CN110487293A - A kind of efficient and privacy paths planning method based on extensive road network - Google Patents
A kind of efficient and privacy paths planning method based on extensive road network Download PDFInfo
- Publication number
- CN110487293A CN110487293A CN201910805967.XA CN201910805967A CN110487293A CN 110487293 A CN110487293 A CN 110487293A CN 201910805967 A CN201910805967 A CN 201910805967A CN 110487293 A CN110487293 A CN 110487293A
- Authority
- CN
- China
- Prior art keywords
- vertex
- outsourcing
- point
- path
- road network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/34—Route searching; Route guidance
- G01C21/3446—Details of route searching algorithms, e.g. Dijkstra, A*, arc-flags, using precalculated routes
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/34—Route searching; Route guidance
- G01C21/3453—Special cost functions, i.e. other than distance or default speed limit of road segments
- G01C21/3476—Special cost functions, i.e. other than distance or default speed limit of road segments using point of interest [POI] information, e.g. a route passing visible POIs
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/34—Route searching; Route guidance
- G01C21/3453—Special cost functions, i.e. other than distance or default speed limit of road segments
- G01C21/3492—Special cost functions, i.e. other than distance or default speed limit of road segments employing speed data or traffic data, e.g. real-time or historical
Abstract
The invention discloses a kind of efficient and privacy the paths planning method based on extensive road network, include the following steps: static road network figure being divided into multiple subgraphs, construct a balance search tree, the subgraph for balancing search tree interior joint and static road network figure corresponds, and balances the corresponding static road network figure of root node in search tree;The each leaf node balanced in search tree is decomposed into one group of outsourcing figure and a linked, diagram, all outsourcing figures are stored in Cloud Server, all-links figure, balance search tree and real-time road condition information are stored in mist server, wherein, any pair of vertex in each outsourcing figure is not directly connected, and the shortest distance between any pair of vertex in each outsourcing figure is measured not less than privacy;The source point and target point for obtaining user's input, cook up driving path using path query model.Cloud Server shares calculating, mist servers' layout more shortest path, and inquiry is made more efficiently with intelligence, and to guarantee the privacy of road net data and user query.
Description
Technical field
The present invention relates to location based service technical fields, and in particular to a kind of based on the efficient of extensive road network
And the paths planning method of privacy.
Background technique
With wireless communication and mobile positioning technique continuous development, location based service (LBS) have become can not or
Scarce application.A underlying issue of LBS is the path planning (also referred to as shortest path query) on road network, it is certain excellent
Change the optimal path found between source point and target point under target.A kind of balance search tree existing at present indexes G-
Tree, it realizes the efficient shortest path query to large-scale road network, but the Shortcomings in terms of availability and intelligence.It
All calculating in search process are executed dependent on client, consume a large amount of client resource.And it can only handle quiet
State path planning does not account for dynamic real time traffic state, so that the path of planning is not always best.
Cloud computing mode provides a potential solution by appointing Cloud Server processing inquiry.However, this
Mode needs to disclose entire road network, user query information and real time traffic data to Cloud Server, and these data all have
There is very high commercial value, is worth the protection of LBS provider.Therefore, the computing capability for how making full use of Cloud Server is not being let out
Active path planning is provided in the case where dew valuable information, is a challenging problem.
Summary of the invention
The present invention proposes a kind of efficient and privacy paths planning method based on extensive road network, existing to solve
There are problems that valuable information disclosure risk while handling and inquire using Cloud Server in technology.
It specifically adopts the following technical scheme that, a kind of efficient and privacy path planning side based on extensive road network
Method includes the following steps:
Static road network figure is divided into multiple subgraphs, constructs a balance search tree, balance search tree interior joint and static state
The subgraph of road network figure corresponds, and balances the corresponding static road network figure of root node in search tree;
The each leaf node balanced in search tree is decomposed into one group of outsourcing figure and a linked, diagram, all outsourcing figures are deposited
It is stored in Cloud Server, all-links figure, balance search tree and real-time road condition information are stored in mist server, wherein each
Any pair of vertex in outsourcing figure is not directly connected, and the shortest distance between any pair of vertex in each outsourcing figure
It is measured not less than privacy;
The source point and target point for obtaining user's input utilize outsourcing figure stored in cloud server based on path query model
The traveling road from source point to target point is searched for linked, diagram, balance search tree and the real-time road condition information stored in mist server
Diameter.
Static road network figure can be modeled asWhereinOne group of vertex, represent one group of POIs (point of interest,
Can be a shop, a house or a bus station etc.) or crossing, ε is the side of one group of expression road connection,It represents every
The static weight on side.Give one group of vertexConnect vertex viAnd vjSide be defined as e (vi,vj), the side
Static weight be defined as w (vi,vj), as vertex viWith vjBetween Euclidean distance (nonnegative value).By static road network
Figure divides and forms a balance search tree G*-tree, by the leaf node of G*-tree under conditions of not revealing road net data
Outsourcing figure is contracted out to Cloud Server, remaining is stored in mist server, and Cloud Server shares calculating, mist server rule in the program
More shortest path is drawn, so that the inquiry of user more efficiently with intelligence, while is guaranteed the privacy of road net data and user query, avoids
Attack by not trusted Cloud Server.
Further, static road network figure is divided into multiple subgraphs, constructs a balance search tree and specifically includes following step
It is rapid:
It obtains static road network figure and it is divided, static road network figure is divided into multiple sons according to default segmentation degree
Figure, then each subgraph is divided into multiple subgraphs according to the default segmentation degree, the iteration above process is up to number of vertex in subgraph
No more than default node size;Assuming that segmentation degree is f, node size τ, i.e., static road network figure is divided into f equal sizes
Subgraph, then each of which subgraph is divided into f subgraph, process as iteration, until number of vertex does not surpass in the subgraph of division
Cross τ;
Father after being divided according to static road network figure schemes one balance search tree of building corresponding with the relationship between subgraph, balance
The subgraph of search tree interior joint and static road network figure corresponds, and balances the corresponding static road network figure of root node in search tree, balance
Each node possesses a unique identifier, one group of boundary point and a distance matrix in search tree.
The division methods can make each sub- boundary point of graph quantity path planning expense with after reduction as few as possible, often
A node possesses a unique identifier, one group of boundary point and a distance matrix, and later period search efficiency can be improved, and realizes efficient
Path planning inquiry.
Further, the distance matrix that balance search tree interior joint possesses meets following require a: nonleaf node
Distance matrix record the information of boundary point in its child nodes, including between a pair of of boundary point the shortest distance and they whether
Adjacent label;The distance matrix of leaf node records the shortest distance between the boundary of the node and its vertex and indicates them
Whether adjacent label.
Each node includes that the distance matrix for meeting above-mentioned requirements improves most convenient for the calculating of the later period shortest distance
The efficiency that short distance calculates.
Further, each leaf node balanced in search tree is decomposed into one group of outsourcing figure and a linked, diagram is specifically wrapped
Include following steps:
Obtain leaf node and privacy measurement;
The boundary point in leaf node is bridged, for each boundary point pair of the leaf node, if between a boundary point pair most
Short path needs to calculate by the boundary point of other leaf nodes, then is the leaf node when collection adds a bridge joint, and be arranged
The bridge joint while weight be the bridge joint while correspond to the shortest distance between two boundary points;
A vertex is initialized to set, for storing all vertex pair in leaf node, and initializes a Priority Queues
For sky, which is a key assignments by vertex to frequency descending for storing candidate outsourcing vertex pair, the Priority Queues
To structure, wherein key is vertex pair, is worth the frequency for vertex pair;
For vertex to each vertex pair in set, if this opposite vertexes is not directly connected, and this opposite vertexes
Between the shortest distance measured not less than privacy, then enumerate all candidate outsourcing vertex pair of the shortest path between this opposite vertexes;
For each of enumerating candidate outsourcing vertex pair, its frequency in Priority Queues need to be updated, that is, candidate's outsourcing vertex of adding up
Between the frequency in the shortest path this opposite vertexes;Then vertex is repeated to this interior vertex is gathered to therefrom removing
The above process is until vertex is combined into sky to collection;
If Priority Queues is not sky, vertex sequence is constructed based on Priority Queues, and generate an outsourcing figure, by the outsourcing
Figure is added in outsourcing atlas, then generates linked, diagram, the vertex set in the linked, diagram belongs to vertex set and the outsourcing of the leaf node
The union of vertex set in atlas, side connection in the linked, diagram is vertex in the leaf node and outsourcing figure, in the linked, diagram
The weight on side is the shortest distance, and the vertex that then will be calculated by outsourcing figure is removed to from Priority Queues;It repeats the above steps
Until Priority Queues is sky;
Export one group of outsourcing figure and a linked, diagram.
It is enumerated using brute-force to find the method for optimal solution for each leaf node and be decomposed into one group of outsourcing figure and a link
Figure be it is expensive and infeasible, each leaf node is decomposed into one group of outsourcing figure and a linked, diagram by above-mentioned didactic mode,
Cost is controllable and feasible.
Further, vertex sequence is constructed based on Priority Queues, and generates an outsourcing figure and specifically comprises the following steps:
Obtain leaf node, privacy measurement and vertex sequence;
It generates the vertex set of outsourcing figure: if vertex sequence is not sky, first vertex in vertex sequence being selected to be added
It is removed from vertex sequence into the vertex set of outsourcing figure, and by it;Then the remaining vertex in vertex sequence is traversed, if from top
First vertex and remaining a certain vertex removed in point sequence be directly connected to or both between the shortest distance be less than it is hidden
Private measurement, then remove from vertex sequence by the remaining a certain vertex, the remaining a certain vertex be otherwise added to outsourcing
In the vertex set of figure, and it is removed from vertex sequence;It repeats the above process until vertex sequence is sky;
Generate the side collection of outsourcing figure: the vertex pair formed for any two vertex in the vertex set of outsourcing figure, to outsourcing
Figure when concentrating and increasing the vertex to composition, the shortest distance of the weight on the side between two vertex be set;
It is to be not directly connected, and the shortest distance between any pair of vertex is not less than that output one, which meets any pair of vertex,
The outsourcing figure of privacy measurement.
Candidate outsourcing vertex pair: given subgraphIn an opposite vertexes (vs,vt), and δ (vs,vt) >=d,When meeting following condition, vertex pairIt is (vs,vt) a candidate outsourcing vertex pair.Its
In, SP (vs,vt) indicate vsTo vtShortest path, δ (vs,vt) indicate vsTo vtThe shortest distance, d be privacy measurement, e (vs,
vt) indicate vsTo vtSide, εiFor the side collection of subgraph.
δ(vs,vt)=δ (vs,vx)+δ(vx,vy)+δ(vy,vt)
δ(vs,vx) < d or e (vs,vx)∈εi, δ (vy,vt) < d or e (vy,vt)∈εi
δ(vx,vy) >=d and
Any pair of vertex in each outsourcing figure decomposed by the above method is not directly connected, and each outsourcing
The shortest distance between any pair of vertex in figure is measured not less than privacy, therefore even if outsourcing figure is stored in Cloud Server,
But Cloud Server can not still obtain the inquiry data of complete road net data, real-time road data and user.
Further, output one meets any pair of vertex and is not directly connected, and between any pair of vertex most
Short distance further includes following steps before the outsourcing figure not less than privacy measurement: any side is concentrated for the side of outsourcing figure, if outside
There are a certain vertex in the vertex set of packet figure, so that the side of the outsourcing figure concentrates the weight of any side corresponding equal to any side
Two vertex the sum of shortest distance between a certain vertex respectively is then concentrated from the side of outsourcing figure and removes any side.
It is complete graph by the outsourcing figure that above step generates, in order to save the memory space on Cloud Server, needs to delete
Except unnecessary side.
Further, cooking up driving path using path query model includes the basic path only considered under static road network
Planning Model and the intelligent path planning mode for considering real-time road condition information.This method includes two kinds of query patterns, can be according to visitor
The demand at family may be selected only to consider the basic path query under static road network, the intelligence for considering real-time road condition information also may be selected
It can path query.
Further, basic path planning mode specifically includes following process:
Position leaf node belonging to source point and target point;
If source point and target point are on same leaf node, if the side that source point and target point connect and compose belongs to the leaf node
Side collection or source point and target point between the shortest distance less than privacy measure, then source point is calculated using dijkstra's algorithm
To the target point shortest distance and shortest path, otherwise call algorithm MD-Inside-leaf obtain source point to target point most short distance
From with incomplete shortest path;
If source point not on same leaf node, calls algorithm MD-Outside-leaf to obtain source point to mesh to target point
The shortest distance of punctuate and incomplete shortest path;
Algorithm PathRecovery is called to restore to obtain source point to the shortest path of target point, and final output source point is to mesh
The shortest path and the shortest distance of punctuate.
In the method, a part of MD-Inside-Leaf is calculated and gives Cloud Server, and Cloud Server possesses by force
Big computing capability keeps this part calculating ratio prior art more efficient.For MD-Outside-Leaf, optimize G*-tree's
Distance matrix decreases computing cost to a certain extent.So the overall calculation efficiency of this method is better than the prior art.
Further, the algorithm MD-Inside-leaf is comprised the following processes:
Obtain source point and target point, the leaf node where source point and target point, and balance search tree;
A set is initialized, by vertex that each neighbours' point of each neighbours' point of source point and target point is built into adding
Enter in the set;
The set is sent to Cloud Server, Cloud Server calculates each vertex pair in the set using one group of outsourcing figure
The shortest distance, for the vertex pair of not calculated result, default its shortest distance be ∞, Cloud Server return the set
In all vertex to the set of the corresponding shortest distance, wherein ∞ indicates infinitely great;
The shortest distance and incomplete shortest path of the source point to target point is calculated;
The algorithm MD-Outside-leaf is comprised the following processes:
Obtain source point and target point, two leaf nodes where source point and target point difference, and balance search tree;
The public ancestor node of minimum that two leaf nodes are found in balance search tree, finds two in balance search tree
Most short node path between leaf node;
The shortest distance and incomplete shortest path of the source point to target point is calculated;
The algorithm PathRecovery is comprised the following processes:
Obtain the shortest distance and imperfect shortest path of the source point to target point, and balance search tree and static road network
Figure;
Each edge in the imperfect shortest path of circular treatment judges whether the side is a line in static road network figure;
If the side is a line in static road network figure, which is added in complete shortest path, and from imperfect
The side is removed in shortest path;
If the side is not a line in static road network figure, the iteration a certain vertex that each starting point with the side is connected, if
The sum of shortest path of terminal is equal to the side when meeting the weight in starting point to a certain vertex and a certain vertex to this
Starting point to the shortest distance of terminal, then this is added to when starting point and a certain vertex connect and compose complete most short
Path is summarized, by a certain vertex and this while terminal constitute while be added in imperfect shortest path, and from it is imperfect most
The side is removed in short path;
Recycle above-mentioned treatment process to imperfect shortest path for sky, obtain source point to target point complete shortest path.
In MD-Inside-leaf algorithm, mobile subscriber is interacted with Cloud Server to accelerate to inquire, in this process,
Cloud does not know that real inquiry and the track of user;Algorithm MD-Outside-leaf can quick search obtain imperfect shortest path
Diameter and the shortest distance;What algorithm MD-Inside-leaf and algorithm MD-Outside-leaf was obtained is imperfect path, and
We must obtain that complete path is just significant, and algorithm PathRecovery can then restore complete shortest path in practice
Further, intelligent path planning mode comprises the following processes:
Mist server obtains the source point that user inputs and target point and moment;
Mist server is arrived using the source point that the path planning process of basic path planning mode obtains under static road network first
The shortest path of target point;
An intelligent path is initialized, the shortest path under obtained static road network is assigned to intelligent path;
Based on the real-time road condition information inscribed when this, one is constructed by congestion sides all in the shortest path under static road network
The sequence of composition, then adjusts the shortest path between source point and target point with heuristics manner, and all congestion paths have adjusted
Afterwards, source point is obtained to the final intelligent path of target point and path cost and is exported to user.
During adjusting the shortest path between source point and target point with heuristics manner, if two vertex vs on congestion sidex
And vyBelong to the same leaf nodeIn this case, first by removing leaf nodeIn all congestion sides update
The sequence of congestion side composition, and fromIn find first vertex v on shortest path under static road networkfAnd the last one
Vertex vl.Then, it is determined that vlIt whether is target point.If it is, it is only necessary to it is calculated in the node using based on dijkstra's algorithm
vfAnd vtBetween shortest path/distance, and use vfTo vtShortest path adjust v in intelligent pathfTo vtPath.
If vlIt is not representative points, in this case, vlFor the outlet (boundary) of the leaf node, other better outlets are found.It calculatesIn each boundary point b to target point vtThe shortest distance, with δ (b, vt) indicate.In δ (b, vt) it is less than threshold θlAll sides
In boundary's point, selection and vfApart from the smallest boundary point.
If two vertex vs on congestion sidexAnd vyIt is the boundary point of different leaf nodes, i.e. their sides for being respectively two subgraphs
Boundary's point.In this case, it finds and vxConnected other boundary points replace vy.For with vxAdjacent each boundary point b, meter
It is calculated away from target point vtShortest distance δ (b, vt).In δ (b, vt) it is less than threshold θyAll boundary points in, selection and vxIt constitutes
The smallest boundary point of side right weight.
In above-mentioned two situations, if shortest path changes, need to adjust IP (vs,vt) in corresponding path and at
This δ * (vs,vt), and update the sequence of congestion side composition.
Since real-time road data have very high commercial value, these data is protected to be necessary.We were both
It is not desired to store it in insecure Cloud Server, is also not desired to distribute them to user and loses its value.So we use
These data are protected than the mist server of Cloud Server more secure and reliable.The mode is related to mobile subscriber, mist server and cloud
Inquiry is directly entrusted to mist server by the interaction between server, mobile subscriber, certainly, in this case, mist server
Once receiving request, complete shortest path and distance are obtained first, obtain more preferably path then in conjunction with real-time road condition information
And relevant cost.Specifically, MD-Inside-Leaf is inquired, mist server may may require that cloud participates in.
Beneficial effect
The invention proposes a kind of efficient and privacy intelligent path planning scheme, Cloud Server shares meter in the program
It calculates, mist servers' layout more shortest path, the inquiry of user is made more efficiently with intelligence, while to guarantee road net data and user query
Privacy, main thought is that former road network is divided to and formed a balance search tree, under conditions of not revealing road net data
The leaf node for balancing search tree is contracted out to Cloud Server, remaining is stored in mist server, mist server can with privacy with
Cloud interaction improves search efficiency, and intelligently cooks up more preferably path for user.Compared with prior art, have following excellent
Point:
1, high availability: mobile subscriber can make full use of the computing capability of Cloud Server, and the resource for reducing client disappears
Consumption;
2, intelligent Service: mist server can comprehensively consider real-time traffic condition and path distance realizes intelligent path rule
It draws, mobile subscriber is allow to enjoy better service;
3, enhance privacy: protection data-privacy and inquiry privacy avoid the attack by not trusted Cloud Server.
Detailed description of the invention
Fig. 1 is the overall framework figure of paths planning method in the embodiment of the present invention;
Fig. 2 is the division schematic diagram of static road network figure in the embodiment of the present invention;
Fig. 3 is that search tree building schematic diagram is balanced in the embodiment of the present invention;
Fig. 4 is the process schematic that outsourcing figure is generated in the embodiment of the present invention;
Fig. 5 is the planning process schematic diagram of basic path planning mode in the embodiment of the present invention;
Fig. 6 is the exemplary diagram of intelligent path planning in the embodiment of the present invention;
Fig. 7 is the flow diagram of intelligent path planning in the embodiment of the present invention.
Specific embodiment
The present invention is described in detail combined with specific embodiments below.Following embodiment will be helpful to the technology of this field
Personnel further understand the present invention, but the invention is not limited in any way.It should be pointed out that the ordinary skill of this field
For personnel, without departing from the inventive concept of the premise, various modifications and improvements can be made.These belong to the present invention
Protection scope.
The present invention provides a kind of efficient and privacy the paths planning method based on extensive road network, including it is as follows
Step:
Static road network figure is divided, a balance search tree, balance search tree interior joint and static road network figure are constructed
Subgraph correspond, balance the corresponding static road network figure of root node in search tree;
Wherein, static road network figure is divided, constructs a balance search tree and specifically comprises the following steps:
Static road network figure is obtained, static road network figure is divided by multiple subgraphs according to default segmentation degree and node size, i.e.,
Static road network figure is divided into multiple subgraphs according to default segmentation degree, then each subgraph is divided into according to the default segmentation degree
Multiple subgraphs, the iteration above process is until number of vertex is not more than default node size in subgraph;
Father after being divided according to static road network figure schemes one balance search tree of building corresponding with the relationship between subgraph, balance
The subgraph of search tree interior joint and static road network figure corresponds, and balances the corresponding static road network figure of root node in search tree, balance
Each node possesses a unique identifier, one group of boundary point and a distance matrix in search tree.Certainly, each node also wraps
Include the weight of vertex set, side collection and each edge.
Wherein, a distance matrix possessing of balance search tree interior joint meets following require: nonleaf node away from
It is whether adjacent with them including the shortest distance between a pair of of boundary point from the information that matrix records boundary point in its child nodes
Label;The distance matrix of leaf node records the shortest distance between the boundary of the node and its vertex and whether indicates them
Adjacent label.
The each leaf node balanced in search tree is decomposed into one group of outsourcing figure and a linked, diagram, all outsourcing figures are deposited
It is stored in Cloud Server, all-links figure, balance search tree and real-time road condition information are stored in mist server, wherein each
Any pair of vertex in outsourcing figure is not directly connected, and the shortest distance between any pair of vertex in each outsourcing figure
It is measured not less than privacy;
Wherein, by each leaf node balanced in search tree be decomposed into one group of outsourcing figure and a linked, diagram specifically include as
Lower step:
Obtain leaf node and privacy measurement;
The boundary point in leaf node is bridged, for each boundary point pair of the leaf node, if between a boundary point pair most
Short path needs to calculate by the boundary point of other leaf nodes, then is the leaf node when collection adds a bridge joint, and be arranged
The bridge joint while weight be the bridge joint while correspond to the shortest distance between two boundary points;
A vertex is initialized to set, for storing all vertex pair in leaf node, and initializes a Priority Queues
For sky, which is a key assignments by vertex to frequency descending for storing candidate outsourcing vertex pair, the Priority Queues
To structure, wherein key is vertex pair, is worth the frequency for vertex pair;
For vertex to each vertex pair in set, if this opposite vertexes is not directly connected, and this opposite vertexes
Between the shortest distance measured not less than privacy, then enumerate all candidate outsourcing vertex pair of the shortest path between this opposite vertexes;
For each of enumerating candidate outsourcing vertex pair, its frequency in Priority Queues need to be updated, that is, candidate's outsourcing vertex of adding up
Between the frequency in the shortest path this opposite vertexes;Then vertex is repeated to this interior vertex is gathered to therefrom removing
The above process is until vertex is combined into sky to collection;
If Priority Queues is not sky, vertex sequence is constructed based on Priority Queues, and generate an outsourcing figure, by the outsourcing
Figure is added in outsourcing atlas, then generates linked, diagram, the vertex set in the linked, diagram belongs to vertex set and the outsourcing of the leaf node
The union of vertex set in atlas, side connection in the linked, diagram is vertex in the leaf node and outsourcing figure, in the linked, diagram
The weight on side is the shortest distance, and the vertex that then will be calculated by outsourcing figure is removed to from Priority Queues;It repeats the above steps
Until Priority Queues is sky;
Export one group of outsourcing figure and a linked, diagram.
Wherein, vertex sequence is constructed based on Priority Queues, and generates an outsourcing figure and specifically comprises the following steps:
Obtain leaf node, privacy measurement and vertex sequence;
It generates the vertex set of outsourcing figure: if vertex sequence is not sky, first vertex in vertex sequence being selected to be added
It is removed from vertex sequence into the vertex set of outsourcing figure, and by it;Then the remaining vertex in vertex sequence is traversed, if from top
First vertex and remaining a certain vertex removed in point sequence be directly connected to or both between the shortest distance be less than it is hidden
Private measurement, then remove from vertex sequence by the remaining a certain vertex, the remaining a certain vertex be otherwise added to outsourcing
In the vertex set of figure, and it is removed from vertex sequence;It repeats the above process until vertex sequence is sky;
Generate the side collection of outsourcing figure: the vertex pair formed for any two vertex in the vertex set of outsourcing figure, to outsourcing
Figure when concentrating and increasing the vertex to composition, the shortest distance of the weight on the side between two vertex be set;
Output one meets any pair of vertex and is not directly connected, and the shortest distance between any pair of vertex is not small
In the outsourcing figure of privacy measurement.
Wherein, output one meets any pair of vertex and is not directly connected, and the most short distance between any pair of vertex
Further include before following steps from the outsourcing figure not less than privacy measurement: any side being concentrated for the side of outsourcing figure, if outsourcing figure
Vertex set in there are a certain vertex so that the outsourcing figure side concentrate any side weight be equal to any side it is two corresponding
Vertex the sum of the shortest distance between a certain vertex respectively is then concentrated from the side of outsourcing figure and removes any side.
The source point and target point for obtaining user's input utilize outsourcing figure stored in cloud server based on path query model
The traveling road from source point to target point is searched for linked, diagram, balance search tree and the real-time road condition information stored in mist server
Diameter.
Wherein, cooking up driving path using path query model includes the basic path planning only considered under static road network
Mode and the intelligent path planning mode for considering real-time road condition information.
Wherein, basic path planning mode specifically includes following process:
Position leaf node belonging to source point and target point;
If source point and target point are on same leaf node, if the side that source point and target point connect and compose belongs to the leaf node
Side collection or source point and target point between the shortest distance less than privacy measure, then source point is calculated using dijkstra's algorithm
To the target point shortest distance and shortest path, otherwise call algorithm MD-Inside-leaf obtain source point to target point most short distance
From with incomplete shortest path;
If source point not on same leaf node, calls algorithm MD-Outside-leaf to obtain source point to mesh to target point
The shortest distance of punctuate and incomplete shortest path;
Algorithm PathRecovery is called to restore to obtain source point to the shortest path of target point, and final output source point is to mesh
The shortest path and the shortest distance of punctuate.
Wherein, the algorithm MD-Inside-leaf is comprised the following processes:
Obtain source point and target point, the leaf node where source point and target point, and balance search tree;
A set is initialized, by vertex that each neighbours' point of each neighbours' point of source point and target point is built into adding
Enter in the set;
The set is sent to Cloud Server, Cloud Server calculates each vertex pair in the set using one group of outsourcing figure
The shortest distance, for the vertex pair of not calculated result, default its shortest distance be ∞, Cloud Server return the set
In all vertex to the set of the corresponding shortest distance, wherein ∞ indicates infinitely great;
The shortest distance and incomplete shortest path of the source point to target point is calculated;
Dijkstra's algorithm is existing known algorithm, is not just repeated herein;
The algorithm MD-Outside-leaf is comprised the following processes:
Obtain source point and target point, two leaf nodes where source point and target point difference, and balance search tree;
The public ancestor node of minimum that two leaf nodes are found in balance search tree, finds two in balance search tree
Most short node path between leaf node;
The shortest distance and incomplete shortest path of the source point to target point is calculated;
The algorithm PathRecovery is comprised the following processes:
Obtain the shortest distance and imperfect shortest path of the source point to target point, and balance search tree and static road network
Figure;
Each edge in the imperfect shortest path of circular treatment judges whether the side is a line in static road network figure;
If the side is a line in static road network figure, which is added in complete shortest path, and from imperfect
The side is removed in shortest path;
If the side is not a line in static road network figure, the iteration a certain vertex that each starting point with the side is connected, if
The sum of shortest path of terminal is equal to the side when meeting the weight in starting point to a certain vertex and a certain vertex to this
Starting point to the shortest distance of terminal, then this is added to when starting point and a certain vertex connect and compose complete most short
Path is summarized, by a certain vertex and this while terminal constitute while be added in imperfect shortest path, and from it is imperfect most
The side is removed in short path;
Recycle above-mentioned treatment process to imperfect shortest path for sky, obtain source point to target point complete shortest path.
Intelligent path planning mode comprises the following processes:
Mist server obtains the source point that user inputs and target point and moment;
Mist server is arrived using the source point that the path planning process of basic path planning mode obtains under static road network first
The shortest path of target point;
An intelligent path is initialized, the shortest path under obtained static road network is assigned to intelligent path;
Based on the real-time road condition information inscribed when this, one is constructed by congestion sides all in the shortest path under static road network
The sequence of composition, then adjusts the shortest path between source point and target point with heuristics manner, and all congestion paths have adjusted
Afterwards, source point is obtained to the final intelligent path of target point and path cost and is exported to user.
Come below with reference to a specific embodiments and the drawings to the elaboration contents of the present invention.
This programme is based on a comprehensive service framework BCloud-IFog, it is by blind Cloud Server (corresponding to BCloud)
With intelligent mist server (corresponding to IFog) composition.On the basis of this frame, a kind of the real-time of outsourcing of this programme is proposed
Paths planning method, to realize the path planning of efficient, privacy and intelligence.
Meaning representated by the symbol occurred in the present embodiment is as shown in the table.
As shown in Figure 1, this BCloud-IFog frame is made of four parts: LBS supplier (LBS provider), movement
User (Mobile user), Cloud Server (Cloud Server), mist server (Fog Server).LBS provider is as former
The data owner of beginning road network figure G, is contracted out to Cloud Server for the non-confidential data of G, by confidential data (such as real-time traffic of G
Data) all mist servers are shared with, so that mobile subscriber can not only entrust to most of calculate in search process
Cloud Server can also enjoy the intelligent path planning service from mist server.
Specifically, in order to improve search efficiency, LBS supplier passes through the static road network figure of division firstTo construct one
Balance search tree indexThen, all leaf figures of G*-tree are decomposed into one group of outsourcing figureWith one group of chain
Map interlinkingWhereinCloud Server is contracted out to,WithIt is stored on mist server.In order to make mobile subscriber enjoy more intelligence
Real-time road data are also stored in the mist server of the path planning of energy, the maintenance of LBS provider.This programme allows mobile subscriber's root
According to needing to select different types of path planning.User can be from mist server download link figureWithIf it is
Infrastructure service, i.e., do not consider real-time road, and user can be interacted with Cloud Server to obtain query result.If it is Intellectual garment
Business, user interact the intelligent path planning of progress with mist server.
Specifically, a road network can be modeled as undirected weighted graph G=(V, E, W, a W*), wherein V is one group
Vertex, represents one group of POIs (point of interest can be a shop, a house or a bus station etc.) or crossing, E are one group
Indicate the side of road connection, W and W* respectively represent the static state and changeable weight of each edge.| | V | | and | | E | | it is respectively defined as G
The quantity on middle vertex and side.Give one group of vertex V=(v1,...,vn), connect vertex viAnd vjSide be defined as e (vi,vj),
The static weight on the side is defined as w (vi,vj), as vertex viWith vjBetween Euclidean distance (nonnegative value), dynamic weigh
It is defined as w again*(vi,vj), i.e., the side is in the real time weight of (T, c), and wherein the T moment, whether congestion was defined with c ∈ { 0,1 }.Road
Diameter p (vs,vt) by a series of sides < e (vs,vi),...,e(vj,vt) > indicate, the sum of the weight in path are vsTo vtDistance Dist
(p(vs,vt)).If distance be it is shortest, the path be shortest path SP (vs,vt), the corresponding shortest distance is δ (vs,vt)。
In the present embodiment,For the static road network figure of G.Consider two kinds of path plannings: rightStatic path rule
It draws and to the intelligent path planning of G.Practical static path planning is the shortest distance inquiry in static road network.Q=(vs,vt)
To inquire vsTo vtShortest path, path cost Φ can calculate with following equation:
Intelligent path planning both considers distance, it is contemplated that real-time road.V is inquired in TsTo vtShortest distance table
It is shown as q=(vs,vt,T).In cost function is α apart from specific gravity, and road conditions are specific gravity β, wherein α, β ∈ R+, alpha+beta=1, then path
Cost Φ can be calculated with following equation:
Wherein λ ∈ Z+For increasing the weight on congestion side, the first part in equation is related to path distance, second part
It is related to congestion state.IP(vs,vt) indicate the water channel principium that intelligent planning goes out, minimum cost δ*(vs,vt)。
Symbol [n] and [n1,n2] respectively indicate set { 1 ..., n } and { n1,...,n2}.We useRepresent with
Machine takes out an element x from set X.The output x of algorithm A is defined asByThe balance search tree generated Each node is correspondingIn a subgraph.For each leaf nodeOne group of outsourcing figure can be generatedWith a linked, diagramIt can be q=(v with thems,vt) calculate shortest distance δ (vs,vt) and shortest path SP
(vs,vt).For real-time query q=(vs,vt, T), possess minimum cost δ*(vs,vt) intelligent path IP (vs,vt) recommended
To user.
Given segmentation degree f, original image is divided into f subgraph by us, i.e.,And it is full
Foot Column Properties:
Vertex integrality:
The non-intersecting property in vertex: given i ≠ j, then
Subgraph integrality: forIf e (vx,vy) ∈ ε, then e (vx,vy)∈εi,
Original image is known as by weHypergraph, wherein i ∈ [f].And if only ifεj∈εiWhen, subgraphReferred to as
Another subgraphHypergraph.
Since after figure divides, the vertex of a part connection is assigned to different subgraphs in original graph.We to
Give a definition and distinguish this vertex and other vertex:
Give a leaf nodeIfAndThen vertexReferred to as one
A boundary point.InIn one group of boundary point be expressed as
In G*-tree, each node has one such as undefined distance matrix:
The distance matrix of one nonleaf node records the information of its child's boundary point, including most short between a pair of of boundary point
Distance and their whether adjacent labels.The distance matrix of leaf node records the most short distance between the boundary of the node and its vertex
From and indicate their whether adjacent labels.
It is a balance search tree, meets following condition:
In each node it is correspondingA subgraph, root node corresponds to original imageThe corresponding figure of father node is its child
The hypergraph of child node;
Each nonleaf node has f child;
Each leaf node at most has τ vertex, and all leaf nodes appear in same layer;
Each node includes a unique node identifierOne boundary setWith a distance matrix
First three above condition ensuresIt is a balance search tree, the last one condition is for efficiently inquiry and path
The important attribute of recovery.
Each leaf node in G*-treeOne group of outsourcing figure will be broken down intoIt is linked with one
Figure It is deployed in Cloud Server,It is deployed in mist server, in order to avoidGraph structure be exposed to cloud, it is each outer
Packet figure requires to meet 1-neighborhood-d-radius attribute:
1-neighborhood-d-radius attribute: for any pair of vertexAnd δ
(vx,vy)≥d。
In this definition, two vertex for having side to connect are considered as (1-neighborhood) of privacy by we first.Its
It is secondary, in road network, if the distance between two vertex relatively close (threshold value d), the relationship even without connection, between them
It is important, and needs to protect.That privacy measurement d is defined is exactly the threshold value (d- of the shortest distance in outsourcing figure between vertex
radius).That is, it is that any two vertex is connected to or distance be less than d, the two vertex be do not appear in it is same
In outsourcing figure.And linked, diagramMaintain original leaf nodeIn vertex and outsourcing figureBetween relationship.Assuming thatFor linked, diagram, whereinForIn each sidevxWithRespectively
It comes fromWithIf side e (vx,vy)∈εi, then increase sideTo linked, diagram, side rightIt is set to δ again
(vx,vy).There are kind of a special circumstances, i.e. sideIts weight is set as 0.
If two vertexIt is disjunct, andIn they the distance between be not less than d, then they
Shortest distance δ (vs,vt) outsourcing atlas can be passed throughAnd linked, diagramTo calculate.Equation is as follows:
Notice that the relevant shortest path of equation (3) is made of multiple summits, each edge by equation each single item two vertex shapes
At.For example, the shortest distanceThen its imperfect path is
Candidate outsourcing vertex pair: given subgraphIn an opposite vertexes (vs,vt), and δ (vs,vt) >=d,When meeting following condition, vertex pairIt is (vs,vt) a candidate outsourcing vertex pair:
δ(vs,vt)=δ (vs,vx)+δ(vx,vy)+δ(vy,vt)
δ(vs,vx) < d or e (vs,vx)∈εi, δ (vy,vt) < d or e (vy,vt)∈εi
δ(vx,vy) >=d and
This programme main thought is: the growth of road network scale makes urgent need of the high search efficiency as path planning.Most
Advanced G-tree structure permission carries out efficient shortest path query on large-scale road network.But it requires client to exist
All calculating are executed in search process, cause availability low.The inquiry of G-tree structure is divided into two classes: inquiry vertex is same
Leaf figure (MD-Inside-Leaf) and inquiry vertex in different leaf figures (MD-Outside-Leaf), their search at
This increases with the increase of leaf node size τ and tree height h respectively, and height h is inversely proportional with leaf node size τ.In other words
It says, the higher query performance that will lead to MD-Inside-Leaf of the query performance of MD-Outside-Leaf reduces.Our target
It is the cost for reducing by two kinds of inquiries simultaneously.Our essential idea is one improved G*-tree structure of building, and subgraph is turned
It changes and generates one group of outsourcing figure with 1-neighborhood-d-radius attribute, Cloud Server is allowed to share MD-Inside-
Searching cost in Leaf.Since incredible cloud may be a potential attacker, our schemes need to protect as far as possible
The data-privacy of LBS provider and the inquiry privacy of mobile subscriber cannot allow cloud to obtain complete road network, can not know use
The real trace at family.In addition to this, we, which also consider, realizes intelligentized inquiry, the i.e. path planning under real-time traffic.
In order to realize privacy and efficient shortest path query, we construct a G*-tree structure, and by each leaf
Node conversion generates one group of outsourcing figure with 1-neighborhood-d-radius attribute, in the case where not revealing privacy
Cloud Server is allowed to participate in calculating.The process of the base case is as follows:
The building of G*-tree
Our one TreeGen of algorithm for design construct a G*-tree.
Algorithm one (TreeGen): input original static road network figure firstSegmentation degree f and node size τ, then executes
Following steps:
Step 1: will be schemed using parameter f and τIt is divided into multiple subgraphs.We using Metis tool (http: //
Glaros.dtc.umn.edu/gkhome/metis/metis/overview/ multistage k k-path partition mode segmentation figure).It is i.e. quiet
State road network figureIt is divided into the subgraph of f equal sizes, then each of which subgraph is divided into f subgraph, mistake as iteration
Journey, until number of vertex is no more than τ in the subgraph of division, as shown in Figure 2, segmentation degree is 2, node size 5.Then according to father
Relationship between figure and subgraph generates a balance search tree G*-tree, and this method makes each sub- boundary point of graph quantity to the greatest extent may be used
It can few query cost with after reduction.
Step 2: construction G*-tree node.We construct the set membership of G*-tree node according to the set membership of figure.
Each leaf node is corresponding with one minimum subgraph (number of vertex is no more than τ), and the quantity of leaf node is equal to the quantity of subgraph.It is each non-
Leaf node is corresponding with the hypergraph of its child nodes, and root node corresponds to figureIn addition, each node possesses a unique identifier
(corresponding with the mark of figure), one group of boundary point and a distance matrix.
Step 3: building distance matrix: each node in G*-tree is owned by a distance matrix as defined 3, this
It is essential for minimum distance calculation.Each subgraph after division have one group define 2 as described in boundary point, and away from
It from matrix is constructed according to boundary point.For the distance matrix of leaf node, the row and column of matrix is respectively the top of the node
Point and boundary point, in matrix each between them the shortest distance and whether adjacent label.For nonleaf node away from
From matrix, row and column is the union of its all child nodes boundary point, each in matrix is the shortest distance between them
Whether adjacent label, if the boundary point of row and column comes from the same child nodes, is not remembered to save memory space
Record this.
Output:
Such as in Fig. 2, f=2, τ=5 are given.First by original static figureIt is divided into 2 subgraphsThey into
One step is divided intoWithG*-tree as shown in Figure 3 can be built into after division.For leaf nodeIt is wrapped
Containing five vertex { v7,v8,v9,v10,v17, wherein vertex v7And v9It is boundary point, thereforeDistance matrix have recorded boundary point
{v7,v9And vertex { v7,v8,v9,v10,v17Between the shortest distance.Wherein, item (v7,v10)=(11,0) mean v7And v10
Between shortest path distance be 11 and non-conterminous.For nonleaf nodeIts child nodesWithSeparately include boundary point
{v0,v5,v6And { v7,v9, therefore its distance matrix records any two boundary point (the same section in the two boundary point sets
Point two boundary points except) between shortest path distance.
The conversion of G*-tree
For each leaf node of G*-treeCan be generated one group by two SubgraphTrans of algorithm has 1-
The outsourcing figure of neighborhood-d-radius attributeWith a linked, diagramIn this way if two non-conterminous vertex it
Between distance be not less than d, then can be used equation (3) calculate the shortest distance.Due to linked, diagramIt will be downloaded simultaneously by mobile subscriber
It is stored in client, therefore our optimization aim is to minimize the size (i.e. number of edges) of linked, diagram.Different figure conversion plans
Generate different size of linked, diagram.It is enumerated using brute-force method allIt is expensive and infeasible to find optimal result.Cause
This, algorithm two reasonably converts each leaf node of G*-tree with heuristics manner.
Algorithm two (SubgraphTrans): input leaf node firstWith privacy measure d, then execute with
Under several steps:
Step 1: the boundary point in bridge joint leaf node, for each boundary point pair of the leaf nodeIf bx
And byShortest path need by other leaf nodes boundary point calculate, it is necessary to be εiAdd a bridge joint side e (bx,by), and
W (b is setx,by)=δ (bx,by).The step makes the shortest distance in each leaf node between vertex be self-contained.Example
Such as, leaf nodeIn vertexv7And v9Between shortest path include leaf nodeIn vertex v6, we must increase by one
Side e (v7,v9), make w (v7,v9)=δ (v7,v9)=9, Cai NengIn be computed correctly the shortest distance between vertex.
Step 2:
A set P is initialized, for storingIn all vertex pair.And initialize a Priority QueuesWith
In storing candidate outsourcing vertex pair, Q is a key-value pair structure by vertex to frequency descending, and wherein key is vertex pair, and value is
The frequency on vertex pair.
For each vertex to (vs,vt) ∈ P, ifIt (meeting outsourcing condition), then enumerates
SP(vs,vt) all candidate outsourcing vertex pair, for each candidate outsourcing vertex to (vx,vy), it is assumed that it is in SP (vs,vt)
Frequency is num, then needs to update its frequency in Q, i.e., cumulative num.(v is handleds,vt) after, it just removes, repeated from P
Journey is 2. until P is sky.
Step 3: if Q is not sky, vertex sequence being constructed based on QThree OutGraphGen of algorithm is called to generate one
Outsourcing figure, i.e.,It is added into outsourcing atlasThen linked, diagram is improvedLink
Vertex in figure isSide connection beWithIn vertex, weight is the shortest distance.Then will
It is removed by the vertex that outsourcing figure calculates to from Q, repeats the step until Q is sky.
Output:
Three OutGraphGen of algorithm is had invoked in algorithm two, the implementation procedure of algorithm three is as follows:
Algorithm three (OutGraphGen): input leaf nodeD is measured with privacy, is formed there are also algorithm two
Vertex sequenceThen following steps are executed:
Step 1 (generates the vertex set V of outsourcing figurei o): ifIt is not sky, then selectsIn first vertex vxIt is added
In vertex set, and by its fromMiddle removal, i.e. Vi o←Vi o∪{vx}.Then it traversesIn remaining vertex, it is assumed that traversal vertex
ForIf e (vx,vy)∈εiOr δ (vx,vy) < d (is unsatisfactory for outsourcing figure condition), then by vyFromMiddle removal.It repeats
The step untilFor sky.
Step 2 (generates the side collection of outsourcing figure): for Vi oIn each vertex to (vx,vy), toIt is middle to increase by one
Side e (vx,vy), setting side right weight w (vx,vy) it is δ (vx,vy)。
Step 3 (deletes unnecessary side): being complete graph by the outsourcing figure that above step generates.In order to save Cloud Server
On memory space, need to delete unnecessary side.I.e. for Vi oIn each vertex to (vs,vt), ifMeet w
(vs,vt)=δ (vs,vx)+δ(vx,vt), then fromMiddle removal e (vs,vt)。
Output: an outsourcing figure with 1-neighborhood-d-radius attribute
As shown in figure 4, we are with subgraphFor conversion process is described.Fig. 4-(a) beSelf-contained figure, wherein e
(v7,v9) it is a bridge joint side, w (v7,v9)=δ (v7,v9).Fig. 4-(b) beIn shortest path between certain vertex pair, with
SP(v8,v10) for, it is assumed that d=2, its candidate outsourcing vertex is to there is (v8,v9),(v8,v10),(v7,v10).According to algorithm two,
A vertex sequence to being stored in frequency Priority Queues Q, and is constructed based on Q in all candidate outsourcing vertex by usRecalling algorithm three can be generated one such as Fig. 4-(c) outsourcing figure, it is clear that the figure meets 1-
The data-privacy of neighborhood-d-radius attribute, road network is protected.
Query process
In this scenario, the Search agreement between mobile subscriber and Cloud Server can inquire source point vsWith target point vt
Between shortest path and its distance.It is set with a Hash table and has recorded pass between each vertex and its affiliated leaf node
System.Given inquiry q (vs,vt), user navigates to v using Hash table firstsAnd vtAffiliated leaf node.Then it is inquired, shape
In formula, inquiry can be divided into two types: (1) MD-Inside-Leaf:vsAnd vtPositioned at identical leaf node.We use
Dijkstra's algorithm calls five MD-Inside-Leaf of algorithm to obtain query result.(2) MD-Outside-Leaf:vsAnd vtPosition
In different leaf nodes.In this case, mobile subscriber will call six MD-Outside-Leaf of algorithm based on G*-tree.For
Both inquiries, the path of generation is all incomplete.Therefore, in the final step, mobile subscriber uses algorithm seven
PathRecovery restores to obtain complete shortest path.Flow chart such as Fig. 5 of basic path planning mode is specifically described such as
Under:
Algorithm four (Search): one inquiry q (v of inputs,vt), then mobile subscriber and Cloud Server execute following respectively
Step:
Mobile subscriber:
Step 1: navigating to v using Hash tablesAnd vtAffiliated leaf node, is used respectivelyWithIt indicates.
Step 2: judging vsAnd vtWhether in the same leaf node.
If e (vs,vt)∈εiOr δ (vs,vt) < d is then existed based on cost equation (1) using dijkstra's algorithmMiddle calculating δ (vs,vt) and SP (vs,vt), otherwise algorithm MD-Inside-leaf is called to obtain δ (vs,vt) and
Algorithm MD-Outside-leaf is called to obtain δ (vs,vt) and
Step 3: algorithm PathRecovery being called to restore to obtain fullpath SP (vs,vt)。
Output: SP (vs,vt) and δ (vs,vt)
Cloud Server: it when mobile subscriber calls MD-Inside-leaf, can be interacted with Cloud Server, user sends one group of time
Vertex is selected to allow cloud to inquire each pair of shortest distance P, steps are as follows:
Step 1: receiving one group of vertex pair of mobile subscriber's transmission
Step 2: usingCalculate the shortest distance on each vertex pair in P.It is assumed thatInclude vertex pairDijkstra's algorithm, which can be used, to be calculatedThe process is repeated until vertex pair all in P
It is all processed, for the vertex pair of not calculated result, it is defaulted as ∞.
Output:
In Search agreement, several other algorithms are had invoked, we are described respectively:
Algorithm five (MD-Inside-Leaf): the algorithm is starting point vsWith target point vtBelong to the same leaf node, andδ(vs,vtIt is called when) >=d.Input inquiry q (v firsts,vt), leaf nodeWithThen it executes
Following steps:
Step 1: one set of initialization
Step 2: by vsEach neighbours' point and vtThe vertex that is built into of each neighbours' point to P is added.For example, vxFor vs
Neighbours' point, vyFor vtNeighbours' point, we willP is added in (expression in outsourcing figure), i.e.,
Step 3: sending P to Cloud Server, Cloud Server returns
Step 4: calculating to obtain δ (v using equation (3)s,vt) and
Output:
In the algorithm, mobile subscriber is interacted with Cloud Server to accelerate to inquire.Such as given inquiry q (v8,v9), it is mobile
User only sends one group of candidate's outsourcing vertex pairTo cloud, then use
Shortest path is calculated using equation (3) in family.In this process, cloud does not know that real inquiry and the track of user.
Algorithm six (MD-Outside-Leaf): the algorithm is starting point vsWith target point vtWhen being not belonging to the same leaf node
It calls.Input inquiry q (v firsts,vt), leaf nodeWith Then following steps are executed:
Step 1:InIn findWithThe public ancestors of minimum Indicate a node,
It is The maximum ancestors of depth.For example, in Fig. 4,
Step 2:InIn findWithBetween most short node path Indicate fromIt arrives
Node is removed in the path that the node of process is constitutedExcept.For example, in Fig. 4,
Step 3: calculating δ (v using equation (4) and equation (5)s,vt) and
Wherein,Indicate G*-tree interior jointFather node.The corresponding shortest path of the two equatioies is by multiple summits group
At each edge is formed by two vertex in equation each single item.For example, shortest distance δ (bs,bt)=δ (bs,bx)+δ(bx,by)+
δ(by,bt), then its imperfect path is
Output:
What algorithm five and algorithm six obtained is imperfect path, and we must obtain complete path just in practice
Significant, seven PathRecovery of algorithm can then restore complete shortest path.
Algorithm seven (PathRecovery): input shortest distance δ (vs,vt), incomplete shortest path With original static road network figureThen following steps are executed:
Step 1: circular treatmentIn each edge e (vx,vy), judge whether the side is a line in the figure of source, is
Then follow the steps 2, it is no to then follow the steps 3.
Step 2:e (vx,vy) be source figure in a line, i.e. e (vx,vy)∈ε∧w(vx,vy)=δ (vx,vy).So e
(vx,vy) it is complete, addition SP (vs,vt), and fromMiddle removal e (vx,vy).Next is handled then according to step 1
Side.
Step 3:e (vx,vy) it is not a line in the figure of source.According to the connectivity of road network, shortest path is bound to by vx
Neighbours' point.So iteration is each and vxConnected vertex, if the point is vnIf meeting w (vx,vn)+δ(vn,vy)=δ
(vx,vy), then v is passed through in pathn, wherein e (vx,vn) it is complete, addition SP (vs,vt), and e (vn,vy) need to be added
Continue to restore, removes e (vx,vy), continue to execute step 1.
Output: SP (vs,vt)
In our scheme, a part of MD-Inside-Leaf is calculated and gives Cloud Server by user, and cloud service
Device possesses powerful computing capability, keeps this part calculating ratio prior art more efficient.For MD-Outside-Leaf, Wo Menyou
Distance value of the boundary point from same node in nonleaf node is omitted, to a certain extent in the distance matrix for having changed G*-tree
Decrease computing cost.So the overall calculation efficiency of our schemes is better than the prior art.
As shown in fig. 6, the path planning in static road network is only considered in base case, without considering real-time road
Have a significant impact to Intelligentize query.For example, inquiry v3To v8Path, the distance inquired in static road network most it is short be Φ (p
(v3,v8))=δ (v3,v8)=14, corresponding shortest path is as shown in Fig. 6-(a) thick line, i.e. SP (v3,v8)=< e (v3,v2),e
(v2,v1),e(v1,v5),e(v5,v15),e(v15,v6),e(v6,v7),e(v7,v8)>.However, found when considering real-time road,
SP(v3,v8) in e (v5,v15) and e (v15,v6) it is congestion, according to equation (2), if α=0.4, β=0.6, λ=2 are then calculated
Obtaining the path cost is 8, if we use e (v5,v6) the two congestion sections are avoided, path cost is only 6.We can make
V is calculated with equation (2)3To v8The aisled path cost of institute, finds the intelligent path of cost minimization, finally calculates to obtain the intelligence
It is IP (v that path, which can be changed,3,v8)=< e (v3,v2),e(v2,v1),e(v1,v5),e(v5,v6),e(v6,v7),e(v7,v8) >, is as schemed
Shown in the thick dashed line of 6- (b).Therefore, it is proposed that an advanced scheme, considers real-time road factor, to cook up more reasonably
Path.
Since real-time road data have very high commercial value, these data is protected to be necessary.We were both
It is not desired to store it in insecure Cloud Server, is also not desired to distribute them to user and loses its value.So we use
These data are protected than the mist server of Cloud Server more secure and reliable.Obviously, mist server level, Search association are increased
View will also change correspondingly, which is indicated with eight Search* of algorithm.As shown in fig. 7, intelligent path planning mode it be related to moving
Interaction between user, mist server and Cloud Server.Different from Search agreement, in Search*, mobile subscriber will directly be looked into
Ask q (vs,vt, T) and mist server is entrusted to, certainly, in this case, user does not need download link figure or G*-tree.Mist
Server once receives request, can work as mobile subscriber in Search agreement first, obtain complete shortest path and
Then distance calls IRoutePlan algorithm to obtain more preferably path and relevant cost.Specifically, for MD-Inside-
Leaf inquiry, mist server may may require that cloud participates in.
Algorithm eight (Search*): one inquiry q (v of inputs,vt, T), then mobile subscriber, mist server and Cloud Server
Following steps are executed respectively:
Mobile subscriber:
Step: q (v is sent to mist servers,vt,T)
Output: and then obtain IP (vs,vt) and δ*(vs,vt)
Mist server:
Step 1: it is identical as the work of mobile subscriber in Search agreement, obtain SP (vs,vt) and δ (vs,vt)
Step 2: algorithm IRoutePlan being called to obtain IP (vs,vt) and δ*(vs,vt)
Output: IP (vs,vt) and δ*(vs,vt)
Cloud Server:
Step: it is identical as the work of Search agreement, but interacted with mist server
Output:
Algorithm nine (IRoutePlan) is an intelligent path planning algorithm.Most direct mode be allow mist server again
Water channel principium is calculated using equation (2).This is a kind of global optimum, but expense is very high.In fact, congestion state is real
Shi Bianhua's, global optimum is temporary.So our algorithm is compromise, local optimum is generated as a result, searching to improve
Rope efficiency.
Main thought is one intelligence path IP (v of initialization firsts,vt)=SP (vs,vt), one is constructed by SP (vs,
vt) in all congestion sides composition sequence C, source point v is then adjusted with heuristics mannersWith target point vtBetween shortest path
SP(vs,vt).We are by congestion side (vx,vy) state be divided into two kinds of situations:
vxAnd vyBelong to the same leaf nodeIn this case, pass through removal firstIn all congestion Bian Laigeng
New sequence C, and fromIn find SP (vs,vt) on first vertex vfWith the last one vertex vl.Then, we determined that vl
It whether is target point.If it is, it is only necessary to v is calculated using the equation (2) based on dijkstra's algorithm in the nodefAnd vtIt
Between shortest path/distance, and use SP (vf,vt) adjustment vfAnd vtPath.If vlIt is not representative points, such case
Under, vlFor the outlet (boundary) of the leaf node, we attempt to find other better outlets.We calculateIn each boundary
Point b to target point vtThe shortest distance, with δ (b, vt) indicate.In δ (b, vt) it is less than threshold θlAll boundary points in, selection with
vfApart from the smallest boundary point.
vxAnd vyIt is the boundary point of different leaf nodes, i.e., they are respectively two sub- boundarys point of graph.In this case,
We find trial and vxConnected other boundary points replace vy.For with vxAdjacent each boundary point b, calculates it away from target
Point vtShortest distance δ (b, vt).In δ (b, vt) it is less than threshold θyAll boundary points in, selection and vxThe side right weight of composition is most
Small boundary point.In both cases, if shortest path changes, need to adjust IP (vs,vt) in corresponding path and at
This δ * (vs,vt), and update congestion sequence C.
After all congestion paths have adjusted, (δ * (v will be obtaineds,vt),IP(vs,vt))。
It is the threshold value that shortest distance between points are pushed up in outsourcing figure that privacy, which measures d, only distance more than or equal to d be possible to it is outer
Packet, less than d by side as linked, diagram.If threshold value d is excessive, that is to say, that we will regard apart from farther away two vertex
For privacy, the scale that will lead to linked, diagram becomes larger, and the calculation amount of user becomes larger.If threshold value d very little, the vertex of outsourcing becomes
More, outsourcing cost becomes larger.We are difficult to find a compromise value.In order to save calculating cost, we are simple during the experiment
Ground utilizes the average value of side right weightSuch as
In IRoutePlan algorithm, when the shortest distance of the boundary point away from target point is less than threshold value, intelligent rule will be considered as
Draw the candidate vertices in path.Therefore, θlAnd θyValue have a significant impact to the result of intelligent path planning.If threshold value very little,
Most of boundary point can then be filtered out.If threshold value is very big, most of boundary point will be considered as candidate point.But restoring
Before complete shortest path between each boundary point and representative points, it is difficult to determine suitable value.In order to save calculating cost,
We are simply by θlAnd θyIt is respectively set toWithThis is consistent with equation (2).
This programme proposes a kind of efficient and privacy intelligent paths planning method based on extensive road network.LBS is provided
Person divides static road network figure and is built into a balance search tree G*-tree, is then converted into leaf node each in G*-tree
Two parts --- outsourcing figure and linked, diagram.Outsourcing figure is stored in Cloud Server, linked, diagram, G*-tree and real-time road condition information
Be stored in mist server, mobile subscriber can in mist server download link figure and G*-tree.
This programme is dedicated to realizing efficient inquiry and the path planning operation of intelligence, our multi-analysis this programme
Privacy and performance, and rigorous experiment has been carried out with large-scale real data set, with the high efficiency of proof scheme.Through examining
It tests, this programme realizes efficient and intelligent shortest path query, and ensure that the privacy of data.In experiment, big rule are used
The real data set of mould is tested, the results show this programme efficiency it is very high (diagram data on 1,000,000 vertex is concentrated,
It is average only to need 2.4 seconds when executing the inquiry of more shortest path).
1) high availability.MD-Inside-Leaf is inquired, mobile subscriber requires Cloud Server to calculate one group of candidate most
Short path, it is calculated using equation (2).Therefore, the calculating cost of client substantially reduces.2) intelligent Service.Mist server can
According to real-time road dynamic adjustment shortest path.Therefore, mobile subscriber can enjoy intelligent Service.3) enhance privacy.Firstly,
All outsourcing figures being stored in Cloud Server meet 1-neighborhood-d-radius attribute.Secondly, mobile subscriber only to
Cloud Server sends candidate outsourcing vertex pair, and Cloud Server can not know starting point and target and the mobile subscriber of user query
Any paths will be selected.Therefore, data-privacy and inquiry privacy are protected.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of efficient and privacy paths planning method based on extensive road network, which is characterized in that including walking as follows
It is rapid:
Static road network figure is divided into multiple subgraphs, constructs a balance search tree, balance search tree interior joint and static road network
The subgraph of figure corresponds, and balances the corresponding static road network figure of root node in search tree;
The each leaf node balanced in search tree is decomposed into one group of outsourcing figure and a linked, diagram, all outsourcing figures are stored in
In Cloud Server, all-links figure, balance search tree and real-time road condition information are stored in mist server;Wherein, each outsourcing
Any pair of vertex in figure is not directly connected, and the shortest distance between any pair of vertex in each outsourcing figure is not small
It is measured in privacy;
The source point and target point for obtaining user's input utilize outsourcing figure stored in cloud server and mist based on path query model
The linked, diagram, balance search tree and the real-time road condition information that store in server search for the driving path from source point to target point.
2. paths planning method according to claim 1, which is characterized in that static road network figure is divided into multiple subgraphs,
A balance search tree is constructed to specifically comprise the following steps:
It obtains static road network figure and it is divided, static road network figure is divided into multiple subgraphs according to default segmentation degree, then
Each subgraph is divided into multiple subgraphs according to the default segmentation degree, the iteration above process is until number of vertex is not more than in subgraph
Default node size;
Father after being divided according to static road network figure schemes one balance search tree of building corresponding with the relationship between subgraph, balance search
The subgraph for setting interior joint and static road network figure corresponds, and balances the corresponding static road network figure of root node in search tree, balance search
Each node possesses a unique identifier, one group of boundary point and a distance matrix in tree.
3. paths planning method according to claim 2, which is characterized in that possess one of balance search tree interior joint away from
Meet following require from matrix: the distance matrix of a nonleaf node records the information of boundary point in its child nodes, including one
To the shortest distance and their whether adjacent labels between boundary point;The distance matrix of leaf node record the node boundary and
The shortest distance between its vertex and indicate their whether adjacent labels.
4. paths planning method according to claim 1, which is characterized in that each leaf node in search tree point will be balanced
Xie Weiyi group outsourcing figure and a linked, diagram specifically comprise the following steps:
Obtain leaf node and privacy measurement;
The boundary point in leaf node is bridged, for each boundary point pair of the leaf node, if the shortest path between a boundary point pair
Diameter needs to calculate by the boundary point of other leaf nodes, then is the leaf node when collection adds a bridge joint, and the bridge is arranged
The weight of edge fit is the shortest distance between corresponding two boundary points in the bridge joint side;
A vertex is initialized to set, for storing all vertex pair in leaf node, and initializing a Priority Queues is sky,
The Priority Queues is a key-value pair knot by vertex to frequency descending for storing candidate outsourcing vertex pair, the Priority Queues
Structure, wherein key is vertex pair, is worth the frequency for vertex pair;
For vertex to each vertex pair in set, if this opposite vertexes is not directly connected, and between this opposite vertexes
The shortest distance is measured not less than privacy, then enumerates all candidate outsourcing vertex pair of the shortest path between this opposite vertexes;For
Candidate outsourcing vertex pair each of is enumerated, its frequency in Priority Queues need to be updated, candidate's outsourcing vertex of adding up is at this
The frequency in shortest path between one opposite vertexes;Then vertex is repeated into above-mentioned mistake to this interior vertex is gathered to therefrom removing
The vertex Cheng Zhizhi is combined into sky to collection;
If Priority Queues is not sky, vertex sequence is constructed based on Priority Queues, and generate an outsourcing figure, which is added
Enter in outsourcing atlas, then generate linked, diagram, the vertex set in the linked, diagram belongs to the vertex set and outsourcing atlas of the leaf node
The union of middle vertex set, side connection in the linked, diagram is vertex in the leaf node and outsourcing figure, side in the linked, diagram
Weight is the shortest distance, and the vertex that then will be calculated by outsourcing figure is removed to from Priority Queues;Repeat the above steps until
Priority Queues is sky;
Export one group of outsourcing figure and a linked, diagram.
5. paths planning method according to claim 4, which is characterized in that vertex sequence is constructed based on Priority Queues, and
An outsourcing figure is generated to specifically comprise the following steps:
Obtain leaf node, privacy measurement and vertex sequence;
It generates the vertex set of outsourcing figure: if vertex sequence is not sky, selecting first in vertex sequence vertex to be added to outer
In the vertex set of packet figure, and it is removed from vertex sequence;Then the remaining vertex in vertex sequence is traversed, if from vertex sequence
First vertex and remaining a certain vertex removed in column be directly connected to or both between the shortest distance be less than anonymity
Amount, then remove from vertex sequence by the remaining a certain vertex, the remaining a certain vertex be otherwise added to outsourcing figure
In vertex set, and it is removed from vertex sequence;It repeats the above process until vertex sequence is sky;
Generate the side collection of outsourcing figure: the vertex pair formed for any two vertex in the vertex set of outsourcing figure, to outsourcing figure
When concentration increases the vertex to composition, the shortest distance of the weight on the side between two vertex is set;
It is to be not directly connected, and the shortest distance between any pair of vertex is not less than privacy that output one, which meets any pair of vertex,
The outsourcing figure of measurement.
6. paths planning method according to claim 5, which is characterized in that it is not that output one, which meets any pair of vertex,
It is directly connected to, and further includes walking as follows before outsourcing figure of the shortest distance between any pair of vertex not less than privacy measurement
It is rapid: any side to be concentrated for the side of outsourcing figure, if there are a certain vertex in the vertex set of outsourcing figure, so that the side collection of the outsourcing figure
The weight of middle any side is equal to corresponding two vertex of any side the sum of the shortest distance between a certain vertex respectively, then
It is concentrated from the side of outsourcing figure and removes any side.
7. paths planning method according to claim 1, which is characterized in that cook up traveling road using path query model
Diameter includes the intelligent path planning mode for only considering basic path planning mode and consideration real-time road condition information under static road network.
8. paths planning method according to claim 7, which is characterized in that basic path planning mode specifically includes as follows
Process:
Position leaf node belonging to source point and target point;
If source point and target point on same leaf node, if source point and target point connect and compose while belong to the leaf node while
Collection or the shortest distance between source point and target point are measured less than privacy, then source point are calculated to mesh using dijkstra's algorithm
The punctuate shortest distance and shortest path, otherwise call algorithm MD-Inside-leaf obtain source point to target point the shortest distance with
Incomplete shortest path;
If source point not on same leaf node, calls algorithm MD-Outside-leaf to obtain source point to target point to target point
The shortest distance and incomplete shortest path;
Algorithm PathRecovery is called to restore to obtain source point to the shortest path of target point, and final output source point is to target point
Shortest path and the shortest distance.
9. paths planning method according to claim 8, which is characterized in that the algorithm MD-Inside-leaf includes such as
Lower process:
Obtain source point and target point, the leaf node where source point and target point, and balance search tree;
A set is initialized, the vertex that each neighbours' point of each neighbours' point of source point and target point is built into should to addition
In set;
The set is sent to Cloud Server, Cloud Server calculates each vertex pair in the set most using one group of outsourcing figure
Short distance, for the vertex pair of not calculated result, defaulting its shortest distance is ∞, and Cloud Server returns to institute in the set
There is vertex to the set of the corresponding shortest distance, wherein ∞ indicates infinitely great;
The shortest distance and incomplete shortest path of the source point to target point is calculated;
The algorithm MD-Outside-leaf is comprised the following processes:
Obtain source point and target point, two leaf nodes where source point and target point difference, and balance search tree;
The public ancestor node of minimum that two leaf nodes are found in balance search tree, finds two leaf segments in balance search tree
Most short node path between point;
The shortest distance and incomplete shortest path of the source point to target point is calculated;
The algorithm PathRecovery is comprised the following processes:
Obtain the shortest distance and imperfect shortest path of the source point to target point, and balance search tree and static road network figure;
Each edge in the imperfect shortest path of circular treatment judges whether the side is a line in static road network figure;
If the side is a line in static road network figure, which is added in complete shortest path, and from imperfect most short
The side is removed in path;
If the side is not a line in static road network figure, the iteration a certain vertex that each starting point with the side is connected, if meeting
The sum of the shortest path of terminal is equal on the side when this is arrived in the weight in starting point to a certain vertex and a certain vertex
This is then added to complete shortest path when starting point and a certain vertex connect and compose to the shortest distance of terminal by initial point
Summarize, by a certain vertex and this while terminal constitute while be added in imperfect shortest path, and from imperfect shortest path
The side is removed in diameter;
Recycle above-mentioned treatment process to imperfect shortest path for sky, obtain source point to target point complete shortest path.
10. paths planning method according to claim 8 or claim 9, which is characterized in that intelligent path planning mode includes as follows
Process:
Mist server obtains the source point that user inputs and target point and moment;
Mist server obtains the source point under static road network to target using the path planning process of basic path planning mode first
The shortest path of point;
An intelligent path is initialized, the shortest path under obtained static road network is assigned to intelligent path;
Based on the real-time road condition information inscribed when this, construction one is made of congestion sides all in the shortest path under static road network
Sequence, the shortest path between source point and target point is then adjusted with heuristics manner and is obtained after all congestion paths have adjusted
It to the final intelligent path of target point and path cost and is exported to source point to user.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910805967.XA CN110487293A (en) | 2019-08-29 | 2019-08-29 | A kind of efficient and privacy paths planning method based on extensive road network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910805967.XA CN110487293A (en) | 2019-08-29 | 2019-08-29 | A kind of efficient and privacy paths planning method based on extensive road network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110487293A true CN110487293A (en) | 2019-11-22 |
Family
ID=68554941
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910805967.XA Pending CN110487293A (en) | 2019-08-29 | 2019-08-29 | A kind of efficient and privacy paths planning method based on extensive road network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110487293A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113361788A (en) * | 2021-06-16 | 2021-09-07 | 东南大学 | Path planning method for multi-type service demand under urban environment |
CN113465612A (en) * | 2021-07-02 | 2021-10-01 | 南京航空航天大学 | Parallel path planning method and system based on double-layer index |
WO2023128868A3 (en) * | 2021-12-27 | 2023-09-07 | Grabtaxi Holdings Pte. Ltd. | System and method for generating a plurality of travel routes for area |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106548646A (en) * | 2016-11-08 | 2017-03-29 | 西安电子科技大学宁波信息技术研究院 | Road information service system and method when being blocked up based on the city that mist is calculated |
CN108259197A (en) * | 2016-12-28 | 2018-07-06 | 航天信息股份有限公司 | A kind of mist computing architecture model based on Internet of Things application demand |
CN108763287A (en) * | 2018-04-13 | 2018-11-06 | 同济大学 | On a large scale can traffic areas driving map construction method and its unmanned application process |
-
2019
- 2019-08-29 CN CN201910805967.XA patent/CN110487293A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106548646A (en) * | 2016-11-08 | 2017-03-29 | 西安电子科技大学宁波信息技术研究院 | Road information service system and method when being blocked up based on the city that mist is calculated |
CN108259197A (en) * | 2016-12-28 | 2018-07-06 | 航天信息股份有限公司 | A kind of mist computing architecture model based on Internet of Things application demand |
CN108763287A (en) * | 2018-04-13 | 2018-11-06 | 同济大学 | On a large scale can traffic areas driving map construction method and its unmanned application process |
Non-Patent Citations (1)
Title |
---|
QIN LIU ET AL: "Intelligent route planning on large road networks with efficiency and privacy", 《JOURNAL OF PARALLEL DISTRIBUTED COMPUTING》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113361788A (en) * | 2021-06-16 | 2021-09-07 | 东南大学 | Path planning method for multi-type service demand under urban environment |
CN113361788B (en) * | 2021-06-16 | 2022-11-01 | 东南大学 | Path planning method for multi-type service demand under urban environment |
CN113465612A (en) * | 2021-07-02 | 2021-10-01 | 南京航空航天大学 | Parallel path planning method and system based on double-layer index |
CN113465612B (en) * | 2021-07-02 | 2024-03-26 | 南京航空航天大学 | Parallel path planning method and system based on double-layer index |
WO2023128868A3 (en) * | 2021-12-27 | 2023-09-07 | Grabtaxi Holdings Pte. Ltd. | System and method for generating a plurality of travel routes for area |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Kaya et al. | A review on the studies employing artificial bee colony algorithm to solve combinatorial optimization problems | |
Cavalcante et al. | Roadside unit deployment for information dissemination in a VANET: An evolutionary approach | |
CN110487293A (en) | A kind of efficient and privacy paths planning method based on extensive road network | |
CN105608896B (en) | Traffic bottlenecks recognition methods in urban traffic network | |
CN108981739B (en) | Path planning method, device, server and storage medium | |
CN108683448B (en) | Influence node identification method and system suitable for aviation network | |
Ercsey-Ravasz et al. | Range-limited centrality measures in complex networks | |
CN109995580B (en) | VN mapping method based on GA _ PSO hybrid algorithm in 5G network slice | |
CN101900565A (en) | Path determining method and device | |
CN106202335A (en) | A kind of big Data Cleaning Method of traffic based on cloud computing framework | |
CN108924192B (en) | Optimal task scheduling method and system based on pseudo tree structure under data center network | |
CN113763700B (en) | Information processing method, information processing device, computer equipment and storage medium | |
CN107133704A (en) | Follow the analogy method of the dynamic emergency evacuation of large-scale crowd of optimal-forgetting rules | |
CN104636457B (en) | A kind of method and device of location finding cognition | |
CN110119150A (en) | A kind of multi-user's real-time route method and system for planning based on ant group algorithm | |
Khanfor et al. | A social IoT-driven pedestrian routing approach during epidemic time | |
CN107677277A (en) | A kind of determining method of path based on dijkstra's algorithm | |
CN112559807B (en) | Graph pattern matching method based on multi-source point parallel exploration | |
CN109359740A (en) | Motor-driven route optimal selection method and storage medium based on more habitat genetic algorithms | |
CN108847993A (en) | Link prediction method based on multistage path intermediate nodes resource allocation | |
CN112797997B (en) | Emergency path planning architecture and method based on grid road network | |
CN108804582A (en) | Method based on the chart database optimization of complex relationship between big data | |
Carmody et al. | Topological analysis of traffic pace via persistent homology | |
Iovanovici et al. | Network motifs uncovering hidden characteristics of urban public transportation | |
Tang et al. | Supporting continuous skyline queries in dynamically weighted road networks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191122 |
|
WD01 | Invention patent application deemed withdrawn after publication |