CN112511256A

CN112511256A - Robust synchronization method for dynamic delay optimization in online game

Info

Publication number: CN112511256A
Application number: CN202011326117.0A
Authority: CN
Inventors: 高小翎; 高宏松
Original assignee: Individual
Current assignee: Individual
Priority date: 2020-11-24
Filing date: 2020-11-24
Publication date: 2021-03-16

Abstract

The invention provides a steady synchronization method for dynamic delay optimization in online games, which aims at the problem that the online game state synchronization efficiency is lower due to the fact that network fluctuation is not considered by an invariant time bucket synchronization algorithm and game messages are processed in a fixed synchronization delay mode, and provides a dynamic synchronization delay time bucket algorithm; aiming at the problem that the game state is damaged and the game is crashed due to the fact that the late messages are directly discarded by the time bucket algorithm, a selective rollback strategy based on the perception intensity is provided, the influence degree of the late messages on the game state is reduced to the minimum, the consistency of game data can be guaranteed, and the game experience is obviously improved.

Description

Robust synchronization method for dynamic delay optimization in online game

Technical Field

The invention relates to a steady synchronization method of games, in particular to a steady synchronization method of dynamic delay optimization in online games, and belongs to the technical field of online game synchronization.

Background

The online game is one of the most popular online entertainment activities for the vast netizens, and the current online games with big fire in the world are undoubtedly the best testimony of the online game. The online game can help the game player break through the limitation of regions, and the game players in different regions and different countries are connected through one piece of application software in daily life. In recent years, online games are gradually getting more and more sophisticated, and are just important entertainment items remaining for people's life, but online games also face a serious technical challenge, and online games are based on internet technology for data interaction, but the network itself has delay, error and packet loss phenomena, which pose a great challenge to online game state synchronization. The synchronization effect in the game world is a decisive factor for the success of a game, so that the performance of a synchronization mechanism adopted by an online game is very important.

With the great improvement of the performance of the online game, the frequency and the quantity of data transmission between the players and the server in the game are greatly increased, but the condition of inconsistent states of the online game is caused by the inherent delay and packet loss of the internet depending on the online game. The inconsistency of the game state directly results in the influence on the interactivity, the entertainment and the fairness of the game. The method ensures that the game state of the online game is strictly synchronized among players, and is the first problem to be solved by the online game.

Regardless of whether the online game server employs the P2P architecture or the C/S architecture, communication among a plurality of player clients requires messages to be transmitted through the network, thereby causing a critical problem in that, due to the delay of network transmission, the messages are not consistent with the information on the screen of the client 1 when they are displayed on the interface after being transmitted to the client two after the client one generates a move operation. As an on-line game, it is most important to ensure that the difference in game state observed by each player through the screen is maintained within a range that is hardly visible to the naked eye. The generation of the online game determines that the online game must be supported by a network, and the online game also determines that the online game must solve the negative influence of the network delay, packet loss, bandwidth and other problems inherent in a computer network on the consistency of the game state.

The robust synchronization algorithm can well ensure the consistency of the game state, and is widely applied to games with low real-time requirements, particularly online games in a local area network under a hybrid P2P mode. In online gaming using the hybrid P2P model, each node updates the game state itself by processing game messages. The most widely used of the robust synchronization algorithms is the constant time bucket synchronization algorithm of the prior art.

The constant time bucket synchronization algorithm in the prior art divides a game process into periods with constant time length, and each period corresponds to one bucket. Determining a constant lag time according to the current network delay, and putting the messages into a corresponding bucket according to the value of the message generation time plus the lag time when the messages are generated locally or from other nodes, and when the simulation time reaches the end of the period, executing all messages belonging to the period to refresh the game state, and directly discarding the late messages. A schematic diagram of the basic principle of the time bucket robust synchronization method is shown in fig. 1.

The invariant time bucket algorithm aims to solve the problem that extra synchronization delay is introduced due to the influence of network delay on the synchronization algorithm, and meanwhile, the condition that a player node in Locksetp is blindly bitter and the like is avoided, so that the consistency of game states can be guaranteed to a certain extent. However, in practical applications, the delay of the network does not have an upper limit, so that it is difficult to determine a delay time that can ensure both game interactivity and game state consistency when network fluctuations are relatively large. The constant time bucket algorithm solves the problem of influence on the synchronization effect due to overhigh delay of part of players to a certain extent, the time bucket determines an upper limit of synchronization delay according to the delay of all players, the game process is divided into periods with constant time length, and each period corresponds to one bucket. When the simulation time progresses to the end of the period, all messages belonging to the period are executed to refresh the game state.

The method comprises the following steps of message processing of an invariant time bucket synchronization algorithm: step 1, setting a maximum tolerance value of synchronous delay according to the characteristics of a game; step 2, setting the period length of a barrel in the game process, namely updating the game state at intervals; step 3, starting game operationThen, the NTP protocol is used for carrying out clock synchronization on the nodes of the players participating in the game; step 4, for the player node, delivering the message to T according to the received message_iInto a corresponding bucket, wherein the message delivery is calculated in a bucket period manner of T_iTime of message generation + maximum tolerance value; step 5, the message generated by the node does not need to be transmitted through the network and does not generate network delay, but still needs to be forced to carry out synchronization delay and put into T_iFor barrel, T_iThe calculation formula of (4) is the same as that in the step (4); step 6, if the player node receives a message, but the message occurs as follows: if the current time-maximum tolerance value is larger than the timestamp carried by the message, the message is indicated to be delayed, and the node directly discards the message; 7, at the end of the period of each barrel, the player node executes the messages in the barrels according to the time stamp sequence and updates the game picture of the node; step 8, the situation that no message exists in the bucket due to the existence of network delay inevitable for the game is possible, and the game state is predicted by using a Dead Reckinning client prediction algorithm only by using data in the previous bucket; and 9, in the game process, the player nodes perform clock synchronization with other node timing, so that the clock error between the player nodes in the whole game process is ensured to be within an acceptable range. The advantages of the invariant time bucket synchronization algorithm are: the game state is ensured to be consistent, meanwhile, the game fluency is not obviously reduced due to the delay of the node height of one player, and the fluency can be ensured to a certain extent.

The disadvantages of the prior art invariant time bucket synchronization algorithm mainly include:

first, the constant time bucket synchronization algorithm of the prior art determines a synchronization delay upper limit according to the delay of the player, and the actual situation is that the network delay is not maintained at a constant level, but fluctuates frequently, when the network condition is good, the messages of the player nodes can reach other player nodes quickly, the excessively long synchronization delay can reduce the responsiveness of the game, and when the network condition is poor, the constant synchronization delay still has a large amount of data delay, and the messages exceeding the synchronization delay upper limit can be directly discarded, but it is not imaginable that the messages are directly discarded for many games, which may possibly destroy the game consistency and cause the collapse of the game.

Secondly, because the online game depends on the condition that the states of the online game are inconsistent due to inherent delay and packet loss of the Internet, the prior art cannot ensure that the states of the online game are strictly synchronous among players, the inconsistency of the states of the online game directly causes the interactivity, entertainment and fairness of the game to be influenced, a time bucket algorithm in the prior art directly discards late messages, the state of the game is damaged, and further the game is crashed, the influence degree of the late messages on the state of the game is large, the consistency of game data cannot be well ensured, and the online game experience is greatly influenced;

thirdly, because the network delay fluctuates frequently, the prior art adopts an invariant time bucket algorithm, and does not consider the network delay and adopts the invariant synchronization delay to process the game message in the whole synchronization process, so that the synchronization performance of the game is seriously reduced. The prior art time bucket algorithm discards late messages directly, which results in a game state being corrupted, the late messages having a large impact on the game state.

Fourthly, by adopting the prior art, the player has poor feeling of responsiveness to the game in the game process and has obvious pause feeling; the condition that the game data is inconsistent and the condition that the game information is inconsistent are easily perceived; the player can feel obvious picture jitter in the game process; due to the existence of network delay, the message is frequently delayed, and the synchronization performance of the online game is seriously influenced by the delayed message; the prior art can roll back the game state frequently, which seriously affects the experience of the player.

Disclosure of Invention

The invention provides a steady synchronization method for dynamic delay optimization in online games, and provides a dynamic synchronization delay time bucket algorithm aiming at the problem that the online game state synchronization efficiency is low because network fluctuation is not considered in an invariant time bucket synchronization algorithm and game messages are processed in a fixed synchronization delay mode. When the network condition is good, the synchronization delay is properly reduced to improve the game responsiveness, and when the condition is poor, the synchronization delay is increased to reduce the influence of message delay on the game consistency, and the steady synchronization method for dynamic delay optimization in the online game can better adapt to the fluctuation of the network, has strong flexibility on the game state synchronization and ensures the reliability of the game state; aiming at the problem that the game state is damaged and the game is crashed due to the fact that the late messages are directly discarded by the time bucket algorithm, a selective rollback strategy based on sensing intensity is provided, the influence degree of the late messages on the game state can be reduced to the minimum, the consistency of game data can be well guaranteed through the time bucket algorithm based on selective rollback optimization of intensity sensing, the influence on online game experience is small, and the game experience is obviously improved.

In order to achieve the technical effects, the technical scheme adopted by the invention is as follows:

a steady synchronization method for dynamic delay optimization in online games optimizes an invariant time bucket synchronization algorithm, the synchronization delay of messages is dynamically adjusted according to the current network condition, when the network condition is good, the synchronization delay is properly shortened, the game responsiveness is improved, when the network condition is poor, the synchronization delay is properly prolonged, the message late rate is reduced, and the sequential processing of online game messages is ensured; the invention provides a selective rollback strategy based on intensity perception, when the induction intensity of a player node pair generating a late message node is higher, the state rollback is selected, and the influence degree of the late message on the game state is reduced to the minimum;

the algorithm optimization based on intensity perception is as follows: when a game starts, firstly, a network time protocol is utilized to carry out clock synchronization among player nodes, the player nodes adopting a time bucket algorithm can put messages into corresponding buckets to be executed after receiving the messages, the time bucket synchronization algorithm for selectively rolling back late messages is optimized based on strength perception among the player nodes, the player nodes store game states after the execution of the bucket period messages in the buckets each time, and the states which are not needed any more are regularly cleaned, when the perception level of the local player nodes on the late messages is an understanding level, the late messages are very important for the player nodes, and at the moment, the game states are rolled back to the bucket states before the late messages; if the node of the local player perceives the grade as other three grades to the late message generating node, the late message is directly discarded;

the message processing steps after the optimization of the time bucket algorithm based on the strength perception selective rollback are as follows:

the first step is as follows: carrying out clock synchronization on each player node through a network time protocol when the game starts;

the second step is that: judging whether the received message is delayed, if not, turning to the third step, otherwise, putting the received message into a corresponding barrel according to the set synchronous delay, and turning to the fifth step;

the third step: calculating the perception intensity of the node pair generating the late message node according to the calculation formula of the intensity perception of the player node to other objects, obtaining the perception grade according to the perception intensity, determining whether to perform state rollback or not, directly abandoning the message if not, turning to the fourth step if the state rollback is selective, and turning to the fifth step if the state rollback is selective;

the fourth step: putting the late message into a corresponding barrel according to the timestamp of the late message, rolling back the game state to the state corresponding to the previous barrel of the barrel, and then quickly performing the game state to the current simulation time;

the fifth step: and judging whether the clock reaches the latest bucket period, if so, refreshing the game state by using the information of the bucket in the period, otherwise, returning to the first step and entering the next information processing period.

The steady synchronization method for dynamic delay optimization in online games further comprises the following steps of network state evaluation: the player node firstly carries out clock synchronization on the node through a network time protocol, and the node after time synchronization can process messages based on the same clock; when the player node processes the received message according to the current synchronous delay, the network state is evaluated by taking Ws as an evaluation period, and the synchronous delay of the next evaluation period is determined according to the delay level of data in the evaluation period; in the time bucket algorithm, each node writes a current time stamp before sending a data packet, so that the network delay can be obtained by the difference between the time stamp in the data packet and a local clock when the data packet is received;

the player node has a quantitative expression of network delay of the evaluation period:

where m represents the total number of data packets received by the player node in the evaluation period, De_kRepresenting the network delay of the kth data packet in the period, and u is the number of the node of the player;

the network delay quantization between the central server node and the node of the evaluation period is as follows:

where n represents the total number of player nodes,

a delay quantization value, De, representing the u-th nodal evaluation period_i-1The network delay evaluation value of the server in the last evaluation period is obtained;

the latest synchronization delay time calculation is as follows:

SyDe₂＝SyDe₁formula 3

Wherein SyDe_iRepresenting the synchronization delay of the i-th evaluation period.

The invention provides a steady synchronization method for dynamic delay optimization in online games, and further provides optimization of time bucket message processing of dynamic synchronization delay, wherein a network state is evaluated in a fixed period in a game process, the delay of the period is quantified in the evaluation period according to the delay of all data packets received in the period, and the synchronization delay of the online games is dynamically adjusted according to the evaluation result;

the message processing steps based on the dynamic synchronization delay are as follows:

the method comprises the following steps: all nodes participating in the game evaluate the network state according to the formula 1 in the evaluation period;

step two: after the evaluation period is finished, each game node sends the evaluation result of the period to a central server node, and the central server node evaluates the overall network environment among the whole players in the evaluation period according to the formula 2;

step three: the central server node calculates the synchronous delay of the player node processing message in the next evaluation period according to the formula 3, and informs all other nodes of the calculation result of the synchronous delay;

step four: all player nodes uniformly adopt the synchronous delay value to calculate the barrel period of the received message or the message which is locally generated by the player nodes to throw in the next period;

step five: the player nodes process the game messages during the evaluation period according to the invariant time bucket synchronization algorithm processing steps of the present invention.

Further, in the game, the strength perception of the game player to the surroundings mainly depends on the visual field range of the game role of the game player and the influence range of the game nodes, and the factors influencing the strength perception of the game role nodes are summarized as follows:

firstly, in the game world, the distance between characters has the most direct influence on the strength perception of a player, the closer the distance is, the higher the strength perception between the characters is, and the farther the distance is, the lower the strength perception between the characters is, and the strength perception of the player is subjected to three stages of perception, identification and understanding in the process that the player is close to other perceived player objects;

secondly, in the three-dimensional space, the field range of the player directly influences the intensity perception of the player on other game objects, the player has strong intensity perception on the role in the field range of the player, the intensity perception outside the field range is weak, and the factors influencing the field range of the player include the visual direction and the visual angle of the player;

thirdly, in the three-dimensional space, the strength perception of the player to other objects is also related to the size of the perceived object, and the strength perception of the player to the object which is larger than the player is higher;

fourthly, the movement direction of the object in the three-dimensional space can influence the strength perception of the player, and if the movement direction of the player and the movement direction of the perceived object enable the player and the perceived object to meet each other in the near future, the strength perception of the player to the object can be improved;

the above factors have a direct relationship to the player's perception of intensity, which is quantified according to the above factors.

The robust synchronization method for dynamic delay optimization in online games further comprises the following steps of defining an intensity perception mathematical model:

the perception entity: the online game can sense the roles of other entities, and the main attributes comprise entity type, entity position, visual angle, visual field direction, visual height, visual distance, action height, action distance and movement speed V (x, y, z);

sensing range: the range which can be perceived by the perception entity is also the range which is interested by the player in the game process;

influence scope: the range that the game entity can affect in the game;

the perception intensity is as follows: the player node senses the strength of other nodes;

perception grade: the player perceives the strength degree of other nodes, namely, the strength perception is graded and quantized, and the perception range is divided into: no perception, recognition, understanding four levels.

The method for the robust synchronization of the dynamic delay optimization in the online game further comprises an intensity perception quantization algorithm:

the sensing range of the player node is a sector area which can be observed by the player node, the sector area is a sector cylinder determined by the position, the visual height, the visual angle, the visual field direction and the visual distance of a player character, the influence range of the player node is a cylinder with the current position coordinate of the player character as the center of a circle, the action distance as the radius and the action height as the height, and the position coordinate of the player node is assumed to be Q (x)₀，y₀，z₀) Then, the sensing range of the player is a sector cylinder with a sector central angle determined by the viewing angle ViA with the size b and the viewing direction ViD (x, y, z) with the center of Q as the center of the circle, the viewing distance ViR as the radius, the viewing height ViH as the height, the player's influence range is a cylinder with Q as the center of the circle, the action distance EfH as the radius, and the action height EfR as the height.

Further, in the game world, the strength perception of the player to other players has great relevance with the perception range of the node of the player and the influence ranges of other nodes, and if two player nodes X and a player node Y in the game are assumed, the size of intersection of the perception range of the player node X and the influence range of the player node Y can be used as an index for evaluating the strength perception of the player X to the player Y; assuming that the intersection area of the perception range of player X and the influence range of player Y is S (SaR, IaR), let f be the coefficient of action of player Y on player X, and f is calculated by equation 4:

the intensity perception between the nodes of the player has a direct relationship with the distance between the two nodes, if the node Y of the player is in the perception range of the node B of the player, the closer the distance between the two nodes, the higher the intensity perception, and the two toursThe distance between player nodes can be used as the X coordinate of evaluation player node as Q (X)_X，y_X，z_X) For player node Y coordinate is Q (x)_Y，y_Y，z_Y) An index of intensity perception, recorded as h as a distance action factor, is calculated as formula 5:

in an actual game, there is also a case where the player node Y is not within the visible range of the player node X or there is no intersection between the sensing range of the player node X and the influence range of the player node Y, but according to the motion rules of the player node X and the player node Y, the player node X is likely to interact with the player node Y in the near future, the potential influence is also an index of the strength sensing of the player node Y by the player node X, w is a potential action factor, and the speed of the player node X is denoted as U (U is a potential action factor, U is a speed of the player node X, and the speed of the player node X is denoted as U (U is a speed of the player node X_Xx，u_Xy，u_Xz) The velocity of the player node Y is denoted as U (U)_Yx，u_Yy，u_Yz) Calculating equation 6:

from the above summary, the strength perception of the player node to other nodes is quantified by three indexes, namely, the action coefficient, the distance action factor, and the potential action factor, and the perceived strength PeI of the player node to other nodes is represented by formula 7:

the calculation formula of the perception intensity is the first formula in formula 7 when the perceived entity is located within the visual range of the perception entity, and the calculation formula of the intensity perception is the second formula in formula 7 when the perceived entity is located outside the visual range of the perception entity.

The steady synchronization method for dynamic delay optimization in the online game further comprises the following steps: according to the perception intensity degree of the perception node to the perceived node, the perception intensity is classified into four grades: an understanding level, a recognition level, a perception level, and an imperceptible level, wherein two constants p and q are set, the perception level is set to be the imperceptible level when the perception intensity PeI (X, Y) < 0, the perception level is set to be the imperceptible level when the perception intensity 0 < PeI (X, Y) ≦ p, the perception level is set to be the recognition level when the perception intensity p < PeI (X, Y) ≦ q, the perception level is set to be the understanding level when the perception intensity PeI (X, Y) > q, and p and q are preset according to the requirements of the game.

Compared with the prior art, the invention has the advantages and innovation points that:

firstly, the invention provides a steady synchronization method for dynamic delay optimization in online games, and provides a dynamic synchronization delay time bucket algorithm aiming at the problem that the online game state synchronization efficiency is low because the network fluctuation is not considered by the constant time bucket synchronization algorithm and game messages are processed by fixed synchronization delay. The method for the steady synchronization of the dynamic delay optimization in the online game can better adapt to the fluctuation of the network, has strong flexibility on the synchronization of the game state, ensures the reliability of the game state, and has high utilization value in practical application;

secondly, the robust synchronization method for dynamic delay optimization in online games provided by the invention provides a selective rollback strategy based on the perception intensity aiming at the problem that the game state is damaged and the game is crashed because the late messages are directly discarded by the time bucket algorithm, so that the influence degree of the late messages on the game state can be reduced to the minimum, the time bucket algorithm for selective rollback optimization based on the perception intensity can better ensure the consistency of game data, has small influence on online game experience, has outstanding advantages and has obvious innovativeness and can ensure the consistency of the game data;

thirdly, the steady synchronization method for dynamic delay optimization in online games provided by the invention optimizes the steady synchronization algorithm of the invariant time bucket, the synchronization delay of the messages can be dynamically adjusted according to the current network condition, the synchronization delay is properly shortened when the network condition is good, the game responsiveness is improved, the synchronization delay is properly prolonged when the network condition is poor, the late rate of the messages is reduced, and further the sequential processing of the online game messages is ensured. The time bucket algorithm for the prior art discards late messages directly, which can result in the game state being corrupted, which in turn leads to the game crashing. The invention provides a selective rollback strategy based on intensity perception, and when the induction intensity of a node pair generating a late message is higher, a player selects a state to rollback. Therefore, the influence degree of the late message on the game state can be reduced to the minimum, the interactivity, the entertainment and the fairness of the game are greatly improved, the state of the online game among players is ensured to be strictly synchronous, and the game experience of the players is obviously improved;

fourthly, according to investigation of participating experimenters, the adoption of the steady synchronization method for dynamic delay optimization in the online game provided by the invention does not feel obvious pause, the adoption of the experimenters based on the dynamic period time bucket algorithm shows that the whole game is good in fluency, although the game has a few tolerable pauses, the fluency is much better than that of the adoption of the invariable time bucket algorithm, the adoption of the method disclosed by the invention does not detect the condition that the game data is inconsistent and the game information is inconsistent, the game process does not feel obvious picture jitter, and the whole game experience is greatly improved.

Drawings

Fig. 1 is a basic principle illustration of a time bucket robust synchronization method.

Fig. 2 is a schematic diagram of a time bucket robust synchronization method based on dynamic delay optimization.

Fig. 3 is a diagram of message processing steps after optimization of a time-bucket algorithm based on intensity-aware selective rollback.

Detailed Description

The technical solution of the robust synchronization method for dynamic delay optimization in online games provided by the present invention is further described below with reference to the accompanying drawings, so that those skilled in the art can better understand the present invention and can implement the present invention.

The invention optimizes the steady synchronization algorithm of the invariable time bucket, because the network delay fluctuates frequently, if the invariable time bucket algorithm is adopted, the invariable synchronization delay is adopted to process the game message in the whole synchronization process without considering the network delay, so the synchronization performance of the game is seriously reduced, for example, when the network condition is good, the game message can be transmitted to other nodes in time, and the nodes can not be processed in time after receiving the message, thus falling into meaningless blind waiting. After the constant time bucket synchronization algorithm is optimized, the synchronization delay of the message can be dynamically adjusted according to the current network condition, when the network condition is good, the synchronization delay is properly shortened, the game responsiveness is improved, and when the network condition is poor, the synchronization delay is properly prolonged, the late rate of the message is reduced, and further the sequential processing of the online game message is ensured. The prior art time bucket algorithm discards late messages directly, which can result in a game state corruption and thus a game crash. The invention provides a selective rollback strategy based on intensity perception, when the induction intensity of a node pair generating a late message is higher, the state rollback is selected, and therefore, the influence degree of the late message on the game state can be reduced to the minimum.

Optimization of time bucket algorithm based on dynamic synchronization delay

The constant time bucket algorithm synchronizes the game state with constant synchronization delay, and the algorithm has no flexibility. Network delay is affected by network congestion and problems of the link itself, so that the network delay fluctuates frequently in a practical environment. If the invariant time bucket does not consider dynamically adjusting the synchronization delay according to the current network state, time waste and unnecessary synchronization waiting are inevitably caused, and the performance of the online game is reduced. The method properly reduces the synchronization delay to improve the game responsiveness under the condition of good network conditions, and properly prolongs the synchronization delay to ensure the game state consistency under the condition of poor network conditions, so that the game state can be timely synchronized well in the network, and certain responsiveness can be sacrificed to ensure the online game state consistency when the network conditions are poor.

For online gaming, the network delay is below 150ms, because of the player's visual delay, the player does not have a noticeable feeling of pause. Synchronization delay initial value consideration set to SyDe₁The period length is set to be W100 ms for 150ms to ensure the game experience of the player. In order to dynamically adjust the synchronization delay according to the network state, the network state needs to be evaluated.

Network state evaluation

The player nodes first clock-synchronize the nodes via the network time protocol, and the time-synchronized nodes can process messages based on the same clock. When the player node processes the received message according to the current synchronous delay, the network state is evaluated by taking Ws as an evaluation period, and the synchronous delay of the next evaluation period is determined according to the delay level of data in the evaluation period. The network delay is the time that it takes for data to be generated at one player node until another node receives the data packet, and in the time bucket algorithm, each node writes the current time stamp before sending the data packet, so the network delay can be obtained by the difference between the time stamp in the data packet and the local clock at the time of receiving the data packet.

where m represents the total number of data packets received by the player node in the evaluation period, De_kRepresenting the network delay of the kth packet in the cycle, u is the number of the player node.

where n represents the total number of player nodes,

a delay quantization value, De, representing the u-th nodal evaluation period_i-1The network delay evaluation value of the server in the last evaluation period is obtained.

The latest synchronization delay time calculation is as follows:

syDe₂＝syDe₁formula 3

(II) dynamically synchronizing delayed time bucket message processing

The invention proposes optimization of time bucket message processing of dynamic synchronization delay, and evaluates the network state in a fixed period in the game process. And in the evaluation period, the delay of the period is quantified based on the delay of all the data packets received in the period, and the synchronous delay of the online game is dynamically adjusted according to the evaluation result.

step two: after the evaluation period is finished, each game node sends the evaluation result of the period to the central server node, and the central server node evaluates the overall network environment among the whole players in the evaluation period according to the formula 2.

The time bucket method based on dynamic delay optimization is shown in fig. 2, and the time bucket algorithm using dynamic synchronization delay can properly adjust the processing waiting time according to the network condition.

Time bucket algorithm optimization based on intensity perception selective rollback

The prior art time bucket algorithm would directly discard late game messages, however, for online games employing non-C/S architectures, directly discarding late messages would be unacceptable and may even cause a game crash. If the late message with higher influence degree on the node state of the player is subjected to state rollback, the influence of the game late message on the game state can be greatly reduced. The game roles in the online game have own perception range and influence range, and according to different senses of the nodes of the game player on the intensities of other game nodes, the late messages are filtered and selectively rolled back, so that the consistency of the online game state is greatly improved by the optimization strategy.

(ii) quantized intensity perception

Any operation of a player node in an online game can only affect players within a certain range near the game node, and the player node in the online game is only interested in a certain visual range away from the player node, and the player node can only sense the existence of the nodes within the interest range, so that the intensity of the player sensing other entities is defined as intensity sensing.

1. Factors influencing the perception of player intensity

The strength perception of the game player to the surroundings in the game mainly depends on the visual field range of the game role of the game player and the influence range of the game node, and the factors influencing the strength perception of the game role node are summarized as follows:

(1) in the game world, the distance between characters has the most direct influence on the strength perception of a player, and as in real life, the closer the distance between people, the higher the strength perception between people, and conversely, the farther the distance between people, the lower the strength perception between people. The strength perception of the player to other player characters in the game follows the same rule, and the strength perception of the player in the process of approaching to other perceived player objects goes through three stages of perception, identification and understanding.

(2) In the three-dimensional space, the field of view range of the player also directly influences the intensity perception of the player on other game objects, the player has strong intensity perception on the role in the field of view range of the player, the intensity perception outside the field of view range is weak, and factors influencing the field of view range of the player include the visual direction and the visual angle of the player.

(3) In the three-dimensional space, the strength perception of other objects by the player is also related to the size of the perceived object, and the strength perception of the object by the player is higher than that of the object by the player.

(4) The direction of movement of the object in three-dimensional space also has an effect on the player's perception of intensity, which is also enhanced if the direction of movement of the player and the direction of movement of the perceived object are such that they will meet in the near future.

The above factors have a direct relationship to the player's perception of intensity, so the player's perception of intensity can be quantified in terms of the above factors.

2. Intensity-aware mathematical model definition

(1) The perception entity: the online game can sense the roles of other entities, and the main attributes comprise entity type, entity position, visual angle, visual field direction, visual height, visual distance, action height, action distance and movement speed V (x, y, z), wherein the object type, the visual angle, the visual height, the visual distance, the action height and the action distance are preset in game design.

(2) Sensing range: the range that the perceiving entity can perceive is also the range that the player is interested in during the game.

(3) Influence scope: the extent to which a gaming entity can affect in a game.

(4) The perception intensity is as follows: the player node senses the strength of the other nodes.

(5) Perception grade: the player perceives the strength degree of other nodes, namely, the strength perception is graded and quantized, and the perception range is divided into: no perception, recognition, understanding four levels.

3. Intensity-aware quantization algorithm

The perceived range of the player nodes is a sector area viewable by the player nodes, the sector area being a sector cylinder determined by the position, viewing height, viewing angle, viewing direction and viewing distance of the player character. The influence range of the player node is a cylinder with the current position coordinate of the player role as the center of a circle, the acting distance as the radius and the acting height as the height. Suppose the position coordinates of the player node are Q (x)₀，y₀，z₀) Then, the player's perception range is a sector cylinder with a sector central angle determined by the viewing angle ViA and the viewing direction ViD (x, y, z) with the viewing distance ViR as a radius, the viewing height ViH as a height, and the size b, with Q as a center of the circle. The player's range of influence is a cylinder centered at Q, with action distance EfH as the radius and action height EfR as the height.

In the game world, the strength perception of the player to other players is greatly related to the perception range of the node of the player and the influence range of other nodes. Assuming that two player nodes X and Y are in the game, the size of the intersection of the range of influence of the player node X and the range of influence of the player node Y can be used as an index for evaluating the strength perception of the player X on the player Y. Assuming that the intersection area of the perception range of player X and the influence range of player Y is S (SaR, IaR), let f be the coefficient of action of player Y on player X, and f is calculated by equation 4:

the intensity perception also has a direct relationship with the distance between two nodes between the player nodes, and if the player node Y is within the perception range of the player node B, the closer the two nodes are, the higher the intensity perception is. The distance between two player nodes can be used as the X coordinate of the evaluation player node as Q (X)_X，y_X，z_X) For player node Y coordinate is Q (x)_Y，y_Y，z_Y) An indicator of intensity perception. Let h be the distance action factor, and the calculation formula of h is 5:

in an actual game, there is also a case where the player node Y is not within the visible range of the player node X or there is no intersection between the sensing range of the player node X and the influence range of the player node Y, but the player node X is likely to interact with the player node Y in the near future according to the movement laws of the player node X and the player node Y. This potential effect is also an indicator of the intensity perception of the player node X on the player node Y, denoted as w as the potential contributing factor, and the velocity of the player node X is denoted as U (U)_Xx，u_Xy，u_Xz) The velocity of the player node Y is denoted as U (U)_Yx，u_Yy，u_Yz) Calculating equation 6:

from the above summary, the strength perception of the player node to other nodes is quantified by three indexes, namely, the action coefficient, the distance action factor and the potential action factor. The perceived strength PeI of the player node to other nodes is represented by equation 7:

4. Classifying perception levels

According to the perception intensity degree of the perception node to the perceived node, the perception intensity is classified into four grades: understanding level, recognition level, perception level, non-perception level. Two constants p, q are set, the perception level is set to a non-perception level when the perception intensity PeI (X, Y) < 0, to a perception level when the perception intensity 0 ≦ PeI (X, Y) ≦ p, to a recognition level when the perception intensity p < PeI (X, Y) ≦ q, to an understanding level when the perception intensity PeI (X, Y) > q, p, q being set in advance according to the requirements of the game.

(II) Algorithm optimization based on intensity perception

When the game starts, the player nodes firstly use the network time protocol to carry out clock synchronization, and the player nodes adopting the time bucket algorithm can put the messages into the corresponding buckets to be executed after receiving the messages. Because the prior art time bucket directly discards received late messages to cause the game state to be damaged and further cause the phenomenon of game collapse, the invention provides time bucket synchronization algorithm optimization for selectively rolling back the late messages based on strength perception among player nodes, and the player nodes store the game state after executing the periodic messages of the bucket in the bucket each time and regularly clear the state which is not needed any more. When the local player node perceives the level as an understanding level to the late message generating node, the late message is important to the player node, and the game state is rolled back to the bucket state before the late message. If the local player node perceives the levels as the other three levels to the late message generating node, the late message is directly discarded.

A flowchart of the message processing steps after optimization of the time-bucket algorithm based on intensity-aware selective rollback is shown in fig. 3.

Third, time bucket synchronization algorithm optimization experiment and result analysis

(I) Experimental Scenario

A game scene is established in the Unity3D, the scene has three roles, 18 clients are divided into two groups, each group has 9 clients, the first group adopts an invariant time bucket algorithm, and the second group adopts a time bucket algorithm based on dynamic synchronization delay and based on intensity perception selective rollback to process the received game time. In the process of each group of the game, each player controls a game role to move in a map with the size of 180 x 180, so that the three roles are always in the same visual field range, and the synchronous effect can be directly observed. Secondly, the map is divided into 90-by-90 grids, each character can only move one grid, a player can control the movement of the character through W, A, S, D on the keyboard, and the character can attack other two characters in the moving process, but the attack frequency is limited within 9 times of a minute. In order to ensure the reliability of the experimental result, two groups of players play games in the same network environment at the same time.

(II) method Performance evaluation index

The purpose of this experiment is to verify whether the optimized time bucket synchronization algorithm can improve the performance of the online game on the premise of ensuring the consistency of the game state. Objective and accurate evaluation of the performance of the game is required. The present invention proposes the following evaluation indexes.

1. Subjective evaluation index

Firstly, the participating experimenters feel the responsiveness of the game in the game process, and whether the experimenters have obvious pause feeling; whether participating experimenters perceive the condition that the game data are inconsistent and the condition that the game information is inconsistent in the game process or not; and thirdly, whether the participating experimenters feel obvious picture jitter in the game process and whether the jitter occurrence frequency is acceptable.

2. Objective evaluation index

Firstly, the message waiting processing time is the time for dynamically changing the message processing time based on the time bucket algorithm of dynamic delay; the late arrival of the message, which is caused by the existence of network delay, often occurs and seriously affects the synchronization performance of the online game; and thirdly, the rolling times of the game state are calculated, the game state is rolled back based on the time bucket algorithm of the intensity perception selective rolling back, and the experience of a player is influenced by the frequency degree of the rolling back.

(III) analysis of the results of the experiment

The experiment was performed in a local area network, with 18 clients divided into two groups. The first group employs the prior art invariant time bucket synchronization algorithm and the second group employs the time bucket synchronization algorithm optimized by the present invention. And performing objective evaluation on each algorithm through subjective feeling of the experiment participants and objective indexes. In order to ensure the tightness of the experiment, all indexes of the two groups of clients are the same except for the adopted message processing algorithm. The game experiment result is ensured to have reliability, the game playing time is ensured to be more than 60 minutes, and the message interaction amount is ensured to be more than 18000 times.

Experiment one:

subject: the invariant time bucket synchronization algorithm and the time bucket algorithm after dynamic synchronization delay optimization are adopted.

Purpose of the experiment: and comparing the message processing response time of the two algorithms with the late number of the messages so as to evaluate the performance optimization effect of the time bucket algorithm based on dynamic delay optimization and the traditional algorithm.

Experimental parameters: the Clumsy software is used to simulate different network states so that the network delay can vary from 100ms to 210 ms.

The experimental results are as follows: according to the conclusion obtained by the investigation of the participating experimenters, no obvious pause is sensed by the two groups of experimenters, and the experimenters based on the dynamic period time bucket algorithm show that the whole game has good fluency, but has several sustainable pauses. Experimenters who employ the invariant time bucket algorithm show that the fluency of the game is marginally acceptable, but need further improvement.

Compared with the prior art algorithm, the optimized algorithm has almost the same message late numbers when the network condition is good, but when the network delay is more than 140ms, the message late numbers of the optimized algorithm of the invention have obvious reduction, while the message late numbers of the prior art algorithm rise linearly, thereby drawing a conclusion that: the time bucket algorithm based on dynamic synchronization delay optimization effectively reduces the late rate of the game. At network delays less than 130ms, the optimized algorithm processes messages at a significantly higher rate than the prior art constant time bucket algorithm, and the optimized algorithm has 1/2 messages that can be executed within 125 ms.

The following conclusions were drawn from the subjective feeling and analysis of the persons involved in the experiment: firstly, the time bucket algorithm based on dynamic synchronization delay optimization can ensure better game consistency; and secondly, the time bucket algorithm based on dynamic synchronization delay optimization effectively reduces the late rate of the messages and improves the responsiveness of the online game.

Experiment two:

subject: the constant time bucket synchronization algorithm of the prior art and the time bucket algorithm of the selective rollback based on the intensity perception of the invention.

Purpose of the experiment: the accuracy of data and states in the game, the number of messages delayed and the rollback times of the game states of the two algorithms are compared, so that the performance optimization effect of the time bucket algorithm based on the intensity perception selective rollback and the algorithm in the prior art is evaluated.

Experimental parameters: the Clumsy software is used for simulating different network states, so that the network delay is changed within 100ms to 210 ms.

The experimental results are as follows: from the investigation of the participating experimenters, it was concluded that neither group of experimenters felt significant katton, and that the experimenters using selective rollback based on intensity perception indicated that a few minor game frame jitters occurred, but within an acceptable range. Experimenters adopting the invariant time bucket algorithm show that the condition that the game result is inconsistent with the expectation occurs in the game process, and the game experience is influenced to a certain extent.

It follows that instead of rolling back the game state once every occurrence of a message late, the late message is not discarded directly, but rather a selective roll back is performed. The rolling back times of the game state are 1/3 which are late, and the investigation of the participating experimenters shows that the experimenters adopting the invariant time bucket algorithm have the condition that the game results are inconsistent with the expectation for many times, but the experimenters adopting the optimized algorithm of the invention have no reaction, and the experimenters adopting the optimized algorithm react to the condition that the game pictures are jittered, but the times are few and are within the acceptable range.

The following conclusions were drawn from the above analysis: the time bucket algorithm based on the intensity perception and selective rollback optimization can better ensure the consistency of game data and has small influence on online game experience.

The optimization summary of the time bucket synchronization algorithm of the invention is as follows: aiming at the problem that the constant time bucket synchronization algorithm does not consider network fluctuation and adopts constant synchronization delay to process game information to cause low game state synchronization efficiency, the following optimization strategy is provided, and if the network condition is good, the synchronization delay is properly reduced to improve the game responsiveness; when the network condition is poor, the synchronization delay is properly prolonged to ensure the consistency with the game state, so that the game state can be synchronized in time when the network is good, and certain responsiveness can be sacrificed to ensure the consistency of the game state when the network condition is poor.

Claims

1. A steady synchronization method for dynamic delay optimization in online games is characterized in that a constant time bucket synchronization algorithm is optimized, the synchronization delay of messages is dynamically adjusted according to the current network condition, when the network condition is good, the synchronization delay is properly shortened, the game responsiveness is improved, when the network condition is poor, the synchronization delay is properly prolonged, the message late rate is reduced, and the sequential processing of online game messages is ensured; the invention provides a selective rollback strategy based on intensity perception, when the induction intensity of a player node pair generating a late message node is higher, the state rollback is selected, and the influence degree of the late message on the game state is reduced to the minimum;

2. The method of claim 1, wherein the network state evaluation: the player node firstly carries out clock synchronization on the node through a network time protocol, and the node after time synchronization can process messages based on the same clock; when the player node processes the received message according to the current synchronous delay, the network state is evaluated by taking Ws as an evaluation period, and the synchronous delay of the next evaluation period is determined according to the delay level of data in the evaluation period; in the time bucket algorithm, each node writes a current time stamp before sending a data packet, so that the network delay can be obtained by the difference between the time stamp in the data packet and a local clock when the data packet is received;

where n represents the total number of player nodes,

the latest synchronization delay time calculation is as follows:

SyDe₂＝SyDe₁formula 3

3. The robust synchronization method for dynamic delay optimization in online games according to claim 2, characterized in that the invention proposes optimization of time bucket message processing for dynamic synchronization delay, evaluates the network state at a fixed period during the game, quantifies the delay of the period in the evaluation period based on the delay of all data packets received in the period, and dynamically adjusts the synchronization delay of the online game according to the evaluation result;

4. The robust synchronization method for dynamic delay optimization in online games as claimed in claim 1, wherein the influence on the intensity perception of the surrounding objects by the player in the game mainly depends on the field of view range of the player's game character and the influence range of the game node, and the factors influencing the intensity perception of the game character node are summarized as follows:

5. The method of claim 1, wherein the intensity-aware mathematical model defines:

influence scope: the range that the game entity can affect in the game;

6. The method of claim 5, wherein the intensity-aware quantization algorithm:

7. The robust synchronization method for dynamic delay optimization in online games as claimed in claim 6, wherein in the game world, the strength perception of the player to other players has a great correlation with the perception range of the node of the player and the influence ranges of other nodes, and assuming that two player nodes X and player node Y in the game, the size of the intersection of the perception range of the player node X and the influence range of the player node Y can be used as an index for evaluating the strength perception of the player X to the player Y; assuming that the intersection area of the perception range of player X and the influence range of player Y is S (SaR, IaR), let f be the coefficient of action of player Y on player X, and f is calculated by equation 4:

the intensity perception between the player nodes is further compared withThe distance between two nodes has direct relation, if the player node Y is in the perception range of the player node B, the closer the distance between the two nodes is, the higher the perception intensity is, the distance between the two player nodes can be used as the coordinate Q (X) of the evaluation player node X_X，y_X，z_X) For player node Y coordinate is Q (x)_Y，y_Y，z_Y) An index of intensity perception, recorded as h as a distance action factor, is calculated as formula 5:

8. The method of claim 1, wherein the perceptual ranking is divided into: according to the perception intensity degree of the perception node to the perceived node, the perception intensity is classified into four grades: an understanding level, a recognition level, a perception level, and an imperceptible level, wherein two constants p and q are set, the perception level is set to be the imperceptible level when the perception intensity PeI (X, Y) < 0, the perception level is set to be the imperceptible level when the perception intensity 0 < PeI (X, Y) ≦ p, the perception level is set to be the recognition level when the perception intensity p < PeI (X, Y) ≦ q, the perception level is set to be the understanding level when the perception intensity PeI (X, Y) > q, and p and q are preset according to the requirements of the game.