.

[0115] The present invention also provides a computer device, comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of resource allocation for VR cloud games in a metaverse in an edge environment as described above are implemented.

[0116] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of resource allocation for VR cloud games in a metaverse in an edge environment as described above.

[0117] The above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, a person skilled in the art should understand that the specific implementation modes of the present invention may still be modified or replaced by equivalents, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention shall be included in the scope of protection of the claims of the present invention.

Claims

1. A resource allocation method for VR cloud games in a metaverse under an edge environment, characterized in that: The following steps are involved: According to the task execution scenario of VR cloud games in the metaverse under the edge environment, an optimization problem is constructed: the edge server has limited GPU resources G, the base station has limited bandwidth resources B, and at a certain moment, N users connect to this edge server through the base station and offload the game process to the server. The user set is represented as for The resolution of the user's device is expressed in s n The user's experience quality score Q n Defined as the frame rate f n and the delay t n The optimization problem is how to allocate GPU resources and bandwidth resources to achieve average quality of experience. The maximization,optimization problem is expressed as: This optimization problem contains two decision variables, namely GPU resource allocation and bandwidth resource allocation, and there is coupling. In addition, it is proved that the optimization problems of GPU resource allocation and bandwidth resource allocation are both convex optimization problems. Therefore, alternating iteration is used to solve the optimization problem. First, GPU resources are evenly allocated, and the bandwidth resource allocation is obtained by solving the optimization problem. Then, the GPU resource allocation is solved using the obtained bandwidth resource allocation. Iterate continuously until the stopping condition is met.

2. The method according to claim 1, characterized in that The task flow in the system includes: The task execution scenario of VR cloud games in the metaverse under the edge environment is constructed, and the task flow of VR cloud games is described in detail. The scenario includes an edge server, a communication base station, and multiple cloud game users. The cloud game user connects to the edge server through the base station and offloads the running and rendering tasks of the cloud game to the edge server. The edge server transmits the rendered picture frames to the user and displays them on the user's device. The task flow of VR cloud games in the metaverse under the edge environment includes the following steps: (1) The user device collects the user's input instructions, encodes them and transmits them to the edge server (2) The server decodes the operation flow and simulates the user operation (3) The server generates the original picture frame (4) The edge server encodes the picture and audio (5) The edge server sends the encoded audio and video data packets to the user through the downlink channel (6) The user decodes on the device and displays the picture on the device.

3. The method according to claim 2, characterized in that The user game frame rate is expressed as: for use TFLOPs means generating a frame with a resolution of s n The edge server allocates computing resources A to user n. nG , ideally the number of frames per second that can be generated for the user 4. The method according to claim 2, characterized in that: The user delay is expressed as: The user's game delay refers to the time from when the user performs any operation to when the corresponding game screen is displayed on the user's device. It consists of five parts: the delay of uploading the user's operation instruction to the edge server t uplink ; The time delay for the edge server to generate a picture frame after receiving the instruction is t generate ; The delay of the edge server encoding the picture frame is t encode ; The delay of transmitting the picture frame to the user is t downlink ; The decoding delay of the user terminal device is t decode Taking user n as an example, the game delay is expressed as follows:

5. The method according to claim 4, characterized in that User latency components include: a) Since the user command data packet is small and the uplink channel is relatively stable, we will not consider optimizing the latency of user command upload. In addition, in the cloud gaming scenario, the computing power of the user device is used for picture decoding, and the workload is relatively low, so we will not consider optimizing the latency of user-side decoding. b) The frame generation time is equal to the frame interval time, c) v represents the GPU encoding rate, μ represents the encoding compression rate, and k represents the bit width of the pixel in the image. d) A nB represents the bandwidth resources allocated by the edge server to user n, r n represents the transmission rate of the edge server transmitting the picture to user n, the transmission power is p, the channel gain is g, and the noise power density is σ 2 , Resolution is s n The frame transmission delay of the picture is expressed as:

6. The method according to claim 2, characterized in that The user experience quality function is expressed as: Q n =ω f *Q f (f n )+ω t *Q t (t n ) f n is the frame rate of the game rendered by the edge server for the user, t n is the user's game delay, weight coefficient ω f and ω t Used to adjust Q f (f n ) and Q t (t n ) as a percentage of user experience quality.

7. The method according to claim 2, characterized in that Solving the optimization problem by alternating iterations includes: First, the GPU resource allocation is fixed and the average allocation method is adopted. At this time, the optimization problem only contains one decision variable, namely, bandwidth allocation, which proves that the optimization problem is a convex optimization problem and is solved by the KKT condition. Then, the bandwidth allocation obtained is substituted back into the optimization problem. At this time, the optimization problem only contains one decision variable, namely, GPU resource allocation, which also proves that the optimization problem is a convex optimization problem and is solved by the KKT condition. After that, it is continuously iterated. Introduce the decision factor ε, Defined as The experience quality score after one iteration is Stop iteration when .

8. An edge VR cloud gaming system, the system consists of an edge server, a communication base station and VR users. VR users connect to the nearest edge server through the communication base station, and the edge server provides cloud gaming services for them.

9. A computer device, characterized in that: include: one or more processors; Memory; And one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of resource allocation for VR cloud games in a metaverse in an edge environment as described in any one of claims 1-7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of resource allocation for VR cloud games in a metaverse in an edge environment as described in any one of claims 1 to 7 are implemented.

Images