WO2023104186A1

WO2023104186A1 - Highly-efficient and low-cost cloud game system

Info

Publication number: WO2023104186A1
Application number: PCT/CN2022/137925
Authority: WO
Inventors: 许磊; 靳文波; 赵庆鹏
Original assignee: 许磊
Priority date: 2021-12-09
Filing date: 2022-12-09
Publication date: 2023-06-15
Also published as: CN114880107A

Abstract

A highly-efficient and low-cost cloud game system, comprising a user-side running system, a smart data access prediction subsystem, a smart service-side subsystem, a data preprocessing subsystem for general data, a data compression subsystem combining general compression and special compression, and a data distribution algorithm combining prediction and instructions. The present application belongs to the technical field of games, and in particular relates to a highly-efficient and low-cost cloud game system. During a game process, a service side distributes game source data to be used, and a user side receives the game source data and locally executes calculation and rendering. Therefore, the requirements for instantly playing once a game is opened, reducing the requirements for a network, reducing game lagging, supporting high-quality game pictures, etc. are met. Moreover, since the service side does not need to perform rendering, and what is distributed is compressed source data, the server cost and the bandwidth cost are greatly reduced, thereby solving the problem of it being impossible to achieve "economies of scale" in the current cloud game.

Description

A high-efficiency and low-cost cloud gaming system

technical field

The invention belongs to the technical field of games, and specifically refers to a high-efficiency and low-cost cloud game system.

Background technique

With the rapid development of computer and game technology, the capacity of games has grown explosively. The capacity of some large games has often exceeded 200GB, and the games of some companies have even exceeded 2000TB. On the other hand, the computing power (processing power of CPU and GPU) required by these games is relatively high. Therefore, under the existing technology, the distribution and operation of these large-scale games has become a problem, so how to reduce the cost of reaching users for games, that is, user acquisition costs and operating costs, has become a top priority.

technical problem

Cloud gaming is a solution that runs the game on a cloud server, transmits the game screen to the client in the form of streaming media, and transmits the input from the client (button, voice, motion sensory input, etc.) to the server. This solves the user's download, installation and running problems, but the cost is high (game purchase cost, server cost, bandwidth cost, operating cost, etc., of which server cost and bandwidth cost are the main parts), and due to the highly dynamic characteristics of the game (real-time generated , it is difficult to reuse), resulting in that the server cost cannot be diluted among users, that is, the server cost and bandwidth cost determine that the current cloud game implementation method cannot achieve "economy of scale", that is, the current cloud game unit cost does not increase with the increase in user scale. There was a marked decline.

technical solution

In order to solve the above problems, the present invention provides an efficient and low-cost cloud game system. During the game process, the server distributes the game source data to be used, and the client receives the game source data and performs calculation and rendering locally. It solves the needs of playing immediately, reducing network requirements, reducing game lag, and supporting high-quality game screens. And since there is no need for server-side rendering, and the compressed source data is distributed, server costs and bandwidth costs are greatly reduced, and the problem of "economies of scale" that cannot be solved in current cloud games is solved.

In order to realize the above-mentioned functions, the technical solution adopted by the present invention is as follows: a high-efficiency and low-cost cloud game system, including a client operating system, an intelligent data access prediction subsystem, an intelligent service terminal system, a data preprocessing subsystem for general data, Data compression subsystem combining general compression and special compression, data distribution algorithm combining prediction and instruction;

The user-side operating system is responsible for starting the game when it is running, and maps the game's access to data into local cache and server-side data access, which is transparent to the user; the runtime will continue to communicate with the server to report the user's current behavior and the game Context information, the server will intelligently predict the data that the user may use next; or pull the data that is currently needed but not in the cache; the runtime accepts the data returned by the server, and performs necessary decompression, decryption and Transcoding to meet the running requirements of the game, adding local cache. And manage the local cache according to user information, remaining storage space, server instructions, and specific algorithms;

During runtime, user behavior and game behavior will be collected, including key presses, key press duration, dwell time of each interface, game response, game running context, game data access and other information, and uploaded to the server for subsequent modeling;

The intelligent data access prediction subsystem selects Aproiri, naive Bayesian, Bayesian network, K-Means, KNN, DBSCAN, SVM, LSTM, CNN, AdaBoost, GBDT, RandomForest and other machine learning algorithms, and uses one of these algorithms or Multiple combinations to build a pool of machine learning algorithms;

Use different algorithms in the machine learning algorithm pool to process the collected user behavior and program behavior, and combine the types and characteristics of the data to establish data access prediction models for different users and different data; these models can be based on user behavior and program behavior, Predict the data that the user will need next;

For each data of each user, in the model evaluation stage, select the top models with the best performance according to the stability under small samples, prediction success rate, computational complexity and other indicators;

The intelligent service terminal system can comprehensively process the user information, device information, network status, user behavior and game context, etc. incoming from the client terminal, including login, authentication, authorization, selection of the appropriate intelligent data access prediction model, selection of the corresponding High-quality model and compression algorithm, send the data required by the user to the user in advance or on demand, and encrypt the data according to the settings during the process;

The data preprocessing subsystem of general data divides the data into data blocks of appropriate size according to the model and data characteristics established by the intelligent data access prediction subsystem, and selects different compression algorithms to compress the data blocks. Based on the user's use of specific data, the nature and function of the data, and the impact of the intelligent data access prediction subsystem on the user experience, the data blocks are dynamically allocated to the corresponding storage area on the server, that is, the cold data area. General data area, hot data area (such as PCIE accelerated storage area, server memory cache), accelerated data storage area (our company or third-party CDN);

A data compression subsystem that combines general and special compression. There are different types of game data, such as code, configuration, text, models, mesh, audio, video, textures, graphics, images, values, other formatted data, and generic data. For the code, configuration, text, value and other data that need to be accurately transmitted, according to its data characteristics and effectiveness requirements (compression, and decompression time consumption), select a specific general compression (lossless) algorithm (such as 7z, LZMA, zip, etc.) Compression with different compression parameters; for formatted data such as audio, video, texture, mesh, graphics, images, etc., use appropriate lossless or lossy compression algorithms (such as image png, jpeg, webp, etc., audio aac, mp3, etc., video mpeg, mpeg4, H264, H265, VP8, VP9, etc., and our self-developed model Mesh compression algorithm) and different bit rates or resolutions and other parameters, generate different quality compressed data. After receiving the data, the client's runtime will decompress and transcode as needed to obtain available data;

A data distribution algorithm that combines prediction and instruction. Under normal circumstances, the system downloads and pulls data based on user behavior and program behavior, as well as the model generated by the intelligent data access prediction subsystem; the data distribution subsystem also supports game developers to implant instructions according to the protocol, when the instruction execution condition is triggered , the data distribution subsystem will pull specific data according to the instruction. In the combination of prediction and instruction, data access can achieve the same zero network delay as accessing local data, so as to achieve no lag in the game process.

Beneficial effect

The present invention adopts the above-mentioned structure to obtain beneficial effects as follows: the high-efficiency and low-cost cloud game system provided by the present invention has a reasonable design and includes the following advantages:

(1) With the support of runtime, the game does not need to be downloaded or installed, and it can be played immediately; in the existing cloud game solution, the game is calculated and rendered on the server, and the game screen is transmitted to the client through the network in video format. Compared with the existing cloud game solutions, the present invention does not require server-side calculation and rendering, which reduces the cost of the server;

(2) The cloud game solution implemented by the present invention only needs to transfer game data during the game process, and the on-demand download and different levels of compression greatly reduce the traffic in distribution. In the existing cloud game solution, the game screen is transmitted to the client through the network in video format. Compared with existing cloud game solutions, the present invention can save over 99% of traffic costs;

(3) In existing cloud game solutions, the game screen is transmitted to the client through the network in video format, and user input also needs to be transmitted to the server through the network. Network delay is inevitable, and network jitter seriously affects user experience. The data access supported by the present invention can achieve the same zero network delay as accessing local data, and user input is processed locally without network transmission, so as to achieve no lag in the game process;

(4) Intelligent prediction of data access + multiple data compression reduces the requirements for network bandwidth and increases the tolerance for network jitter. Compared with existing cloud games, it greatly improves the user experience in weak network environments;

(5) Due to the local calculation and rendering, the game supported by the present invention can run in the ultra-high definition + ultra-high refresh rate mode, and has no additional requirements on the network. However, existing cloud games cannot support such high-quality game quality due to server costs, bandwidth costs, and network delay and jitter;

(6) The requirements of the game on the storage capacity of the user end are greatly reduced. With the support of the present invention, the user can theoretically have unlimited games.

Description of drawings

Fig. 1 is a schematic block diagram of an efficient and low-cost cloud game system provided by the present invention;

Fig. 2 is a block diagram of the user terminal runtime of the high-efficiency and low-cost cloud game system provided by the present invention;

3 is a block diagram of the data preprocessing subsystem of the general data of the high-efficiency and low-cost cloud game system provided by the present invention;

Fig. 4 is a block diagram of an intelligent service terminal system of an efficient and low-cost cloud game system provided by the present invention;

Fig. 5 is a data compression flow chart of the high-efficiency and low-cost cloud game system provided by the present invention.

Embodiments of the present invention

The technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Apparently, the described embodiments are part of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer" etc. The indicated orientation or positional relationship is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, or in a specific orientation. construction and operation, therefore, should not be construed as limiting the invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and should not be construed as indicating or implying relative importance. The present invention will be described in further detail below in conjunction with the accompanying drawings.

As shown in Figures 1-5, the high-efficiency and low-cost cloud game system provided by the present invention includes a client operating system, an intelligent data access prediction subsystem, an intelligent service subsystem, a data preprocessing subsystem for general data, general compression and Data compression subsystem combined with special compression, data distribution algorithm combined with prediction and instruction;

A data compression subsystem that combines general and special compression. There are different types of game data such as code, configuration, text, models, audio, video, textures, graphics, images, values, others, etc. For the code, configuration, text, model, value and other data that need to be accurately transmitted, according to its data characteristics and effectiveness requirements (compression, and decompression time consumption), select a specific general compression (lossless) algorithm (such as 7z, LZMA, zip etc.) and different compression parameters; and for audio, video, texture, graphics, images and other data, according to the effectiveness requirements and user service level, use appropriate lossless or lossy compression algorithms (such as image png, jpeg, webp, etc., audio aac, mp3, etc., video mpeg, mpeg4, H264, H265, VP8, VP9, etc., and our self-developed algorithm) and different bit rates or resolutions and other parameters to generate compressed data of different quality . After receiving the data, the client's runtime will decompress and transcode as needed to obtain available data;

The present invention and its implementations have been described above, and this description is not limiting. What is shown in the drawings is only one of the implementations of the present invention, and the actual structure is not limited thereto. All in all, if a person of ordinary skill in the art is inspired by it, without departing from the inventive concept of the present invention, without creatively designing a structure and an embodiment similar to the technical solution, it shall fall within the scope of protection of the present invention.

Claims

A high-efficiency and low-cost cloud game system, characterized in that: it includes a user-side operating system, an intelligent data access prediction subsystem, an intelligent service terminal system, a data preprocessing subsystem for general data, and a data combination of general compression and special compression A data distribution algorithm that combines compression subsystems, predictions, and instructions.
A high-efficiency and low-cost cloud gaming system according to claim 1, characterized in that: the client running system is responsible for starting the game when running, and maps the game's access to data into local cache and server data access , and transparent to the user; the runtime will continue to communicate with the server, report the user's current behavior and game context information, and the server will intelligently predict the data that the user can use next; or pull the data that is currently needed but not in the cache ;Accept the data returned by the server at runtime, and decompress, decrypt, and transcode according to the protocol to meet the running requirements of the game, and add the local cache; and manage the local cache according to the user information, remaining storage space, and server instructions according to the algorithm ; Collect user behavior and game behavior during runtime, and upload to the server for subsequent modeling.
A high-efficiency and low-cost cloud game system according to claim 2, characterized in that: the collection of user behavior and game behavior during runtime includes key presses, key press duration, dwell time of each interface, game response, game running context, and game data access information.
A high-efficiency and low-cost cloud game system according to claim 1, characterized in that: the intelligent data access prediction subsystem selects machine learning algorithms, and combines one or more of these algorithms to build a machine learning algorithm pool ;

Use different algorithms in the machine learning algorithm pool to process the collected user behavior and program behavior, and combine the types and characteristics of the data to establish a data access prediction model for different users and different data; and predict the data that users will need next;

For each data of each user, in the model evaluation stage, the top models with good performance are selected according to indicators such as stability under small samples, prediction success rate, and computational complexity.
A high-efficiency and low-cost cloud game system according to claim 1, characterized in that: the intelligent service terminal system comprehensively processes the user information, device information, network status, user behavior and game context transmitted from the client terminal , including login, authentication, authorization, selection of intelligent data access prediction model, selection of corresponding quality model and compression algorithm, sending the data required by the user to the user in advance or on demand, and encrypting the data according to the settings during the process.
A high-efficiency and low-cost cloud game system according to claim 1, characterized in that: the data preprocessing subsystem of the general data is based on the model and data characteristics established by the intelligent data access prediction subsystem, and the data preprocessing subsystem Divide the data into data blocks, and select different compression algorithms to compress the data blocks; comprehensively integrate the user's rating of the specific data usage, the nature and function of the data, and the impact of the intelligent data access prediction subsystem on the user experience. Data blocks are dynamically allocated to corresponding storage areas on the server, namely cold data area, general data area, hot data area and accelerated data storage area.
A high-efficiency and low-cost cloud game system according to claim 1, characterized in that: the data compression subsystem combining general compression and special compression; according to different types of game data, for codes that need to be accurately transmitted, Configuration, text, model, and value, according to their data characteristics and effectiveness requirements, select a specific general compression algorithm and different compression parameters for compression; for audio, video, texture, graphics, images, according to the effectiveness requirements and user services Level, use lossless or lossy compression algorithm and different parameters such as bit rate or resolution to generate compressed data of different quality; after receiving the data, the runtime of the client performs decompression and transcoding as needed to obtain usable data .
A high-efficiency and low-cost cloud game system according to claim 1, characterized in that: the data distribution algorithm combining prediction and instruction is based on user behavior and program behavior, as well as the model generated by the intelligent data access prediction subsystem. Download and pull data; the data distribution subsystem supports game developers to implant instructions according to the protocol. When the instruction execution condition is triggered, the data distribution subsystem will pull specific data according to the instruction; the combination of prediction and instruction can achieve data access. Same as accessing local data, there is no network delay, so there is no lag in the game process.