Mobile game cross-platform real-time interaction optimization method and device based on cloud data synchronization

Abstract

The application discloses a hand game cross-platform real-time interaction optimization method and device based on cloud data synchronization, and belongs to the technical field of electronic digital data processing, and comprises the following steps: collecting parallel operation instructions from each client and a server, and collecting to form a decision event set of a current logical frame; generating a candidate decision event pair and calculating a decision overlap index; when the decision overlap index is greater than a decision overlap index threshold value, a decision dependence directed graph is constructed; a comprehensive decision sequence is formed; a sequence stability evaluation parameter set is calculated, and whether there is an unstable risk is determined; when there is an unstable risk, sequence compensation adjustment is performed, otherwise, decision processing is performed according to the comprehensive decision sequence; if a cloud logical node migration event is detected in the sequence compensation process, the phase correction adjustment of a new node logical clock is performed, and the adjustment of a logical clock advancing rate is performed, and the problem of the decision partial order caused by the non-commutativity of state update under the parallel decision of multiple logical threads in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, and in particular to a method and apparatus for optimizing cross-platform real-time interaction in mobile games based on cloud data synchronization. Background Technology

[0002] Non-commutativity means that within the same logical frame, if the execution order of two adjudication events is different, the final game state will be different, and the order cannot be arbitrarily changed.

[0003] In cloud-based cross-platform real-time mobile games, multiple adjudication events from various clients and internal server modules are typically aggregated within the same logical frame. Due to the high coupling between adjudication events in terms of time intervals, affected objects, and state impacts, the execution order of different adjudication events directly affects the server's state evolution path. When there are non-commutative relationships between adjudication events, changes in the adjudication order during multi-node scheduling, multi-threaded execution, or cross-platform merging will result in a non-unique final game state within the same logical frame, leading to issues such as cross-platform state inconsistencies, abnormal skill judgments, and shifts in combat results. Existing technologies mostly rely on timestamp sorting or queue-based execution, failing to identify the natural causal constraints between adjudication events and unable to proactively detect and compensate for the risk of unstable order.

[0004] For example, the data synchronization method, device, and system disclosed in Chinese invention patent CN117453807B include: for each application supporting personal cloud on multiple terminal devices, a cloud device maintains multiple sets of transaction logs, each set of transaction logs recording a set of transaction IDs in chronological order. When a terminal device initiates synchronization, the terminal device first reports the changed data and / or the terminal device's previous transaction ID to the cloud device. If the previous transaction ID differs from the latest transaction ID in the transaction log, the cloud device or the terminal device resolves the conflict. Afterwards, the cloud device distributes the changed data between the latest transaction ID and the terminal device's previous transaction ID.

[0005] The above-mentioned technology has at least the following technical problems:

[0006] In existing technologies, due to the complex dependencies between different adjudication operations in terms of time, space, rules, and objects of action, and the fact that existing cloud adjudication devices usually only execute adjudication based on the order of operation arrival or thread scheduling, without modeling and constraining the non-commutative relationships between adjudication operations, the non-commutative nature of state updates under parallel adjudication by multiple logical threads leads to the problem of partial order of judgment caused by the non-commutative nature of state updates. Summary of the Invention

[0007] To address the issue of biased decision order caused by the non-commutativity of state updates under parallel adjudication of multiple logical threads in existing technologies, this invention provides a method and apparatus for optimizing cross-platform real-time interaction in mobile games based on cloud data synchronization. The technical solution is as follows:

[0008] On the one hand, a method for optimizing cross-platform real-time interaction in mobile games based on cloud data synchronization is provided. This method includes: the cloud server collecting parallel operation instructions from various clients and servers in real time within a logical frame period, converting them into corresponding adjudication event feature vectors to form an adjudication event set; indexing and classifying the adjudication event set to generate candidate adjudication event pairs, and performing overlap determination to form corresponding adjudication overlap indices; when the adjudication overlap indices exceed a preset threshold, constructing an adjudication dependency directed graph for the candidate adjudication event pair; otherwise, not constructing an adjudication dependency directed graph; when the adjudication dependency directed graph does not have cyclic dependencies, performing topological sorting on the adjudication events based on the adjudication dependency directed graph to generate a basic adjudication execution sequence; and when the adjudication... When a directed graph with cyclic dependencies exists, a strongly connected component decomposition is performed on the directed graph to generate a minimum conflict decision sequence, thus obtaining a comprehensive decision sequence. The comprehensive decision sequence includes the basic decision execution sequence and the minimum conflict decision sequence. The set of order stability evaluation parameters for the comprehensive decision sequence is obtained. If any order stability evaluation parameter is greater than or equal to the corresponding order stability evaluation threshold, it is determined that the current logical frame decision order has an unstable risk, and order compensation adjustment is performed. Otherwise, the comprehensive decision sequence remains unchanged, and decision processing is performed according to the comprehensive decision sequence. During the order compensation process, when a cloud logical node migration event is detected, the phase correction adjustment of the new node's logical clock and the logical clock advance rate are adjusted.

[0009] On the other hand, a mobile game cross-platform real-time interaction optimization device based on cloud data synchronization is provided. This device applies a method for optimizing cross-platform real-time interaction in mobile games based on cloud data synchronization. The device includes: a parallel analysis module, an adjudication overlap determination module, an adjudication sequence generation module, an adjudication processing module, and a compensation correction module. The parallel analysis module is used by the cloud server to collect parallel operation instructions from various clients and servers in real time within a logical frame period, and convert them into corresponding adjudication event feature vectors to form an adjudication event set. The adjudication overlap determination module is used to index and classify the adjudication event set, generate candidate adjudication event pairs, and perform overlap determination to form a corresponding adjudication overlap index. When the adjudication overlap index is greater than a preset adjudication overlap index threshold, an adjudication dependency directed graph of the candidate adjudication event pair is constructed; otherwise, the adjudication dependency directed graph is not constructed. The adjudication sequence generation module is used to process adjudication sequences when the adjudication dependency directed graph is not... When cyclic dependencies exist, a topological sorting of the adjudication events is performed based on the adjudication dependency directed graph to generate a basic adjudication execution sequence. When cyclic dependencies exist in the adjudication dependency directed graph, strongly connected component decomposition is performed on the adjudication dependency directed graph to generate a minimum conflict adjudication sequence, thus obtaining a comprehensive adjudication sequence. The comprehensive adjudication sequence includes the basic adjudication execution sequence and the minimum conflict adjudication sequence. The adjudication processing module is used to obtain the order stability evaluation parameter set of the comprehensive adjudication sequence. When any order stability evaluation parameter is greater than or equal to the corresponding order stability evaluation threshold, it is determined that the adjudication order of the current logical frame has an unstable risk, and order compensation adjustment is performed. Otherwise, the comprehensive adjudication sequence remains unchanged and adjudication processing is performed according to the comprehensive adjudication sequence. The compensation correction module is used to perform phase correction adjustment on the logical clock of the new node and adjust the logical clock advance rate when a cloud logical node migration event is detected during the execution of order compensation.

[0010] The beneficial effects of the technical solutions provided by the embodiments of the present invention include at least the following:

[0011] 1. The mobile game cross-platform real-time interaction optimization method based on cloud data synchronization provided by the present invention forms a set of adjudication events and constructs a feature vector of adjudication events, thereby performing unified adjudication modeling on parallel operation instructions within the same logical frame period, and thus realizing the correlation analysis of the temporal relationship and conflict relationship of adjudication events, effectively solving the problem of partial order of judgment caused by the non-commutativity of state updates under parallel adjudication of multiple logical threads in the prior art.

[0012] 2. This invention calculates the adjudication overlap index and constructs an adjudication dependency directed graph when the adjudication overlap index is greater than the adjudication overlap index threshold. This establishes the sequential constraint relationship between adjudication events in the order of execution. When there is a cyclic dependency, it performs strongly connected component decomposition to generate the minimum conflict adjudication sequence. This enables the sortable processing of adjudication order under complex concurrent adjudication relationships and effectively solves the problems of topological sorting failure and adjudication blocking caused by cyclic adjudication dependency.

[0013] 3. This invention forms a set of parameters for evaluating order stability based on partial order stability index, order determinism index, and dependency chain depth index. It also dynamically adjusts the phase parameter and decision delay window parameter for decision events on the critical path, thereby reducing the fluctuation amplitude of decision order and restoring the stability of decision sequence. This enables long-term stable output of decision order in cross-platform high-concurrency interaction scenarios, effectively solving the problem of state rollback and interaction disorder caused by decision order jitter.

[0014] 4. This invention obtains the logical time phase offset and performs phase acceleration correction or phase deceleration correction adjustment. At the same time, it adjusts the logical clock advance rate according to the phase offset and restores normal adjudication scheduling when the remaining phase offset is within a preset safe range. This achieves smooth convergence between the logical time of the new node and the reference logical time of the adjudication domain, thereby ensuring the consistency of the logical clock and the continuous and stable adjudication scheduling cycle. It effectively solves the problem of adjudication misordering and time inconsistency caused by logical clock drift during the migration and access of new nodes. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A macro-flowchart of the mobile game cross-platform real-time interaction optimization method based on cloud data synchronization provided in the embodiments of this application;

[0017] Figure 2 The overall adjudication scheduling and sequence stability control flowchart of the mobile game cross-platform real-time interaction optimization method based on cloud data synchronization provided in the embodiments of this application;

[0018] Figure 3 Flowchart of the loop decision dependency processing and minimum conflict decision sequence generation of the mobile game cross-platform real-time interaction optimization method based on cloud data synchronization provided in the embodiments of this application;

[0019] Figure 4This is a structural framework diagram of a mobile game cross-platform real-time interactive optimization device based on cloud data synchronization, provided in an embodiment of this application. Detailed Implementation

[0020] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0021] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0022] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent.

[0023] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0024] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0025] like Figure 1 As shown, Figure 1This is a macro-flowchart of a mobile game cross-platform real-time interaction optimization method based on cloud data synchronization provided in this application embodiment. The method includes the following steps: The cloud server collects parallel operation instructions from each client and server in real time within a logical frame period, and converts them into corresponding adjudication event feature vectors to form an adjudication event set; the adjudication event set is indexed and classified to generate candidate adjudication event pairs. Specifically, the indexing and classification of the adjudication event set refers to constructing a multi-dimensional index key based on the identifier parameter set, establishing a time index dimension with the server frame number and intra-frame phase number as the time index dimension, a spatial index dimension with the bounding volume of the action area as the spatial index dimension, and an object index dimension with the object set and rule identifier as the object index dimension. The multi-index structure initially categorizes adjudication events into buckets based on server frame number and intra-frame phase number. Within the same time bucket, spatial overlap index filtering is performed based on the bounding volume of the action region. The filtering results are further subdivided into subsets of adjudication events based on the intersection of object sets and the consistency of rule identifiers. Only within the same subset of adjudication events, pairwise combinations of adjudication events are constructed based on the proximity of server receiving timestamps, generating candidate adjudication event pairs. Overlap is then determined, forming a corresponding adjudication overlap index. When the adjudication overlap index exceeds a preset threshold, a directed graph of adjudication dependency for that candidate adjudication event pair is constructed; otherwise, no directed graph is constructed. The process involves constructing a directed graph of decision dependencies, adding corresponding decision events to a basic decision sorting queue according to a pre-defined stable sorting key, and then executing subsequent decisions. When the directed graph of decision dependencies does not have cyclic dependencies, a topological sort is performed on the decision events based on the graph to generate a basic decision execution sequence. When the directed graph of decision dependencies has cyclic dependencies, strongly connected component decomposition is performed on the graph to generate a minimum conflict decision sequence, thus obtaining a comprehensive decision sequence. The comprehensive decision sequence includes the basic decision execution sequence and the minimum conflict decision sequence. The order stability evaluation parameter set of the comprehensive decision sequence is obtained. If any order stability evaluation parameter is greater than or equal to the corresponding... When the order stability assessment threshold is reached, it is determined that the current logical frame adjudication order is at risk of instability, and order compensation adjustment is performed; otherwise, the comprehensive adjudication sequence remains unchanged and adjudication processing is performed according to the comprehensive adjudication sequence. During the execution of order compensation, when a cloud logical node migration event is detected, the phase correction adjustment of the new node's logical clock and the logical clock advance rate are adjusted. A cloud logical node migration event refers to an event in which, during the adjudication execution process, due to reasons such as load balancing, container scheduling, instance reconstruction, or fault takeover, the cloud logical node undertaking the current logical frame adjudication task undergoes an execution subject switch, causing the adjudication control to be transferred from the original logical node to the new logical node.

[0026] In this embodiment, it should be noted that the comprehensive decision sequence is the final and unique execution order of the current logical frame, that is, the decision execution order.

[0027] like Figure 2 As shown, Figure 2 The flowchart of the overall adjudication scheduling and sequence stability control process of the mobile game cross-platform real-time interaction optimization method based on cloud data synchronization provided in this application embodiment is as follows: First, an adjudication event feature vector is constructed, and candidate adjudication event pairs are generated. Then, an adjudication overlap index is calculated for the candidate adjudication event pairs, and it is determined whether the adjudication overlap index is greater than a preset threshold. When the adjudication overlap index is greater than the threshold, an adjudication dependency directed graph is constructed; otherwise, no adjudication dependency directed graph is constructed, and the sorting stage is directly entered. Next, it is determined whether there is a circular dependency. When there is no circular dependency, topological sorting is performed to generate a basic adjudication execution sequence. When there is a circular dependency, strongly connected component decomposition is performed to generate a minimum conflict adjudication sequence, and then a comprehensive adjudication sequence is formed. After obtaining the comprehensive adjudication sequence, the sequence stability evaluation parameter set is calculated, and it is determined whether there is an instability risk. When there is an instability risk, sequence compensation adjustment is performed and the process ends. When there is no instability risk, adjudication processing is performed according to the comprehensive adjudication sequence and the process ends.

[0028] Furthermore, a set of adjudication events is formed. Specifically, within a logical frame period, the corresponding set of identifier parameters is extracted from each received parallel operation instruction, and an adjudication event feature vector is constructed based on the set of identifier parameters. The set of identifier parameters includes event identifier, server frame number, intra-frame phase number, coordinates of the action center, bounding volume of the action area, action strength, object set, rule identifier, rule priority, and server receiving timestamp. The adjudication event feature vectors generated within the same logical frame period are aggregated to form the adjudication event set of the current logical frame.

[0029] In this embodiment, the adjudication event feature vector is constructed based on the identifier parameter set. The specific method is as follows: Step S21: Within the logical frame period, extract the event identifier, server frame number, intra-frame phase number, action center coordinates, action region bounding volume, action strength, object set, rule identifier, rule priority, and server receiving timestamp for each received parallel operation instruction to form a corresponding identifier parameter set; Step S22: Arrange the identifier parameters extracted in Step S21 according to a preset field order (the field order includes event identifier, server frame number, intra-frame phase number, action center coordinates, action region bounding volume, action strength, object set, rule identifier, rule priority, and server receiving timestamp in sequence); Step S23: Perform numerical normalization processing on the action center coordinates, action region bounding volume, action strength, rule priority, and server receiving timestamp to obtain normalized parameters; Step S24: Combine the event identifier, server frame number, intra-frame phase number, action center coordinates, action region bounding volume, action strength, rule priority, and server receiving timestamp into a normalized parameter set. The frame number, intra-frame phase number, normalization parameter, object set encoding result, and rule identifier encoding result are concatenated to obtain the adjudication event feature vector of the corresponding parallel operation instruction. The specific concatenation method is as follows: the event identifier, server frame number, and intra-frame phase number are directly mapped to the front of the feature vector as discrete index fields. The normalized action center coordinates, action area bounding volume, action strength, rule priority, and server receiving timestamp are arranged in the order of continuous numerical fields. The object set is encoded into a fixed-length object set encoding vector according to the preset object index table. The rule identifier is encoded into a rule identifier encoding field according to the rule index table. Finally, the adjudication event feature vector is obtained by concatenating the discrete index field, continuous numerical field, object set encoding field, and rule identifier encoding field in the existing order. Step S25: The adjudication event feature vectors generated within the logical frame period are collected to form the adjudication event set of the current logical frame.

[0030] The event identifier is a globally unique number assigned by the cloud server to the parallel operation instruction upon receipt, used to uniquely identify an adjudication event; the server frame number is the value of the logical frame counter maintained by the cloud server within the current logical frame period, used to identify the logical frame period to which the adjudication event belongs; the intra-frame phase number is an intra-frame sequence number assigned by the server to the parallel operation instruction within the logical frame period according to the receiving order or time slice division rules, used to characterize the relative sequential stages of the adjudication event within the same logical frame; the center of action coordinates are obtained from the coordinates of the point of action of the target position parameter carried in the parallel operation instruction, representing the spatial center position of the adjudication event; the bounding volume of the action area is determined according to the skill, attack, or... corresponding to the parallel operation instruction. The three-dimensional or two-dimensional bounding volume generated by the preset scope parameters of the interaction rules describes the spatial influence area of ​​the adjudication event; the influence strength is obtained from the numerical parameters carried in the parallel operation instructions, which represents the degree of influence of the adjudication event on the set of objects, such as the displacement distance of the operator; the set of objects is obtained from the set of affected objects retrieved by the server in the spatial index structure; the rule identifier is parsed by the server rule engine and is used to identify the business rule applicable to the adjudication event; the rule priority is a preset priority parameter in the rule base corresponding to the rule identifier, which is used to characterize the rule level between different adjudication events; the server receiving timestamp is obtained from the local logical time recorded by the cloud server when it receives the parallel operation instructions.

[0031] By extracting event identifiers, server frame numbers, intra-frame phase numbers, action center coordinates, action region bounding volume, action strength, object set, rule identifiers, rule priorities, and server reception timestamps from each parallel operation instruction within a logical frame period, and constructing adjudication event feature vectors based on the identifier parameter set, heterogeneous operation instructions with different arrival times are uniformly mapped into a consistent adjudication event representation. This enables unified modeling and precise positioning of adjudication events across spatial, object, rule, and temporal dimensions, solving the problems of difficulty in direct comparison of cross-platform operation instructions, inconsistent adjudication granularity, and difficulty in aligning intra-frame adjudication timing in existing technologies. Furthermore, by aggregating the adjudication event feature vectors generated within the same logical frame period to form an adjudication event set, the consistency, reproducibility, and determinism of cross-node adjudication results in cloud-based adjudication modeling are improved.

[0032] Furthermore, the adjudication overlap index is specifically obtained by extracting adjudication overlap judgment parameters for each candidate adjudication event pair. These parameters include overlap of the target object, overlap of the spatial region, similarity value of the rule domain, and overlap of the temporal phase. After normalization, these parameters are weighted according to a preset adjudication overlap weight set and then summed. Specifically, the adjudication overlap judgment parameters are multiplied by their corresponding adjudication overlap weight sets, and the multiplication results are summed to obtain the adjudication overlap index for the corresponding candidate adjudication event pair. The adjudication overlap weight set includes the overlap weight of the target object, the overlap weight of the spatial region, the similarity weight of the rule domain, and the overlap weight of the temporal phase.

[0033] In this embodiment, the overlap of the action objects represents the degree of overlap between two sets of adjudication event objects. Specifically, it is calculated by dividing the intersection of the two sets of adjudication event objects by their union; the quotient is the overlap of the action objects. The spatial region overlap represents the degree of spatial overlap between the bounding volumes of the action regions of two adjudication events. Specifically, it is calculated by dividing the overlapping volume of the bounding volumes of the action regions of two adjudication events by the volume of their union. The rule domain overlap similarity value represents the degree of similarity between the rules applied to the two adjudication events. Specifically, it is obtained as follows: when the rule identifiers of the two adjudication events are the same, the rule domain overlap parameter is 1; when the rule identifiers are different but the rule categories are the same, the rule domain overlap parameter is a preset intermediate value; when both the rule identifiers and rule categories are different, the rule domain overlap parameter is 0. The temporal phase overlap represents the degree of temporal overlap between the two adjudication events within a logical frame. Specifically, it is calculated as follows: Where D represents the temporal phase overlap, A represents the intra-frame phase number of the first adjudication event, B represents the intra-frame phase number of the second adjudication event, and C represents the preset maximum phase span. The sum of the weights for the overlap of the target object, the spatial region overlap, the rule domain overlap similarity, and the temporal phase overlap equals 1.

[0034] The adjudication overlap weight set, including the overlap weight of the affected object, the overlap weight of the spatial region, the overlap similarity weight of the rule domain, and the overlap weight of the temporal phase, can be obtained from a database. For example, the overlap of the affected object, the overlap of the spatial region, the overlap similarity value of the rule domain, and the overlap of the temporal phase of candidate adjudication event pairs are collected over multiple historical logical frame periods, and the sequence conflict or state offset results of the corresponding candidate adjudication event pairs are recorded simultaneously during the actual adjudication execution, thereby constructing a historical conflict sample set. Based on the historical conflict sample set, the influence intensity coefficient between each overlap judgment parameter and the frequency of adjudication conflict is calculated. For any overlap judgment parameter X, the influence intensity coefficient is calculated according to the formula. The calculation is performed, where Xi is the value of the overlap judgment parameter corresponding to the i-th candidate adjudication event pair, and Yi is the conflict identifier value corresponding to the i-th candidate adjudication event pair. Yi is 1 when a conflict occurs, and 0 otherwise. Thus, the influence intensity coefficients of the overlap of the target object, the overlap of the spatial region, the overlap similarity of the rule domain, and the overlap of the time phase are obtained respectively. Based on the above coefficients, the weights in the adjudication overlap weight set are normalized so that the sum of the four weights is always equal to 1.

[0035] Furthermore, the adjudication dependency directed graph is constructed as follows: when the adjudication overlap index of a candidate adjudication event pair is greater than the corresponding adjudication overlap index threshold, the object causal relationship and logical time phase of the candidate adjudication event pair are obtained, and the execution order of the adjudication events within the candidate adjudication event pair is determined based on the rule priority; each adjudication event within the candidate adjudication event pair is used as a node, and the execution order is used as a directed edge to construct the corresponding adjudication dependency directed graph that describes the sequential constraint relationship between each adjudication event within the candidate adjudication event pair.

[0036] In this embodiment, the object causal relationship is obtained by comparing the object sets and rule identifiers of two adjudication events. When the effect of one adjudication event changes the input state or the attribute of the object of another adjudication event, it is determined that there is an object causal relationship between the two, and the causal direction is recorded. The logical time phase is obtained by directly reading the server frame number and intra-frame phase number corresponding to the two adjudication events, and using the intra-frame phase number as the logical time phase.

[0037] The specific steps for determining the execution order of internal adjudication events within a candidate adjudication event include: S41: Obtaining the object causal relationship between the two adjudication events within the candidate adjudication event; S42: Obtaining the logical time phase of the two adjudication events within the candidate adjudication event; S43: When an object causal relationship exists, determining the execution order according to the causal direction, where "first" refers to the adjudication event that has a pre-constraining effect on subsequent adjudication events in the object state evolution or resource occupation relationship, that is, the creation, initialization, occupation, or state writing events of the same object are executed first as causal precursor events, while events that depend on the state result of that object, the resource release result, or the preceding state change are executed first. The outcome of the ruling event is executed as the causal successor event, thus ensuring the causal consistency of the object's state evolution. The latter is the other event among the two ruling events; S44: When there is no causal relationship between objects, the logical time phase is compared, and the ruling event corresponding to the smaller one is executed first; S45: When the logical time phase is the same, the rule priority is compared, and the ruling event corresponding to the rule with the higher priority is executed first; S46: Otherwise, according to the sorting key of this stable sorting, the candidate ruling events are deterministically sorted according to the event identifier and the server receiving timestamp, and the events with the earlier sorting are executed first, and the events with the later sorting are executed later.

[0038] Furthermore, the basic adjudication execution sequence is generated as follows: when there is no circular dependency in the directed graph of adjudication dependency, the adjudication events are topologically sorted based on the directed graph of adjudication dependency; during the same round of topological sorting, when there are two or more adjudication event nodes with an in-degree of zero, the adjudication event nodes are sorted and selected according to the preset stable sorting key to generate the basic adjudication execution sequence.

[0039] In this embodiment, the basic adjudication execution sequence is generated as follows: S51: During the same round of topological sorting, when there are two or more adjudication event nodes with an in-degree of zero in the candidate queue, the in-degree refers to the number of directed edges pointing to a certain node in the directed graph, that is, the number of predecessor adjudication events that the current adjudication event node depends on in the adjudication dependency directed graph. When the in-degree is zero, it means that the adjudication event node does not have any incomplete predecessor dependency constraints and is qualified to be scheduled for execution in the current round of topological sorting. The sorting key of the stable sorting corresponding to the adjudication event node with an in-degree of zero is extracted; S52: The adjudication event nodes with an in-degree of zero are sorted according to the ascending order rule of the sorting key of the stable sorting; S53: The adjudication event nodes are taken out in order according to the sorting result and added to the adjudication execution sequence. Specifically, adding nodes to the adjudication execution sequence can be achieved through topological sorting output rules. The specific steps are as follows: After sorting, adjudication event nodes with an in-degree of zero are appended sequentially to the end of the current basic adjudication execution sequence according to the order determined by the stable sorting key, rather than being inserted in the middle of the sequence. This maintains the linear output characteristic of "first-choose-first-execute, last-choose-last-execute" during topological sorting. After each node is added to the basic adjudication execution sequence, the node and its outgoing edges are simultaneously deleted, and the in-degree value of its successor adjudication event nodes is updated. Once a new node with an in-degree of zero enters the candidate queue, the above selection and appending process continues to ensure that the final generated basic adjudication execution sequence simultaneously satisfies the adjudication dependency constraint relationship and the stable sorting determinism requirement. Only adjudication event nodes with an in-degree of zero are analyzed here, because the set of selectable nodes for topological sorting is limited to the current set of nodes with an in-degree of zero. Only these nodes are qualified to be "selected for execution," and sorting all nodes would be meaningless and would violate dependency constraints.

[0040] By triggering the construction of the decision dependency directed graph only when the decision overlap index of a candidate decision event pair exceeds the corresponding decision overlap index threshold, the system establishes sequential constraints only for decision event pairs with substantial coupling relationships at the levels of the target object, spatial region, rule domain, and temporal phase. This avoids introducing unnecessary dependencies on a large number of unrelated or weakly related decision events, reducing the scale of dependency modeling and topological sorting complexity, and minimizing the impact of invalid constraints on the efficiency of parallel decision-making. Simultaneously, by introducing object causality, logical temporal phase, and rule priority to jointly determine the execution direction after triggering the construction condition, each directed edge originates from natural causal relationships and rule constraints rather than thread scheduling order. This explicitly characterizes the sequential constraints between non-commutative decision operations in the decision dependency directed graph, enabling decision events within the same logical frame to generate a consistent sequential constraint structure under different thread scheduling, different node migration, or different platform merging conditions. This provides a deterministic constraint basis for subsequent topological sorting and cycle stabilization, effectively eliminating the problem of non-unique decision results caused by random scheduling order during parallel decision-making, and improving the determinism, reproducibility, and cross-platform consistency of authoritative cloud decision results.

[0041] Furthermore, the minimum conflict resolution sequence is generated by the following method: when there is a cyclic dependency in the directed graph of the decision dependency, the strongly connected component decomposition is performed on the directed graph of the decision dependency to obtain each cyclic decision event group; the internal sorting of each cyclic decision event group is fixed according to the sorting key of the preset stable sorting, and the sorted cyclic decision event group is regarded as a super node to participate in the outer topology sorting process, thereby generating the minimum conflict resolution sequence corresponding to each cyclic decision event group.

[0042] In this embodiment, as Figure 3 As shown, Figure 3 The flowchart for the cyclic decision dependency processing and minimum conflict decision sequence generation of the mobile game cross-platform real-time interaction optimization method based on cloud data synchronization provided in this application embodiment is as follows: First, a decision dependency directed graph is input and it is determined whether a cyclic dependency exists; when there is no cyclic dependency in the decision dependency directed graph, a topological sort is performed on the decision events based on the decision dependency directed graph; when a cyclic dependency exists, a strongly connected component decomposition is performed on the decision dependency directed graph to obtain a cyclic decision event group, and a fixed sort is performed within each cyclic decision event group according to the sort key of the stable sorting, thereby generating a comprehensive decision sequence and ending the process.

[0043] The strongly connected component decomposition is performed on the directed graph of decision dependency to obtain each cyclic decision event group. The specific method is as follows: S61: Taking each decision event node in the directed graph of decision dependency as the starting point of traversal, the directed graph of decision dependency is recursively visited according to the preset depth-first traversal rule, and an access sequence number is assigned to each decision event node during the visit and the current backtracking minimum access sequence number is recorded; S62: During the traversal, when the backtracking minimum access sequence number of a certain decision event node is equal to its own access sequence number, the decision event node and its successor nodes on the current recursive path are popped from the node stack in sequence to form a cyclic decision event group; S63: The traversal and grouping process is repeated until all decision event nodes in the directed graph of decision dependency have been visited.

[0044] It should be noted that the Depth-First Search (DFS) algorithm is used for the sub-process of strongly connected component decomposition. The input object of this DFS algorithm is a directed graph with decision dependency, the operation objects are decision event nodes and directed edges, the intermediate quantities are the access sequence number, the backtracking minimum access sequence number, and the node stack, and the output result is several cyclic decision event groups (strongly connected components).

[0045] By proactively detecting cyclic dependencies in the directed graph of decision dependencies and performing strongly connected component decomposition, the set of decision events that were originally mutually constrained and sequentially locked and could not be directly linearly ordered is automatically identified into several cyclic decision event groups. This avoids the sorting failure or randomization problems caused by direct topological sorting under the presence of cyclic constraints. At the same time, by fixing the sorting key based on stable sorting within each cyclic decision event group, the same cyclic group always generates a consistent decision order within the group under different scheduling environments, different nodes, or different execution rounds, eliminating the drift of decision results caused by differences in thread scheduling within the loop. The cyclic decision event groups after fixed sorting are regarded as super nodes participating in the outer topological sorting, so that the complex cyclic dependency relationship is compressed into a linearly sortable high-level structure, minimizing the number of events that must be forcibly sorted while ensuring the correctness of global order constraints.

[0046] Furthermore, order compensation adjustment is performed. Specifically, based on comprehensive adjudication sequence analysis, partial order stability index, order determinism index, and dependency chain depth index are obtained, thereby forming an order stability evaluation parameter set. When any index in the order stability evaluation parameter set exceeds the corresponding order stability evaluation threshold, the execution phase parameter and adjudication delay window parameter of the adjudication event on the preset critical path in the adjudication dependency directed graph within the current logical frame are dynamically adjusted to reduce the fluctuation amplitude of the adjudication order and restore the stability of the adjudication sequence. The order stability evaluation threshold includes the partial order stability threshold, the order determinism threshold, and the dependency chain depth threshold.

[0047] In this embodiment, the partial order stability index represents the degree to which the comprehensive decision sequence maintains the partial order relationship of decision dependencies within consecutive logical frames. It is obtained by extracting the corresponding comprehensive decision sequences in two adjacent logical frame periods, and based on the set of partial order constraints defined in the decision dependency directed graph, calculating the ratio of the number of decision event pairs satisfying the same order relationship to the total number of constrained decision event pairs in the two frame sequences. Specifically, the ordered decision event pair set in the comprehensive decision sequence of the previous logical frame can be matched with the corresponding ordered decision event pair set in the current logical frame, and the proportion of unchanged partial order relationship pairs to the total number of partial order relationship pairs can be used as the partial order stability index. The order determinism index represents the degree of certainty in the decision order selection during topological sorting and conflict resolution within the same logical frame. It is obtained by: when generating the comprehensive decision sequence, calculating the size of the candidate set of decision event nodes with a zero in-degree in each round of topological sorting, and calculating the average size of the candidate set in each round. A larger candidate set size indicates more possible execution order branches and higher order uncertainty. The reciprocal of the average candidate set size is used to obtain the order determinism index. The dependency chain depth metric represents the length of the longest dependency path in a directed graph of decision dependencies. It is obtained by taking a node with an in-degree of zero as the starting point and a node with an out-degree of zero as the ending point in the directed graph of decision dependencies, and calculating the sum of the number of nodes in each dependency path using the longest path search method (such as a dynamic programming algorithm based on topological sorting).

[0048] In a directed graph where adjudication depends on a predetermined critical path, the adjudication event refers to the sequence of adjudication events that have the longest dependency chain or the largest cumulative delay weight, formed from a node with an in-degree of zero to a node with an out-degree of zero. The out-degree refers to the number of directed edges originating from a node in the directed graph, which represents the number of subsequent adjudication events constrained by the current adjudication event node as its predecessor node.

[0049] The dynamic adjustment of the execution phase parameters and the adjudication delay window parameters is achieved as follows: Within each logical frame, based on the actual execution time sequence of each adjudication event in the comprehensive adjudication sequence and its target execution time sequence in the reference stable sequence, time alignment processing is performed to obtain the phase deviation vector of adjacent adjudication events on the critical path. This vector is then averaged and filtered to obtain the average phase offset. Specifically, within the same logical frame, the difference between the actual execution time of each adjacent adjudication event on the critical path and its target execution time difference in the reference stable sequence are extracted, and the corresponding time deviation components are calculated. The time deviation components corresponding to each adjacent adjudication event on the critical path are arranged in the path order to form a phase deviation vector. The phase deviation vector is used to characterize the overall phase drift direction and amplitude of the current critical path. When the average phase offset is greater than the preset execution phase threshold, the execution phase parameters of each adjudication event on the critical path within the current logical frame are updated by subtracting the product of the current adjudication event execution phase parameters and the preset phase adjustment coefficient. The waiting time distribution of adjudication events on the critical path is statistically analyzed. This distribution can be obtained by statistically analyzing the scheduling and queuing process of each adjudication event on the critical path. Specifically, when an adjudication event enters the executable state (in-degree becomes zero or enters the candidate queue), its executable timestamp is recorded. When it is actually scheduled for execution, its execution start timestamp is recorded. The difference between the two timestamps is the waiting time of the adjudication event. Waiting time samples of all adjudication events on the critical path within the current logical frame are aggregated to form a waiting time distribution. Based on this distribution, the waiting time variance is calculated as an adjudication delay dispersion index. This index is used to characterize the fluctuation of the critical path scheduling delay and guide the adaptive adjustment of the delay window parameters. When the waiting time variance exceeds the adjudication delay dispersion threshold, the adjudication delay is adjusted accordingly. The window parameters are adaptively expanded and shrunk. Specifically, the waiting time variance is subtracted from the current decision delay window, multiplied by the preset delay window adjustment coefficient, and then added back to the current delay window adjustment coefficient to obtain a new decision delay window. After updating the phase parameters and delay window parameters, the phase parameters and decision delay window are reordered according to the updated decision events, and the critical path decision events are executed. This forms a new comprehensive decision sequence in the next logical frame and enters the next round of stability assessment. It should be noted that the reference stable sequence can be the comprehensive decision sequence when the sequence stability assessment parameter sets in multiple historical logical frames have not exceeded the threshold. The average execution time of each decision event in this sequence is used as the target execution time sequence. By aligning and comparing the current actual execution time with the target execution time, the impact of instantaneous scheduling disturbances can be eliminated.

[0050] By evaluating the sequential stability of the comprehensive adjudication sequence, when anomalies are detected in the partial order stability index, sequential determinism index, or dependency chain depth index, the execution phase parameter and adjudication delay window parameter of the adjudication event on the critical path are dynamically adjusted. This corrects the actual execution time of the adjudication event and adaptively compensates for the adjudication waiting time, thereby reducing the fluctuation amplitude of the adjudication order, suppressing the cascading delay of the dependency chain, and restoring the temporal consistency of the adjudication sequence. This solves the problems of easy jitter in the adjudication order, uncontrolled dependency propagation, and asynchronous state between logical frames in the cross-platform real-time interaction process in the existing technology, and improves the consistency and stability of multi-terminal interaction.

[0051] Furthermore, phase correction adjustment is performed on the logic clock of the new node. The specific method is as follows: obtain the local logic time of the new node and the reference logic time of the current decision domain, and perform difference processing to obtain the logic time phase offset; when the absolute value of the logic time phase offset exceeds the preset phase offset threshold and the logic time phase offset is negative, the logic time of the new node is lagging, and phase acceleration correction adjustment is performed; when the absolute value of the logic time phase offset exceeds the preset phase offset threshold and the logic time phase offset is positive, the logic time of the new node is ahead, and phase deceleration correction adjustment is performed; otherwise, no phase correction adjustment is performed.

[0052] In this embodiment, phase correction adjustment includes phase acceleration correction adjustment and phase deceleration correction adjustment. Phase correction adjustment specifically includes: S81: Obtaining the new node's local logic time Tn and the current decision domain reference logic time Tr; S82: Calculating the logic time phase offset ΔT, where ΔT = Tn - Tr; S83: When |ΔT| ≥ the phase offset threshold and ΔT < 0, determining that the new node's logic time is lagging, and increasing the logic clock advance rate according to a preset acceleration coefficient. Specifically, this is achieved by adding the current logic clock advance rate to the product of the current logic clock advance rate and the acceleration coefficient. The logic clock advance rate is given by: S84: When |ΔT|≥ phase offset threshold and ΔT>0, the logic time of the new node is determined to be ahead, and the logic clock advance rate is reduced according to the preset deceleration coefficient. Specifically, the reduced logic clock advance rate is obtained by subtracting the product of the current logic clock advance rate and the deceleration coefficient from the current logic clock advance rate; S85: The logic time of the new node is updated and the remaining phase offset is recalculated within each sampling time window; S86: When the remaining phase offset enters the preset safety range, the phase correction adjustment is terminated and the normal logic clock advance rate is restored.

[0053] It should be noted that the local logical time of the new node can be obtained by reading the current logical count value in the local logical clock module of the new node. The logical count value is continuously accumulated by the logical clock generator inside the new node based on the local event triggering frequency and clock advancement strategy. The current adjudication domain reference logical time is obtained by sending a time synchronization request to the reference node in the adjudication domain. The reference node periodically broadcasts its current logical timestamp, thereby obtaining the unified logical time reference of the current adjudication domain. The logical clock advancement rate is the logical time increment step of the new node's logical clock within a unit physical time. This advancement rate is determined by the new node's logical clock module according to the preset basic advancement rate parameters and is dynamically updated in combination with the current event processing frequency. The phase offset recalculated within each sampling time window is the time phase offset of the new node's logical time relative to the current decision domain reference logical time. It is obtained by reading the updated local logical time of the new node again at the end of each sampling time window, and synchronously obtaining the decision domain reference logical time at the corresponding moment. The remaining phase offset is obtained by subtracting the decision domain reference logical time at the corresponding moment from the updated local logical time of the new node. The remaining phase offset is used to determine whether the preset safe zone has been entered and to control the termination of the phase correction adjustment.

[0054] By acquiring the local logical time of the new node and the reference logical time of the adjudication domain and calculating the phase offset of the logical time, phase acceleration correction or phase deceleration correction is performed respectively when the offset exceeds the phase offset threshold. This corrects the leading or lagging state of the new node's logical clock in a targeted manner, thereby quickly eliminating time drift and phase mismatch in the early stage of new node access. This avoids problems such as time base splitting in the adjudication domain, event sorting disorder, and cross-platform state playback deviation.

[0055] Furthermore, the logic clock advance rate is adjusted using the following method: The logic clock advance rate adjustment value is obtained based on the phase offset and a preset correction window length. The logic clock advance rate is then adjusted under preset adjustment constraints. These constraints are that the logic clock advance rate is within a preset maximum and minimum advance rate range. If the logic clock advance rate exceeds the constraints during adjustment, it is adjusted according to the maximum or minimum advance rate within the constraints. If the logic clock advance rate is greater than or equal to the maximum advance rate in the constraints, it is adjusted according to the maximum advance rate; conversely, if the logic clock advance rate is less than or equal to the minimum advance rate in the constraints, it is adjusted according to the minimum advance rate. During the logic clock advance rate adjustment, sampling is performed according to a preset sampling time window size to obtain each logic advance cycle. The new node logic time is updated according to the logic clock advance rate adjustment value within each logic advance cycle, and the current remaining phase offset is updated synchronously. When the remaining phase offset is within a preset safety range, the advance rate adjustment is terminated, and normal adjudication scheduling is restored.

[0056] In this embodiment, the logic clock advance rate adjustment value is obtained by analyzing the phase offset and the preset correction window length. The specific method is as follows: obtain the preset correction window length, which represents the target logic time span within which phase convergence adjustment is allowed to be completed. Divide the current phase offset by the correction window length to obtain the logic time offset that needs to be compensated within a unit advance cycle. Use this unit compensation amount as the basis for adjusting the logic clock advance rate, that is, add or subtract the corresponding advance step size on the basis of the original basic logic clock advance rate to obtain the logic clock advance rate adjustment value. If the current phase offset is negative, it indicates that the logic time of the new node is lagging, so the corresponding proportion of the advance step size is increased on the basic advance rate. If the current phase offset is positive, it indicates that the logic time of the new node is ahead, so the corresponding proportion of the advance step size is reduced on the basic advance rate.

[0057] By calculating the logic clock advance rate adjustment value based on the phase offset and the correction window length, and limiting the advance rate within the adjustment constraints formed by the maximum and minimum advance rates, and continuously updating the logic time of the new node and the remaining phase offset according to the sampling time window in each logic advance cycle, the logic clock can be smoothly converged within the controlled rate range. This avoids time jitter, decision jumps and event reversal caused by one-time forced correction, and effectively ensures the continuity of the logic time correction process.

[0058] like Figure 4 As shown, Figure 4This is a structural framework diagram of a mobile game cross-platform real-time interaction optimization device based on cloud data synchronization provided in this application embodiment. The device is used for a mobile game cross-platform real-time interaction optimization method based on cloud data synchronization. The device includes: a parallel analysis module, an adjudication overlap determination module, an adjudication sequence generation module, an adjudication processing module, and a compensation correction module. The parallel analysis module is used by the cloud server to collect parallel operation instructions from various clients and the server in real time within a logical frame period, and convert them into corresponding adjudication event feature vectors to form an adjudication event set. The adjudication overlap determination module is used to index and classify the adjudication event set, generate candidate adjudication event pairs, and perform overlap determination to form corresponding adjudication overlap indices. When the adjudication overlap indices are greater than a preset adjudication overlap indices threshold, an adjudication dependency directed graph of the candidate adjudication event pair is constructed; otherwise, the adjudication dependency directed graph is not constructed. The adjudication sequence generation module is used to perform topological sorting of the adjudication events based on the adjudication dependency directed graph when there is no cyclic dependency. The system generates a basic decision execution sequence. When a cyclic dependency exists in the directed graph of decision dependencies, it performs strongly connected component decomposition on the directed graph to generate a minimum conflict decision sequence, thus obtaining a comprehensive decision sequence. The comprehensive decision sequence includes the basic decision execution sequence and the minimum conflict decision sequence. The decision processing module is used to obtain the order stability evaluation parameter set of the comprehensive decision sequence. When any order stability evaluation parameter is greater than or equal to the corresponding order stability evaluation threshold, it is determined that the decision order of the current logical frame has an unstable risk, and order compensation adjustment is performed. Otherwise, the comprehensive decision sequence remains unchanged and decision processing is performed according to the comprehensive decision sequence. The compensation correction module is used to perform phase correction adjustment on the logical clock of the new node and adjust the logical clock advance rate when a cloud logical node migration event is detected during the execution of order compensation.

[0059] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0060] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0061] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0062] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0063] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0064] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for optimizing cross-platform real-time interaction in mobile games based on cloud data synchronization, characterized in that: Includes the following steps: The cloud server collects parallel operation instructions from various clients and servers in real time within the logical frame period, and converts them into corresponding adjudication event feature vectors to form an adjudication event set. The set of adjudication events is indexed and classified to generate candidate adjudication event pairs. Overlap is determined to form a corresponding adjudication overlap index. When the adjudication overlap index is greater than a preset adjudication overlap index threshold, an adjudication dependency directed graph of the candidate adjudication event pair is constructed; otherwise, an adjudication dependency directed graph is not constructed. When there is no cyclic dependency in the directed graph of adjudication dependency, a topological sort is performed on the adjudication events based on the directed graph of adjudication dependency to generate a basic adjudication execution sequence. When there is cyclic dependency in the directed graph of adjudication dependency, a strongly connected component decomposition is performed on the directed graph of adjudication dependency to generate a minimum conflict adjudication sequence, thereby obtaining a comprehensive adjudication sequence, which includes the basic adjudication execution sequence and the minimum conflict adjudication sequence. Obtain the set of order stability evaluation parameters for the comprehensive decision sequence. If any order stability evaluation parameter is greater than or equal to the corresponding order stability evaluation threshold, it is determined that the current logic frame decision sequence has an unstable risk and order compensation adjustment is performed. Otherwise, the comprehensive decision sequence is maintained unchanged and decision processing is performed according to the comprehensive decision sequence. During the execution of sequence compensation, when a cloud logical node migration event is detected, the phase correction adjustment is performed on the logical clock of the new node, and the logical clock advance rate is adjusted.

2. The mobile game cross-platform real-time interaction optimization method based on cloud data synchronization as described in claim 1, characterized in that: The specific method for forming the set of adjudication events is as follows: Within a logical frame period, the corresponding set of identifier parameters is extracted from each received parallel operation instruction, and an adjudication event feature vector is constructed based on the set of identifier parameters. The set of identifier parameters includes event identifier, server frame number, intra-frame phase number, coordinates of the action center, bounding volume of the action area, action strength, object set, rule identifier, rule priority, and server receiving timestamp; The feature vectors of each adjudication event generated within the same logical frame period are aggregated to form the adjudication event set of the current logical frame.

3. The mobile game cross-platform real-time interaction optimization method based on cloud data synchronization as described in claim 1, characterized in that: The method for the aforementioned adjudication overlap index is as follows: For each candidate adjudication event pair, the adjudication overlap determination parameters are extracted. These parameters include the overlap of the target object, the overlap of the spatial region, the overlap similarity value of the rule domain, and the overlap of the temporal phase. After normalization, the parameters are weighted according to a preset set of adjudication overlap weights and then summed to obtain the adjudication overlap index of the corresponding candidate adjudication event pair. The adjudication overlap weight set includes the overlap weight of the target object, the overlap weight of the spatial region, the overlap similarity weight of the rule domain, and the overlap weight of the temporal phase.

4. The mobile game cross-platform real-time interaction optimization method based on cloud data synchronization as described in claim 2, characterized in that: The adjudication depends on a directed graph, and the specific construction method is as follows: When the adjudication overlap index of a candidate adjudication event pair is greater than the corresponding adjudication overlap index threshold, the object causal relationship and logical time phase of the candidate adjudication event pair are obtained, and the execution order of the adjudication events within the candidate adjudication event pair is determined based on the rule priority. Using each decision event within a candidate decision event pair as a node and the execution order as directed edges, a corresponding decision dependency directed graph is constructed to characterize the sequential constraints between each decision event within a candidate decision event pair.

5. The mobile game cross-platform real-time interaction optimization method based on cloud data synchronization as described in claim 1, characterized in that: The specific method for generating the basic adjudication execution sequence is as follows: When there is no circular dependency in the directed graph of decision dependency, perform topological sorting on the decision events based on the directed graph of decision dependency; During the same round of topology sorting, when there are two or more adjudication event nodes with an in-degree of zero, the adjudication event nodes are sorted and selected according to the preset stable sorting key to generate a basic adjudication execution sequence.

6. The mobile game cross-platform real-time interaction optimization method based on cloud data synchronization as described in claim 1, characterized in that: The specific method for generating the minimum conflict resolution sequence is as follows: When a directed graph with decision dependency has cyclic dependencies, perform strongly connected component decomposition on the directed graph with decision dependency to obtain each cyclic decision event group. The cyclic decision event groups are sorted according to a preset stable sorting key, and the sorted cyclic decision event groups are regarded as super nodes to participate in the outer topology sorting process, thereby generating the minimum conflict decision sequence corresponding to each cyclic decision event group.

7. The method for optimizing cross-platform real-time interaction in mobile games based on cloud data synchronization as described in claim 1, characterized in that: The specific method for performing sequential compensation adjustment is as follows: Based on comprehensive decision sequence analysis, partial order stability index, order determinism index, and dependency chain depth index are obtained respectively, thus forming a set of order stability evaluation parameters; When any index in the sequence stability evaluation parameter set exceeds the corresponding sequence stability evaluation threshold, the execution phase parameter and the decision delay window parameter of the decision event on the preset critical path in the decision dependency directed graph within the current logical frame are dynamically adjusted to reduce the fluctuation amplitude of the decision order and restore the stability of the decision sequence. The order stability evaluation thresholds include the partial order stability threshold, the order determinism threshold, and the dependency chain depth threshold.

8. The method for optimizing cross-platform real-time interaction in mobile games based on cloud data synchronization as described in claim 1, characterized in that: The specific method for performing phase correction and adjustment on the logic clock of the new node is as follows: Obtain the local logical time of the new node and the reference logical time of the current adjudication domain, and perform interpolation to obtain the logical time phase offset; When the absolute value of the logical time phase offset exceeds the preset phase offset threshold and the logical time phase offset is negative, the logical time of the new node is lagging, and a phase acceleration correction adjustment is performed. When the absolute value of the logical time phase offset exceeds the preset phase offset threshold and the logical time phase offset is positive, the logical time of the new node is ahead, and a phase deceleration correction adjustment is performed. Otherwise, phase correction will not be performed.

9. The mobile game cross-platform real-time interaction optimization method based on cloud data synchronization as described in claim 8, characterized in that: The specific method for adjusting the logic clock advance rate is as follows: The logic clock advance rate adjustment value is obtained by analyzing the phase offset and the preset correction window length. The logic clock advance rate is adjusted based on the logic clock advance rate adjustment value under preset adjustment constraints. The adjustment constraints are that the logic clock advance rate is within the preset maximum and minimum advance rate range. If the logic clock advance rate exceeds the constraints during the adjustment process, the adjustment is carried out according to the maximum or minimum advance rate within the constraints. During the adjustment of the logic clock advance rate, sampling is performed according to the preset sampling time window size to obtain each logic advance cycle. The logic time of the new node is updated according to the logic clock advance rate adjustment value within each logic advance cycle, and the current remaining phase offset is updated synchronously. When the remaining phase offset is within the preset safety range, the advance rate adjustment is terminated and normal adjudication scheduling is restored.

10. An apparatus for applying the cloud-based data synchronization-based mobile game cross-platform real-time interaction optimization method as described in any one of claims 1-9, characterized in that, include: The system includes a parallel analysis module, an adjudication overlap determination module, an adjudication sequence generation module, an adjudication processing module, and a compensation and correction module. The parallel analysis module is used by the cloud server to collect parallel operation instructions from each client and server in real time within the logical frame period, and convert them into corresponding adjudication event feature vectors to form an adjudication event set. The adjudication overlap determination module is used to index and classify the adjudication event set, generate each candidate adjudication event pair, and perform overlap determination to form a corresponding adjudication overlap index. When the adjudication overlap index is greater than the preset adjudication overlap index threshold, an adjudication dependency directed graph of the candidate adjudication event pair is constructed; otherwise, an adjudication dependency directed graph is not constructed. The adjudication sequence generation module is used to perform topological sorting on the adjudication events based on the adjudication dependency directed graph when there is no cyclic dependency, and generate a basic adjudication execution sequence. When there is cyclic dependency in the adjudication dependency directed graph, it performs strongly connected component decomposition on the adjudication dependency directed graph to generate a minimum conflict adjudication sequence, thereby obtaining a comprehensive adjudication sequence. The comprehensive adjudication sequence includes the basic adjudication execution sequence and the minimum conflict adjudication sequence. The adjudication processing module is used to obtain the sequence stability evaluation parameter set of the comprehensive adjudication sequence. When any sequence stability evaluation parameter is greater than or equal to the corresponding sequence stability evaluation threshold, it is determined that the current logical frame adjudication sequence has an unstable risk and sequence compensation adjustment is performed. Otherwise, the comprehensive adjudication sequence is maintained unchanged and adjudication processing is performed according to the comprehensive adjudication sequence. The compensation and correction module is used to perform phase correction and adjustment on the logic clock of the new node and adjust the logic clock advance rate when a cloud logic node migration event is detected during the execution of sequential compensation.

Images