Mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index

Abstract

The application discloses a mobile terminal two-dimensional geometry library resource scheduling method based on a spatial grid index, relates to the technical field of mobile terminal two-dimensional geometry spatial index scheduling, and comprises the following steps: receiving and deconstructing an original query request input by a user in real time, performing query intention analysis, and generating an initial computing power scheduling strategy; according to a real-time load state, performing dynamic constraint adjustment on the initial computing power scheduling strategy, and outputting an efficiency-aware computing power distribution strategy; scheduling a main spatial index library and an auxiliary attribute index library to perform parallel index retrieval; performing gradual retrieval range convergence of multi-thread index cross-validation multi-stage result screening, and positioning a target geometry resource. The application solves the technical problems that, in the prior art, mobile terminal geometry query computing power distribution is unreasonable, retrieval positioning is inefficient, and precision is insufficient, and achieves the technical effects of realizing mobile terminal two-dimensional geometry resource dynamic computing power scheduling and parallel retrieval, and improving the efficiency and precision of geometry resource query positioning.

Description

Technical Field

[0001] This invention relates to the field of mobile terminal two-dimensional geometric spatial index scheduling technology, specifically to a mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index. Background Technology

[0002] Currently, mobile applications of 2D geometric resources are becoming increasingly widespread. Traditional geometric library retrieval methods have fixed and rigid computing power allocation, which cannot adapt to the real-time load and power consumption fluctuations of devices. The index retrieval mode is singular, multi-dimensional data queries are difficult to process in parallel, and the retrieval scope cannot be efficiently converged. This results in problems such as slow query response, low positioning accuracy, and unbalanced device resource consumption, making it difficult to meet the actual usage needs of mobile devices for efficient and accurate retrieval of geometric resources.

[0003] Existing technologies suffer from problems such as unreasonable allocation of computing power for geometric queries on mobile devices, resulting in inefficient retrieval and positioning with insufficient accuracy. Summary of the Invention

[0004] This application provides a resource scheduling method for mobile two-dimensional geometry libraries based on spatial grid indexing, which is used to address the technical problems of unreasonable allocation of computing power for mobile geometry queries, and inefficient and inaccurate retrieval and positioning in existing technologies.

[0005] In view of the above problems, this application provides a method for scheduling mobile terminal two-dimensional geometry library resources based on spatial grid index.

[0006] This application provides a method for scheduling mobile two-dimensional geometry library resources based on spatial grid indexing, the method comprising: The system receives and deconstructs raw query requests input by real-time users, performs query intent parsing, and generates an initial computing power scheduling strategy. Based on the real-time load status of the main spatial index library and the auxiliary attribute index library, combined with the real-time power consumption status of the mobile device, the system dynamically adjusts the constraints of the initial computing power scheduling strategy, outputting an efficiency-aware computing power allocation strategy. This includes indexing and pre-coding the original geometric resource set to construct the main spatial index library and the auxiliary attribute index library. Driven by the efficiency-aware computing power allocation strategy, the system schedules the main spatial index library and the auxiliary attribute index library to perform parallel index retrieval based on the raw query request. During the parallel index retrieval process, a progressive retrieval range convergence is achieved through multi-level result filtering under multi-threaded index cross-validation to locate the target geometric resource.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: The system receives and deconstructs raw query requests from real-time user input, performs query intent parsing, and generates an initial computing power scheduling strategy. Based on the real-time load status of the main spatial index and auxiliary attribute index, combined with the real-time power consumption status of the mobile device, the system dynamically adjusts the constraints of the initial computing power scheduling strategy, outputting an efficiency-aware computing power allocation strategy. Driven by the efficiency-aware computing power allocation strategy, the system schedules the main spatial index and auxiliary attribute index to perform parallel index retrieval based on the raw query request. During the parallel index retrieval process, a progressive retrieval range convergence is achieved through multi-threaded index cross-validation and multi-level result filtering to locate the target geometric resource. This achieves the technical effect of realizing dynamic computing power scheduling and parallel retrieval of two-dimensional geometric resources on mobile devices, improving the efficiency and accuracy of geometric resource query and location. Attached Figure Description

[0008] 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.

[0009] Figure 1 A schematic diagram of the mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index provided in the embodiments of this application; Figure 2 This is a schematic diagram illustrating the process of constructing a main spatial index library and an auxiliary attribute index library in the mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index provided in the embodiments of this application. Detailed Implementation

[0010] This application provides a resource scheduling method for mobile two-dimensional geometry libraries based on spatial grid indexing, which is used to address the technical problems of unreasonable allocation of computing power for mobile geometry queries, resulting in inefficient retrieval and positioning accuracy in existing technologies.

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0012] Examples, such as Figure 1 As shown, this application provides a method for scheduling mobile terminal two-dimensional geometry library resources based on spatial grid indexing, the method comprising: Step S100: Receive and deconstruct the original query request input by the real-time user, perform query intent parsing, and generate an initial computing power scheduling strategy.

[0013] Specifically, the system receives and deconstructs raw query requests input by real-time users, performs syntactic and semantic parsing on the raw query requests to generate standardized query expressions, performs structured parsing on the standardized query expressions to extract spatial constraints and attribute constraints, maps multiple attribute constraint sub-conditions in the attribute constraints to corresponding attribute dimension index sub-libraries, calculates the auxiliary index computing power weights guided by energy efficiency by backtracking historical query performance data locally, obtains the auxiliary attribute index computing power allocation weights, determines the query intent pattern based on the parsing of spatial constraints and attribute constraints, and calculates the main spatial index computing power allocation weights accordingly. The auxiliary attribute index computing power allocation weights and the main spatial index computing power allocation weights together constitute the initial computing power scheduling strategy.

[0014] Step S200: Based on the real-time load status of the main space index library and the auxiliary attribute index library, and combined with the real-time power consumption status of the mobile terminal, the initial computing power scheduling strategy is dynamically constrained and adjusted, and an efficiency-aware computing power allocation strategy is output. In this process, the original geometric resource set is indexed and pre-encoded and marked to construct the main space index library and the auxiliary attribute index library.

[0015] Specifically, the original geometric resource set is pre-indexed and pre-encoded. A quadtree spatial grid index is constructed through hierarchical main label spatial grid encoding to form a main spatial index library. The original geometric resource set is marked with multi-dimensional auxiliary label attributes and multiple attribute dimension index sub-libraries are constructed through inverted index processing. After label mapping configuration, they are integrated to form an auxiliary attribute index library. Then, the real-time load data of the main spatial index library and each attribute dimension index sub-library are retrieved. Combined with the real-time power consumption status of the mobile terminal, power consumption-aware load margin calculation is carried out to obtain the main spatial computing power margin and the computing power margin of each auxiliary index. Based on the computing power margin, the computing power allocation weight of the auxiliary attribute index and the computing power allocation weight of the main spatial index library in the initial computing power scheduling strategy are flexibly and dynamically adjusted. Finally, the performance-aware computing power allocation strategy is output.

[0016] Step S300: Driven by the performance-aware computing power allocation strategy, schedule the main space index library and the auxiliary attribute index library to perform parallel index retrieval based on the original query request.

[0017] Specifically, driven by the performance-aware computing power allocation strategy, the spatial constraints and multiple attribute constraint sub-conditions corresponding to the original query request are mapped to the main index query task and multiple auxiliary index query tasks, respectively. The performance-aware computing power allocation strategy allocates corresponding computing resources to the main index query task and each auxiliary index query task. The index retrieval is executed synchronously and in parallel on the main spatial index library and each attribute dimension index sub-library. During the parallel index retrieval process, the main spatial index library and each attribute dimension index sub-library send incremental result heartbeats containing candidate set bitmaps, retrieval progress indicators and resource utilization efficiency to the retrieval scheduling platform according to a preset time window.

[0018] Step S400: During the parallel index retrieval process, a progressive retrieval range convergence is performed through multi-level result filtering under multi-threaded index cross-validation to locate the target geometric resource.

[0019] Specifically, during the parallel index retrieval process, the retrieval scheduling platform receives incremental result heartbeats from the main spatial index library and each auxiliary attribute index library. Using the performance-aware computing power allocation strategy as the confidence basis, it performs fast intersection calculation on the multi-way candidate set bitmap to obtain the intersection candidate set bitmap and the auxiliary attribute index computing power correction weight. The intersection candidate set bitmap is fed back to each index library, and the computing power of each auxiliary attribute index library is rescheduled in real time according to the correction weight. This drives the main spatial index library and each attribute dimension index sub-library to execute the query task in parallel again with the intersection candidate set bitmap as the retrieval range constraint. Through iterative multi-threaded index cross-validation, multi-level result filtering, and progressive convergence of the retrieval range, the target geometric resource is finally accurately located.

[0020] In one possible implementation, such as Figure 2 As shown, step S200 further includes: Step S210: Perform hierarchical master tag spatial grid encoding on the original geometric resource set to obtain a spatial encoding set, and then construct a quadtree spatial grid index to form the master spatial index library.

[0021] Step S220: Mark the original geometric resource set with multi-dimensional auxiliary tag attributes to obtain multiple attribute tag sets, and then construct multiple attribute dimension index sub-libraries through inverted index processing.

[0022] Step S230: Based on the multiple attribute constraint sub-conditions of the multiple attribute tag sets, after configuring the tag mapping of the multiple attribute dimension index sub-libraries, integrate the multiple attribute dimension index sub-libraries to obtain the auxiliary attribute index library.

[0023] Specifically, the original geometric resource set is divided and encoded into a main label spatial grid according to the spatial distribution hierarchy. Each geometric resource object is assigned a hierarchical grid code that matches its spatial location, thus forming a spatial code set covering all geometric resources. Then, using the spatial code as the index key, the spatial region is divided from top to bottom according to the spatial coordinates and hierarchical relationship of the geometric resources, and a quadtree hierarchical index structure is constructed. Each geometric resource is assigned to the corresponding quadtree node according to its spatial code, thus completing the construction of the quadtree spatial grid index. Finally, a main spatial index library that can support fast spatial positioning and query is formed.

[0024] The original geometric resource set is marked with multi-dimensional auxiliary tags. Each geometric resource object is added with corresponding attribute tags according to multiple attribute dimensions such as geometry type, size, style, and rendering priority, forming multiple attribute tag sets corresponding to different attribute dimensions. Inverted index processing is performed on each attribute tag set. Using the attribute tag as the key and the geometric resource identifier with the tag as the value, a reverse mapping relationship from attribute tag to geometric resource is established. Based on this, multiple attribute dimension index sub-libraries corresponding one-to-one with each attribute dimension are constructed.

[0025] Based on multiple attribute constraint sub-conditions corresponding to multiple attribute tag sets, establish tag association mapping and query condition matching configuration between each attribute dimension index sub-library, clarify the combination constraint relationship and linkage retrieval rules between different attribute dimensions, and then uniformly encapsulate, associate and integrate, and uniformly schedule the multiple attribute dimension index sub-libraries that have completed tag mapping configuration to form a unified auxiliary attribute index library that supports multi-attribute parallel retrieval and cross-validation.

[0026] In one possible implementation, step S100 further includes: Step S110: Perform syntactic and semantic parsing on the original query request to obtain a standardized query expression.

[0027] Step S120: Perform structured parsing on the standardized query expression to extract spatial constraints and attribute constraints.

[0028] Step S130: Calculate the auxiliary index computing power weight guided by energy efficiency for the attribute constraints to obtain the auxiliary attribute index computing power allocation weight.

[0029] Step S140: Based on the spatial constraints and attribute constraints, perform query intent parsing to determine the query intent pattern.

[0030] Step S150: Calculate the computing power allocation weight of the main space index library based on the query intent pattern, wherein the computing power allocation weight of the auxiliary attribute index and the computing power allocation weight of the main space index library constitute the initial computing power scheduling strategy.

[0031] Specifically, the system sequentially performs lexical analysis, syntax verification, and semantic normalization on the user's original query request. First, the request text is segmented, symbol recognition is performed, and illegal characters are filtered. Then, the validity of the statement structure is verified according to the preset geometric query syntax rules, and key semantic units such as spatial range, attribute filtering, and query target are identified. Finally, requests in different forms are uniformly converted into structured and standardized query expressions that the system can directly parse, thus achieving accurate conversion from original requests to standard query instructions.

[0032] The generated standardized query expression is subjected to structured parsing, including field decomposition, logical partitioning, and constraint identification. First, the expression is decomposed into basic units such as query objects, range limits, logical operators, and filtering conditions according to the query syntax tree structure. Then, through keyword matching and dimension classification, the constraints related to spatial location and regional range are extracted as spatial constraints, and the constraints related to non-spatial attributes such as geometric type, size, style, and priority are extracted as attribute constraints, thus completing the separation and extraction of the two types of core constraints.

[0033] Based on the P attribute constraint sub-conditions contained in the attribute constraint conditions, the corresponding P attribute dimension index sub-databases are accurately mapped and located in the auxiliary attribute index database. Then, the historical performance data of the P attribute dimension index sub-databases are retrieved locally, including query time, CPU utilization, retrieval hit rate, and data return volume. Each data point is linearly normalized to obtain a single performance score. An energy efficiency weighted calculation model is constructed with low power consumption and high retrieval efficiency as optimization objectives. Query time and CPU utilization are used as energy consumption penalty terms, and retrieval hit rate and data return volume are used as efficiency gain terms. Preset weighting coefficients are set for the four scores, and the scores are summed to obtain the comprehensive energy efficiency score of each attribute dimension index sub-database. The comprehensive energy efficiency scores of all sub-databases are then normalized to convert the score values ​​into relative weight values ​​in the range of 0 to 1. Finally, the auxiliary attribute index computing power allocation weight suitable for parallel retrieval of multiple sub-databases is calculated.

[0034] Based on the extracted spatial and attribute constraints, the query intent is analyzed by pattern recognition. First, keyword matching and logical judgment are used to identify whether the query is mainly spatial range retrieval or multi-attribute filtering. Then, it is determined whether it is a joint spatial and attribute retrieval. Thus, the query is divided into three standard query intent patterns: spatial-dominant, attribute-dominant, and spatiotemporal attribute joint. Finally, the core retrieval direction and resource call priority of the current query are clarified, and the query intent pattern is accurately determined.

[0035] Based on the determined query intent pattern, the computing power allocation weight is calculated for the main spatial index: when the query is space-dominant, the computing power ratio of the main spatial index is increased according to preset rules; when the query is attribute-dominant, the computing power ratio of the main spatial index is decreased according to preset rules; when the query is a combination of spatial and attribute types, the computing power weight of the main spatial index is allocated in a balanced proportion; then the computing power allocation weight of the main spatial index calculated above is combined and normalized with the computing power allocation weight of the auxiliary attribute index to form a complete and directly executable computing power allocation ratio. Finally, the two types of weights together constitute the initial computing power scheduling strategy.

[0036] In one possible implementation, step S300 further includes: Step S310: Map the spatial constraints and P attribute constraint sub-conditions to the main index query task and the P auxiliary index query tasks.

[0037] Step S320: Allocate computing resources for the main index query task and P auxiliary index query tasks according to the performance-aware computing power allocation strategy, and perform index retrieval in parallel in the main spatial index library and the P attribute dimension index sub-libraries.

[0038] During the parallel index retrieval process, the main spatial index library and the P attribute dimension index sub-libraries send incremental result heartbeats to the retrieval scheduling platform based on a preset time window.

[0039] Specifically, the spatial constraints extracted through structured parsing are mapped one-to-one to the main index query tasks for retrieval in the main spatial index library. At the same time, the extracted P attribute constraint sub-conditions are mapped independently to form P auxiliary index query tasks that correspond one-to-one with the index sub-libraries of each attribute dimension, thus completing the standardized transformation from query conditions to executable retrieval tasks.

[0040] Read the computing power weight values ​​in the performance-aware computing power allocation strategy, and allocate hardware computing resources such as computing kernels and memory to the main index query task and P auxiliary index query tasks according to the weight ratio. Bind them to the corresponding main space index library and each attribute dimension index sub-library, and simultaneously issue retrieval instructions to drive each index library to carry out index retrieval operations based on the allocated resources.

[0041] During the retrieval operation, the main spatial index library and the sub-index libraries of each attribute dimension periodically summarize the current candidate retrieval data, operation progress and resource consumption information at predetermined fixed time intervals, encapsulate them into incremental result heartbeat messages, and uniformly upload them to the retrieval scheduling platform to complete the status feedback.

[0042] In one possible implementation, step S320 further includes: The incremental result heartbeat includes the candidate set bitmap, retrieval progress indicators, and resource utilization efficiency.

[0043] Specifically, the incremental result heartbeat contains three types of core data: the candidate set bitmap is a distribution map of the geometric resource identifier set initially screened in the current retrieval stage, used to intuitively lock the range of candidate resources; the retrieval progress indicator statistics show the completed retrieval volume and the remaining scale to be retrieved, reflecting the overall progress of the query; and the resource utilization efficiency records the operating parameters such as computing power usage and memory consumption during the retrieval process, reflecting the status of the index database hardware resource consumption.

[0044] In one possible implementation, step S400 further includes: S1: The retrieval scheduling platform receives incremental result heartbeats from the main space index library and P auxiliary attribute index libraries, performs retrieval status optimization, and obtains the intersection candidate set bitmap and auxiliary attribute index computing power correction weight.

[0045] S2: Feed back the intersection candidate set bitmap to the main space index library and P auxiliary attribute index libraries.

[0046] S3: Real-time computing power rescheduling of the P auxiliary attribute index libraries is performed based on the auxiliary attribute index computing power correction weight.

[0047] S4: Drive the main spatial index library and P attribute dimension index sub-libraries, using the intersection candidate set bitmap as the retrieval range constraint, and execute the main index query task and P auxiliary index query tasks in parallel.

[0048] Iteration steps S1 to S4 are performed to progressively narrow the search range until the target geometric resource is located.

[0049] Specifically, the retrieval scheduling platform receives incremental result heartbeats reported by all indexes, breaks down and extracts bitmap data for each candidate set, with each bit in the bitmap corresponding to a unique geometric resource, and the bit value indicating whether the resource meets the current retrieval constraints. Multiple candidate set bitmaps are matched one-to-one according to their positions, and a logical AND operation is performed bit by bit. Only when the bit values ​​at the same position in all bitmaps are in a valid matching state is the geometric resource determined to meet all retrieval constraints and retained in the candidate range; other resources that do not meet the conditions are directly eliminated, thus generating an intersection candidate set bitmap. Simultaneously, the actual operating parameters of each index are compared with the initial computing power standard, the deviation value is calculated, and the computing power allocation is adjusted to obtain the auxiliary attribute index computing power adjustment weight, completing the retrieval state optimization.

[0050] The bitmap of the intersection candidate set obtained by the operation is uniformly packaged into a standardized data format. Relying on the dedicated data interaction link inside the system, according to the corresponding relationship of the preset index library, the bitmap data is first sent back and stored in the designated cache area of ​​the main space index library, and then distributed and pushed to P auxiliary attribute index libraries in sequence according to the sequence number. After receiving the data, each index library completes the verification and parsing and caches it locally, providing data support for subsequent joint retrieval and filtering.

[0051] First, the system reads the determined auxiliary attribute index computing power correction weights, which include the computing power allocation ratio for each of the P auxiliary attribute index libraries. Then, it retrieves the system's current total available computing power resources, including CPU core utilization, GPU computing unit idle capacity, and available memory bandwidth. Following the calculation rule that the computing power allocated to a single library = total computing power resources × corresponding correction weight ratio, the target computing power allocation for each of the P auxiliary attribute index libraries is calculated. Next, the computing power scheduling engine is activated, dynamically adjusting the computing resources of each auxiliary attribute index library through the process scheduling interface: for index libraries with increased computing power ratios, the number of dedicated computing threads is increased, memory cache space is expanded, and GPU computing power is improved. The system prioritizes core resource allocation and optimizes I / O read / write channel bandwidth. For indexes with reduced computing power, redundant threads are systematically reduced and some non-core memory resources are released to ensure that computing power recovery does not affect the continuity of current retrieval tasks. During scheduling, the computing power reception status and resource adaptation of each index are monitored in real time. A heartbeat detection mechanism is used to confirm whether the computing power configuration has taken effect. If there are resource allocation conflicts, such as multiple indexes requesting GPU cores at the same time, arbitration is conducted based on the adjusted weight priority to prioritize the computing power needs of the high-weight indexes. Finally, the real-time computing power rescheduling of P auxiliary attribute indexes is completed, so that the computing power of each index is accurately matched with the adjusted weight, ensuring the efficient execution of subsequent retrieval tasks.

[0052] The scheduling instructions are issued to drive the main spatial index library and P attribute dimension index sub-libraries to run. The intersection candidate set bitmap limits the boundaries of the legal search objects. Each index library uses this as the filtering basis to synchronously and in parallel process the main index query task of the spatial dimension, as well as the P auxiliary index query tasks corresponding to different attribute conditions.

[0053] The entire process from S1 to S4 is executed repeatedly. Each calculation relies on the intersection of bitmaps to continuously compress the search candidate range, dynamically adjusts the computing power allocation ratio, continuously narrows the search interval, and iterates and compares repeatedly until a precise match is found and the unique target geometric resource is locked.

[0054] In one possible implementation, step S400 further includes: Using the performance-aware computing power allocation strategy as the basis for confidence allocation, multi-path bitmap fast intersection is performed on P+1 candidate bitmaps to obtain the intersection candidate bitmap.

[0055] The retrieval performance is evaluated based on the P retrieval progress indicators and P resource utilization efficiency of the P attribute dimension index sub-libraries. The real-time performance scores of the P auxiliary indexes are quantified, and the computing power allocation weights of the auxiliary attribute indexes are dynamically redistributed to obtain the computing power correction weights of the auxiliary attribute indexes.

[0056] Specifically, using the performance-aware computing power allocation strategy as the criterion, one main space candidate set bitmap and P attribute candidate set bitmaps are combined into P+1 groups of bitmaps to be processed. All bitmaps are aligned sequentially according to bit order, and logical AND operations are performed on the bit data at the same position in batches. Only when all bits of the same position in all bitmaps are valid identifiers is the resource at that position retained as valid; all other states are uniformly set to invalid. After all bit operations are completed in batches, the final bitmaps are merged to generate the filtered intersection candidate set bitmaps.

[0057] Extract P retrieval progress indicators reported by P attribute dimension index sub-databases, including the number of completed retrieval entries, retrieval time, and remaining retrieval volume, and P resource utilization efficiency indicators, including CPU utilization, memory utilization, and IO read / write speed. Based on a preset efficiency quantification model, standardize each set of data: convert the retrieval progress indicators into progress scores of 0-100 points according to the actual completion percentage / preset progress threshold, and convert resource utilization efficiency into efficiency scores of 0-100 points according to the actual resource utilization rate / optimal resource utilization rate range. Then, use a weighted summation formula: real-time performance score = progress score × 0.6 + efficiency score × 0.4. The weights can be adjusted according to the business scenario. Calculate the real-time performance score for each of the P auxiliary indexes. Subsequently, based on the real-time performance scores of P auxiliary indexes, a dynamic reallocation of computing power allocation weights is performed: first, the sum of the P scores is calculated, and then the score ratio of each auxiliary index (single index score ÷ total score) is calculated. This ratio is used as the computing power allocation coefficient. Combined with the total available computing power resources of the system, the computing power allocation ratio of each auxiliary attribute index is recalculated according to the calculation method of single index correction weight = computing power allocation coefficient × total computing power quota. At the same time, abnormal fluctuation data is removed. For example, the allocation ratio of indexes with scores below the threshold is appropriately reduced. Finally, P auxiliary attribute index computing power correction weights that match the real-time performance are generated, and the dynamic update of the weights is completed.

[0058] In one possible implementation, step S130 further includes: Step S131: Based on the P attribute constraint sub-conditions in the attribute constraint conditions, map and locate the P attribute dimension index sub-databases in the auxiliary attribute index database.

[0059] Step S132: Locally backtrack the P historical query performance data of the P attribute dimension index sub-libraries, perform auxiliary computing power pre-allocation calculation, and obtain the auxiliary attribute index computing power allocation weight.

[0060] Specifically, the complete attribute constraints are first broken down into P independent attribute constraint sub-conditions. The system then retrieves the pre-configured attribute-index binding mapping table, compares the field types and filtering rules of each attribute constraint sub-condition one by one, and matches and locates the corresponding storage partitions based on the corresponding relationships, ultimately accurately finding the P corresponding attribute dimension index sub-databases.

[0061] The system locally retrieves and backtracks the historical query performance data corresponding to each of the P attribute dimension index sub-databases. This historical data includes core indicators such as the average retrieval response time, number of entries processed per unit of computing power, peak resource utilization, and retrieval success rate for each sub-database in handling queries with similar attribute constraints. Then, each set of historical performance data is standardized and normalized to eliminate differences in the dimensions of different indicators. Next, a weighted summation formula is used, with weight coefficients set according to the business's priority for retrieval speed and resource cost, to calculate the comprehensive historical performance score for each attribute dimension index sub-database. The total comprehensive historical performance scores of the P sub-databases are then calculated, and the proportion of each sub-database's score to the total score is calculated, serving as the basic allocation coefficient. Finally, based on the system's current total available computing power resources, and following the calculation rule of "computing power allocation weight for a single sub-database = basic allocation coefficient × total computing power," the computing power allocation weights corresponding to each of the P auxiliary attribute indexes are calculated one by one, completing the auxiliary computing power pre-allocation calculation.

[0062] In one possible implementation, step S200 further includes: Step S240: Locally retrieve the main space index library and the P auxiliary attribute library level real-time loads of the main space index library and the P attribute dimension index sub-libraries.

[0063] Step S250: Based on the real-time power consumption status of the mobile terminal, perform power-aware load margin calculations on the main space library-level real-time load and the P auxiliary attribute library-level real-time loads to obtain the main space computing power margin and the P auxiliary index computing power margins.

[0064] Step S260: Based on the main space computing power margin and the P auxiliary index computing power margins, perform flexible adjustment of the auxiliary attribute index computing power allocation weight and the main space index library computing power allocation weight, and output the performance-aware computing power allocation strategy.

[0065] Specifically, relying on the local data reading interface, the main space index library and P attribute dimension index sub-libraries are accessed respectively to collect the running status parameters of each database in real time, and the real-time load data at the main space library level is summarized. At the same time, the real-time load data at the auxiliary attribute library level corresponding to each attribute dimension index sub-library is obtained one by one, and the local retrieval and collection of all load data is completed.

[0066] Using the real-time power consumption of the mobile device as a constraint, we first set the rated load limit and power consumption conversion factor for each index library. We then use the formula: Computing power margin = Rated load limit - Real-time load - Real-time power consumption ÷ Conversion factor. We substitute the corresponding parameters of the real-time load of the main space library level and the real-time load of each auxiliary attribute library level into the calculation to calculate the computing power margin of the main space and the computing power margin of P auxiliary indexes one by one.

[0067] By comparing the computing power margin of the main space with the computing power margin values ​​of P auxiliary indexes, the higher the margin, the more computing power resources can be allocated. The weight ratio is recalculated according to the margin ratio rule, the allocation weight of the surplus computing power library is increased, the weight ratio of the computing power shortage library is reduced, the computing power allocation weight of the main space index library and all auxiliary attribute index libraries is simultaneously corrected, all adjusted weight parameters are integrated, and finally a complete performance-aware computing power allocation strategy is generated and output.

[0068] In one possible implementation, step S400 further includes: After scheduling rendering resources based on the target geometric resources, graphics rendering calculations are performed to obtain the target rendering output.

[0069] Specifically, based on the located target geometric resources, its core attributes are first analyzed, including vertex coordinates, texture mapping information, material parameters, and lighting response characteristics. Combined with preset rendering quality standards, such as resolution, frame rate requirements, and anti-aliasing levels, rendering resources are scheduled: the local GPU rendering core and video memory resources are called, the corresponding rendering thread pool is allocated, the adapted shader program, namely the vertex shader and fragment shader, is loaded, and the preprocessed data of the target geometric resources, such as vertex cache and index cache, is cached to reduce redundant calculations. At the same time, the rendering batch is dynamically adjusted according to the complexity of the resources, and complex geometric resources are broken down into reasonable rendering primitives, such as triangles and quadrilaterals. The subsequent graphics rendering calculations are as follows: First, the vertex shader transforms the vertex data of the target geometric resource into screen coordinates, while simultaneously calculating vertex lighting intensity. Second, primitive assembly is performed, combining the processed vertices into preset rendering primitives according to the index cache order, and performing a clipping test to remove primitives outside the screen. Third, the fragment shader performs texture sampling, material shading, and lighting overlay calculations on the pixels of each primitive, while performing optimization processing such as depth testing and alpha blending to ensure the correct pixel rendering hierarchy. Fourth, post-processing is performed on the rendered pixel data, such as color correction, anti-aliasing, and shadow enhancement, and finally, the processed pixel matrix is ​​output as the target rendering result that meets the display requirements.

[0070] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Specific embodiments of this specification have been described above. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0071] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0072] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for scheduling mobile terminal 2D geometry library resources based on spatial grid indexing, characterized in that, The method includes: Receive and deconstruct the raw query requests input by real-time users, perform query intent parsing, and generate an initial computing power scheduling strategy; Based on the real-time load status of the main space index library and the auxiliary attribute index library, combined with the real-time power consumption status of the mobile terminal, the initial computing power scheduling strategy is dynamically constrained and adjusted, and an efficiency-aware computing power allocation strategy is output. In this process, the original geometric resource set is indexed, pre-encoded, and marked to construct the main space index library and the auxiliary attribute index library. Specifically, based on the real-time load status of the main space index library and the auxiliary attribute index library, combined with the real-time power consumption status of the mobile device, the initial computing power scheduling strategy is dynamically adjusted to output an efficiency-aware computing power allocation strategy, including: Locally retrieve the main spatial index library and P attribute dimension index sub-libraries. The real-time load at the main spatial library level and the real-time load at the P auxiliary attribute library level are also retrieved. Based on the real-time power consumption status of the mobile terminal, the load margin calculation of the main space library-level real-time load and the P auxiliary attribute library-level real-time load is performed to obtain the main space computing power margin and the P auxiliary index computing power margin. The computing power margin is expressed as: computing power margin = rated load limit - real-time load - real-time power consumption ÷ conversion factor. Based on the main space computing power margin and the computing power margin of P auxiliary indexes, the computing power allocation weights of the auxiliary attribute indexes and the main space index library are flexibly adjusted, and the performance-aware computing power allocation strategy is output. Driven by the performance-aware computing power allocation strategy, the main space index library and the auxiliary attribute index library are scheduled to perform parallel index retrieval based on the original query request; During parallel index retrieval, a progressive retrieval range convergence is achieved through multi-level result filtering under multi-threaded index cross-validation to locate the target geometric resource.

2. The mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index as described in claim 1, characterized in that, The method involves indexing and precoding the original geometric resource set, and constructing a main space index library and an auxiliary attribute index library. The original geometric resource set is subjected to hierarchical master tag spatial grid encoding to obtain a spatial encoding set. Then, a quadtree spatial grid index is constructed to form the master spatial index library. After marking the original geometric resource set with multi-dimensional auxiliary tag attributes to obtain multiple attribute tag sets, multiple attribute dimension index sub-libraries are constructed through inverted index processing. Based on the multiple attribute constraint sub-conditions of the multiple attribute tag sets, after configuring the tag mapping of the multiple attribute dimension index sub-libraries, the multiple attribute dimension index sub-libraries are integrated to obtain the auxiliary attribute index library.

3. The mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index as described in claim 2, characterized in that, The method includes receiving and deconstructing raw query requests input by real-time users, performing query intent parsing, and generating an initial computing power scheduling strategy. The original query request is parsed using syntax and semantics to obtain a standardized query expression; The standardized query expression is parsed in a structured manner to extract spatial constraints and attribute constraints; The auxiliary index computing power weight is calculated based on the energy efficiency guided by the attribute constraints to obtain the auxiliary attribute index computing power allocation weight. Based on the spatial and attribute constraints, query intent is parsed to determine the query intent pattern. The computing power allocation weight of the main space index is calculated based on the query intent pattern, wherein the computing power allocation weight of the auxiliary attribute index and the computing power allocation weight of the main space index constitute the initial computing power scheduling strategy.

4. The mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index as described in claim 3, characterized in that, Driven by the performance-aware computing power allocation strategy, the main spatial index library and the auxiliary attribute index library are scheduled to perform parallel index retrieval based on the original query request. The method includes: The spatial constraints and P attribute constraint sub-conditions are mapped to the main index query task and the P auxiliary index query task. The computing resources of the main index query task and P auxiliary index query tasks are allocated according to the performance-aware computing power allocation strategy, and the index retrieval is performed in parallel in the main spatial index library and the P attribute dimension index sub-libraries; During the parallel index retrieval process, the main spatial index library and the P attribute dimension index sub-libraries send incremental result heartbeats to the retrieval scheduling platform based on a preset time window.

5. The mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index as described in claim 4, characterized in that, The incremental result heartbeat includes the candidate set bitmap, retrieval progress indicators, and resource utilization efficiency.

6. The mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index as described in claim 5, characterized in that, In the parallel index retrieval process, a progressive retrieval range convergence is achieved through multi-level result filtering under multi-threaded index cross-validation to locate the target geometric resource. The method includes: S1: The retrieval scheduling platform receives incremental result heartbeats from the main space index library and P auxiliary attribute index libraries, performs retrieval status optimization, and obtains the intersection candidate set bitmap and auxiliary attribute index computing power correction weight. S2: Feed back the intersection candidate set bitmap to the main space index and P auxiliary attribute indexes; S3: Real-time computing power rescheduling of the P auxiliary attribute index libraries is performed based on the auxiliary attribute index computing power correction weight. S4: Drive the main spatial index library and P attribute dimension index sub-libraries, using the intersection candidate set bitmap as the retrieval range constraint, and execute the main index query task and P auxiliary index query tasks in parallel; Iteration steps S1 to S4 are performed to progressively narrow the search range until the target geometric resource is located.

7. The mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index as described in claim 6, characterized in that, The retrieval scheduling platform receives incremental result heartbeats from the main spatial index and P auxiliary attribute indexes, performs retrieval status optimization, and obtains an intersection candidate set bitmap and auxiliary attribute index computing power correction weights. The method includes: Using the performance-aware computing power allocation strategy as the basis for confidence allocation, multi-way bitmap fast intersection is performed on P+1 candidate bitmaps to obtain the intersection candidate bitmap. The retrieval performance is evaluated based on the P retrieval progress indicators and P resource utilization efficiency of the P attribute dimension index sub-libraries. The real-time performance scores of the P auxiliary indexes are quantified, and the computing power allocation weights of the auxiliary attribute indexes are dynamically redistributed to obtain the computing power correction weights of the auxiliary attribute indexes.

8. The mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index as described in claim 3, characterized in that, The method involves calculating the auxiliary index computing power weights based on energy efficiency guidance for the attribute constraints, thereby obtaining the auxiliary attribute index computing power allocation weights. Based on the P attribute constraint sub-conditions in the attribute constraint conditions, the P attribute dimension index sub-databases are mapped and located in the auxiliary attribute index database; Locally backtrack the P historical query performance data of the P attribute dimension index sub-databases, perform auxiliary computing power pre-allocation calculation, and obtain the auxiliary attribute index computing power allocation weight.

9. The mobile terminal two-dimensional geometry library resource scheduling method based on spatial grid index as described in claim 1, characterized in that, After scheduling rendering resources based on the target geometric resources, graphics rendering calculations are performed to obtain the target rendering output.

Images