Efficient Index Construction Method for Road Networks

By using an edge-order-based index building method and parallel algorithms, the index building of multi-label road networks is optimized, solving the problem of low efficiency in existing technologies and achieving efficient index building and fast shortest path query.

CN118093589BActive Publication Date: 2026-11-17NANJING UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410243519.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-03-04
Publication Date
2026-11-17
Estimated Expiration
2044-03-04

AI Technical Summary

Technical Problem

Existing CH and CHLR algorithms cannot efficiently solve the shortest path problem in multi-label road networks under the high-speed data updates and large data volumes of the big data era, resulting in excessively long index building times and low efficiency.

Method used

An edge-order-based index construction method is adopted. By traversing the edges of the multi-label road network model one by one, the concept of fully dependent edges is used to optimize the index construction process. Parallel and batch parallel algorithms are introduced to reduce redundant indexes and avoid unnecessary shortest path queries.

Benefits of technology

It significantly improves the efficiency of index building, shortens the building time, enhances query efficiency in multi-label road networks, and can quickly find the shortest path.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118093589B_ABST
    Figure CN118093589B_ABST
Patent Text Reader

Abstract

This invention proposes an efficient indexing method for road networks, comprising: constructing a multi-label road network model, wherein the multi-label road network model uses path intersections as vertices, roads between two intersections as edges, and the distance between two intersections as the edge weight, and each edge is labeled to represent road attributes; numbering all vertices sequentially, and traversing the edges in the multi-label road network model one by one according to the edge weights to construct an index. This invention changes the existing technology's mode of indexing according to the order of points, reduces the number of redundant indexes during the indexing process, and greatly speeds up the indexing time.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of index construction technology in shortest path query in road networks, specifically a method for efficient index construction for road networks. Background Technology

[0002] In recent years, with the development of mobile internet, the Internet of Things, and other technologies, the scale and speed of data acquisition have exploded, making big data-related technologies a hot topic worldwide. In the field of big data analytics, graphs, as an effective abstract data structure for describing data relationships, are playing an increasingly crucial role, attracting close attention from academia, industry, and government. Particularly in the field of road networks, the analysis and processing of graph data has been extensively studied and has been successfully applied to various aspects such as route planning and traffic analysis, providing new research directions and methods for solving practical traffic problems and improving route planning efficiency.

[0003] The shortest path problem is of great significance in road network research. A road network can be represented using a graph model. To represent. In this graph, each vertex This represents a path intersection, and each edge This represents the weight of the edge on the road between two intersections. This is used to represent the distance between two intersections. The core of the shortest path problem is to study how to... (The sentence is incomplete and requires more context to be fully translated.) This involves finding the shortest path from the source vertex to the target vertex. By establishing this graph model, we can effectively describe and analyze shortest path conditions in road networks, providing a foundation for solving practical traffic and navigation problems.

[0004] The shortest path problem is widely used in road networks, primarily in route planning and traffic management and optimization. On one hand, the shortest path problem directly impacts the efficiency and accuracy of route planning services. By determining the shortest or lowest-cost path between two points, drivers can be provided with the most efficient navigation, reducing travel time and costs. On the other hand, the shortest path problem influences the management and optimization strategies of traffic management departments. With the continuous development of urban transportation, expanding traffic demand, and increasingly complex road networks, modern urban traffic management systems cannot function without the shortest path problem. Traffic management departments can more accurately allocate traffic flow by analyzing the shortest paths between nodes in the road network, thereby reducing congestion, improving road utilization efficiency, reducing environmental pollution, lowering traffic accident rates, and saving energy.

[0005] To solve the shortest path problem, numerous index-based algorithms have been proposed. These algorithms pre-calculate path information between vertices and organize them in a reasonable manner, reducing computation and improving efficiency during queries. Among the existing index-based algorithms for solving the shortest path problem, Contraction Hierarchy (CH) is widely used due to its excellent performance in terms of preprocessing time, space cost, and query processing time. However, with the development of big data, the efficiency of CH can no longer meet the current data volume. As early as 2010, Beijing's 67,000 taxis generated more than 1.2 million trip records daily, or 13.8 new data points per second. In recent years, with the popularization of ride-sharing services like Uber and Didi, the number of registered drivers has grown rapidly, and the speed of data addition and update has also increased significantly. Therefore, it is necessary to improve the efficiency of CH to adapt to the larger data volume.

[0006] Given the characteristics of road networks, different paths such as railways and highways may exist between two locations. Therefore, research on shortest paths in multi-label road networks has attracted considerable attention from scholars. Multi-label road networks can be modeled using graph models. To indicate, among which This represents the label set of each edge in a multi-label road network. The core of the shortest path problem in a multi-label road network is to study how to... The goal is to find the shortest path from the source vertex to the target vertex, where the labels of all edges in this path are restricted to label set A. In shortest path research in multi-label road networks, CHLR (Contraction Hierarchy with Label Restrictions) inherits the ideas of CH and applies them to multi-label networks, introducing indexes to improve the efficiency of solving the shortest path problem. However, similar to CH, with the advent of the big data era, CHLR cannot keep up with the ever-increasing data volume and update speed, and cannot efficiently solve the shortest path problem. Therefore, how to efficiently solve the shortest path problem in multi-label road networks has become an important task. Summary of the Invention

[0007] The purpose of this invention is to propose an efficient index construction method for road networks.

[0008] The technical solution to achieve the purpose of this invention is: an efficient index construction method for road networks, the specific steps of which are as follows:

[0009] A multi-label road network model is constructed, wherein the multi-label road network model uses the path intersection as the vertex, the road between two intersections as the edge, and the distance between two intersections as the edge weight. Each edge is labeled, and the label is used to represent the road attribute.

[0010] All vertices are numbered sequentially, and the edges in the multi-label road network model are indexed and constructed one by one according to their weights.

[0011] Preferably, the specific method for constructing the index by traversing the edges in the multi-label road network model one by one according to the edge weight is as follows:

[0012] Step 2.1: Initially, the result graph is set to empty. All edges in the multi-label road network model are placed into the edge set to be processed. The result graph is indexed and constructed to obtain the road network graph.

[0013] Step 2.2: Sort the edges in the set of edges to be processed according to their weights, in ascending order of weight; take the union of the resulting graph and the set of edges to be processed as the global graph;

[0014] Step 2.3: When the edge set to be processed is not empty, determine whether there is an edge at the top of the sorted order in the result graph. If there are no fully dependent edges, proceed to step 2.4; otherwise, proceed to step 2.6.

[0015] Step 2.4: Add the top edge to the result graph and remove it from the set of edges to be processed;

[0016] Step 2.5: Traverse the newly added edges in the resulting graph, prioritizing the vertices. Find the neighbors of the vertices that are sorted later. All vertices sorted after Excluding the other vertex on the added edge Based on the relationship between the edges formed by the found vertices and the newly added edges, add or update the index.

[0017] The traversal is complete; return to step 2.2.

[0018] Step 2.6: If the resulting graph contains a fully dependent edge of the first edge in the sorted graph, do not perform any operation on that edge, delete it directly, and select the next edge in the sorted graph to repeat Step 2.1.

[0019] Preferably, the weight of the index is The edges formed and The sum of the weights of the edges formed, labeled as .

[0020] Preferably, the specific method for adding or updating the index based on the relationship between the edges formed by the found vertices and the newly added edges is as follows:

[0021] Determine the global graph and Are there any tags between them? If there is no edge, then directly on and Insert an index between the edges; and add the index to the set of edges to be processed;

[0022] If in the global graph and There are labels between them An edge whose weight is greater than 1. The edges formed and The sum of the weights of the edges formed is used to update the weight of that edge to the sum of the weights of the two edges, and it is then added to the set of edges to be processed; if the global graph... and There are labels between them An edge whose weight is not greater than 1. The edges formed and The sum of the weights of the edges formed will not be processed.

[0023] Preferably, determine whether the result graph has an edge that is first in the sorted order. The specific method for achieving full dependency edges is as follows:

[0024] Vertices in the search results graph and vertex Does an edge exist between them? , making the edge The set of labels is the edge If a subset of the label set contains an edge that satisfies a certain condition, it means that the resulting graph contains an edge that is at the top of the sorted order. A complete dependence on edges, and vice versa.

[0025] Compared with the prior art, the present invention has the following significant advantages: the present invention changes the existing technology's mode of index construction that must follow point order, reduces the number of redundant indexes in the index construction process, and greatly speeds up the index construction time; it avoids the situation in the prior art where shortest path queries need to be performed continuously during the index construction process, and greatly improves the efficiency of index construction. Attached Figure Description

[0026] Figure 1 This is a flowchart of an efficient index construction method for road networks.

[0027] Figure 2 This is a multi-label road network map.

[0028] Figure 3A diagram illustrating the steps involved in constructing an index on a multi-label road network based on edge order.

[0029] Figure 4 Construct a result graph for a multi-label road network index.

[0030] Figure 5 Experimental dataset.

[0031] Figure 6 The runtime of different algorithms on different multi-label road networks. Detailed Implementation

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0033] The shortest path problem on road networks is a fundamental and meaningful problem. This invention addresses the high computational complexity and construction time of traditional index building schemes in multi-label road network queries. It restructures the traditional point-order-based index building method by adopting an edge-order-based approach, accelerating index building on locally optimal subgraphs. Furthermore, it introduces the concept of fully dependent edges to optimize the index building process. Within this framework, this invention proposes an efficient index building method for road networks, avoiding unnecessary shortest path queries during index building, thus significantly improving the efficiency of index building in multi-label road networks. To further enhance performance, this invention proposes a parallel index building algorithm that processes edges with equal weights in parallel, accelerating index building time. To maximize the utilization of parallel computing resources, we also propose a batch parallel algorithm, which significantly improves the utilization rate of parallel computing resources in each round, effectively shortening the index building time.

[0034] like Figure 1 As shown, an efficient index construction method for road networks includes the following steps:

[0035] Step 1: Construct a multi-label road network model, with path intersections as vertices. The road between two intersections is represented as an edge. The distance between two intersections represents the weight of the edge. In the road network model, each edge is labeled to represent road attributes, such as railway, highway, or ship.

[0036] The constructed multi-label road network model is specifically represented as follows:

[0037] The label set of each edge in a multi-label road network is represented as follows: .

[0038] Set the vertex order in the multi-label road network model;

[0039] Step 2: Build an index by iterating through the edges in the multi-label road network model according to their weights. The specific steps are as follows:

[0040] Step 2.1: Initially, the result graph is set to empty. All edges in the multi-label road network model are placed into the edge set to be processed. The result graph is indexed and constructed to obtain the road network graph.

[0041] Step 2.2: Sort the edges in the set of edges to be processed according to their weights, in ascending order of weight; take the union of the resulting graph and the set of edges to be processed as the global graph;

[0042] Step 2.3: When the edge set to be processed is not empty, determine whether there is an edge at the top of the sorted order in the result graph. If there are no fully dependent edges, proceed to step 2.4; otherwise, proceed to step 2.6.

[0043] The determination of whether a graph is completely dependent on edges is based on the vertices in the search result graph. and vertex Does an edge exist between them? , making the edge The set of labels is the edge If a subset of the label set contains an edge that satisfies a certain condition, it means that the resulting graph contains an edge that is at the top of the sorted order. A complete dependence on edges, and vice versa.

[0044] Step 2.4: Delete the top-ranked edge from the set of edges to be processed, and add the top-ranked edge to the result graph; among the top-ranked edges, The vertex that appears earlier in the sequence. For another vertex;

[0045] Step 2.5: Traverse the result graph Find the order of the neighbors. After that and not vertex ;

[0046] Determine the global graph and Are there any tags between them? If there is no edge, then directly on and Insert an index between them, the weight of which is... The edges formed and The sum of the weights of the edges formed, labeled as And add the index to the set of edges to be processed;

[0047] If in the global graph and There are labels between them An edge whose weight is greater than 1. The edges formed and The sum of the weights of the edges formed is used to update the weight of that edge to the sum of the weights of the two edges, and it is then added to the set of edges to be processed; if the global graph... and There are labels between them An edge whose weight is not greater than 1. The edges formed and The sum of the weights of the edges formed will not be processed.

[0048] The traversal is complete; return to step 2.2.

[0049] Step 2.6: Do not perform any operation on this edge, delete it directly, select the next edge to sort, and repeat Step 2.2.

[0050] In the shortest path query process, using an index can help the algorithm find the starting point more quickly. To the finish line The shortest path index can be understood as a hierarchical structure compressed from the original graph by pre-computing and storing key path information. This allows the shortest path algorithm to skip edges and nodes that cannot contain the shortest path during a query, significantly improving query efficiency. Specifically, when the shortest path algorithm starts searching from the starting point, it first uses the index to locate the level that may contain the shortest path, and then performs a depth-first search or breadth-first search within that level. The index constructed using this invention can quickly find the shortest path from the starting point to the ending point, avoiding ineffective searches across the entire graph.

[0051] Example: Figure 2 Taking the multi-label road network shown as an example, assuming that the vertex number is its vertex order, the index construction steps based on the edge order are as follows: Figure 3 As shown. Figure 3 The gray squares in the first row represent the final index construction result graph. Each row in the graph represents the edge put into the result graph in each round. The gray part in each row is the valid index generated in that row. The index with a white background is the case where there is a fully dependent edge, and this index is not processed.

[0052] The algorithm first shrinks each edge with a weight of 1, at which point no index is constructed. Then, it shrinks the edges... After that, index It will be constructed because of the vertices in the global graph. and If there are no connected edges, the index will be directly inserted into the set of edges to be processed and participate in the sorting. Note that in Figure 3 In the middle, the shrinking edge At that time, index It will still be constructed and added to the set of edges to be processed for sorting. However, when the algorithm performs the edge shrinking operation on the edge, it will determine that there are fully dependent edges in the result graph and will not process it.

[0053] In this example, the index constructed in this way results in the following: Figure 4 As shown.

[0054] To clearly and objectively describe the effects of the invention's technical innovations, this invention compares the index construction time of different algorithms. The experimental subjects are the traditional index construction algorithm ("CHLRBaseline") and the index construction method on multi-label road networks based on edge order ("ECHLR"). The experimental data consists of four real multi-label road networks; specific data details are as follows... Figure 5 As shown. The maximum acceptable runtime is set to 12 hours, denoted as "INF".

[0055] In this experiment, the index creation time of the two algorithms was compared on four real-world road network datasets, and the results were summarized in... Figure 6 middle.

[0056] As Figure 6 As shown, with the same 8 threads, the algorithm of this invention takes less time than the traditional index building algorithm, and the efficiency of index building on real multi-label road networks is greatly improved.

[0057] For example, in a DC multi-label road network, the construction times of the two algorithms are 192 seconds and 14 seconds, respectively.

[0058] In summary, the algorithm proposed in this invention demonstrates a significant performance improvement, greatly reduces index construction time, and achieves efficient index construction on multi-label road networks.

[0059] The scope of protection of this invention is not limited to the above embodiments. Any variations and advantages that can be conceived by those skilled in the art without departing from the spirit and scope of the inventive concept are included in this invention and are protected by the appended claims.

Claims

1. A method for constructing an efficient index for a road network, characterized in that, The specific steps are as follows: A multi-label road network model is constructed, wherein the multi-label road network model uses the path intersection as the vertex, the road between two intersections as the edge, and the distance between two intersections as the edge weight. Each edge is labeled, and the label is used to represent the road attribute. All vertices are numbered sequentially, and the edges in the multi-label road network model are indexed and constructed one by one according to their weights. The specific method is as follows: Step 2.1: Initially, the result graph is set to empty. All edges in the multi-label road network model are placed into the edge set to be processed. The result graph is indexed and constructed to obtain the road network graph. Step 2.2: Sort the edges in the set of edges to be processed according to their weights, in ascending order of weight; take the union of the resulting graph and the set of edges to be processed as the global graph; Step 2.3: When the edge set to be processed is not empty, determine whether there is an edge with the highest order in the result graph completely dependent edge, if not, go to Step 2.4, otherwise go to Step 2.6; Step 2.4: Add the top edge to the result graph and remove it from the set of edges to be processed; Step 2.5: Traverse the newly added edges in the resulting graph, prioritizing the vertices. Find the neighbors of the vertices that are sorted later. All vertices sorted after Excluding the other vertex on the added edge Based on the relationship between the edges formed by the found vertices and the newly added edges, add or update the index. The traversal is complete; return to step 2.

2. Step 2.6: If the resulting graph contains a fully dependent edge of the first edge in the sorted graph, do not perform any operation on that edge, delete it directly, and select the next edge in the sorted graph to repeat Step 2.

1.

2. The efficient index construction method for road networks according to claim 1, characterized in that, The weight of the index is The edges formed and The sum of the weights of the edges formed, labeled as .

3. The efficient index construction method for road networks according to claim 1, characterized in that, Based on the relationship between the edges formed by the found vertices and the newly added edge, the specific method for adding or updating the index is as follows: Determine the global graph and Are there any tags between them? If there is no edge, then directly on and Insert an index between the edges; and add the index to the set of edges to be processed; If in the global graph and There are labels between them An edge whose weight is greater than 1. The edges formed and The sum of the weights of the edges formed is used to update the weight of that edge to the sum of the weights of the two edges, and it is then added to the set of edges to be processed; if the global graph... and There are labels between them An edge whose weight is not greater than 1. The edges formed and The sum of the weights of the edges formed will not be processed.

4. The efficient index construction method for road networks according to claim 1, characterized in that, Determine if the result graph contains the first edge in the sorted order. The specific method for achieving full dependency edges is as follows: Vertices in the search results graph and vertex Does an edge exist between them? , making the edge The set of labels is the edge If a subset of the label set contains an edge that satisfies a certain condition, it means that the resulting graph contains an edge that is at the top of the sorted order. A complete dependence on edges, and vice versa.