KR101797207B1 - Method of Processing Moving Object Trajectory Data With User Defined Functions - Google Patents

Abstract

The present invention relates to a method to process trajectory data of a moving object based on a user defined function (UDF), capable of enhancing a query function about the trajectory data. According to the present invention, the method comprises the following steps: using a trajectory table storing trajectory data, a trajectory segment table storing sequentially generated trajectory segments, and a meta table storing metainformation of each trajectory to construct a database storing generated trajectory data in the predetermined number of trajectory segment sets; generating a temporary table with mpids (moving point identification) of the trajectory segment table satisfying a query condition when the occurrence of a user query event is detected; generating a generalized search tree (GiST) index dataset from the trajectory segment table and searching for a candidate record set satisfying the query condition from the GiST index dataset; and examining whether or not an actual record exists through a refining step and returning a true/false value depending on the existence of a matched record.

Description

Field of the Invention < RTI ID = 0.0 > [0001] < / RTI &

The present invention relates to a UDF-based moving object trajectory data processing method, and more particularly, to a modeling method for storing and managing large-capacity moving object trajectory data, a UDF-based trajectory index method and a query line for improving trajectory data query performance. And a UDF-based moving object trajectory data processing method implemented by an instantiation table technique.

The present invention has been made from the research carried out as part of the research project of the national land space information research project of the Ministry of Land, Infrastructure and Transport Technology (Project No: 1615007698 (80795), Project title: Development of open source based application link technology for activation of spatial information industry).

Recently, various mobile devices such as smart phones and tablet computers have become popular due to the development of mobile related technologies. In addition, due to the development of mobile computing environment, attempts to integrate existing technology with advanced technology and provide new services are continuing. Especially, as the spread of mobile devices equipped with GPS becomes active, various application services using location information can be provided. In the early days of location-based services (LBS), only the location information was interested, so other additional information was not received much attention. However, recent location information based services use location data and time data together to analyze past, present and future correlations.

For example, an intelligent navigation service that accumulates and analyzes the GPS coordinates of a real-time traffic information service or a moving object providing real-time road traffic status and surrounding situation information, provides an optimal road guidance service, And a carpool service that matches people traveling on similar routes at certain times such as commuting time. However, there is a limit to providing these services quickly and accurately with location information alone.

Therefore, time attributes should be considered together to analyze historical data and continuous data. The trajectory is a set of moving object location information acquired in successive times. Also, due to the nature of the moving object trajectory data, even if only a few days are collected, a large data set is generated. Such large-volume trajectory data is not suitable for processing in a general database. Therefore, a system model for storing and managing the trajectory of the moving object and a method for quickly querying the large moving object data are needed.

DOMINO (Database fOr MovINg Objects) proposed moving object space-time model (MOST) to support moving objects in DBMS. This MOST model stores the direction vector to obtain the current position of the moving object. We also proposed a FTL (Future Temporal Logic) query language that can predict the near future location. HERMES, developed as an extension of the Oracle DBMS, provides operators that enable spatiotemporal queries on ORDBMS to support spatiotemporal queries in Oracle. The main goal of HERMES is to support continuous modeling and querying of moving objects. HERMES defines a set of data types and their corresponding operators. HERMES 'ORDBMS layer has been enhanced to fit the internal structure query capabilities for storing the trajectory data and supporting LBS in the Oracle ORDBMS server. Guting et al proposed a data model for storing both past and present data in an ORDBMS. We also proposed an abstract data type that can represent all segments with an infinite set of curves and a discrete data type that can be represented by a finite number of polylines. SECONDO defines a moving object data type and operators by extending a spatial object, and proposes an algorithm for implementing these operators. In addition, SECONDO supports BerlinMOD, a benchmark system for moving object databases.

These conventional researches focus on modeling methods for storing and managing the trajectory data in a general DBMS. However, since the trajectory data is continuously collected at regular intervals, some applications may create a large data set. Therefore, it is necessary to study methods for efficiently querying large-volume trajectory data sets. Therefore, it is required to develop a new and more efficient technology for storing and managing large-capacity moving object trajectory data.

SUMMARY OF THE INVENTION The present invention has been made to overcome the above problems, and it is an object of the present invention to provide a modeling method for storing and managing large-capacity moving object trajectory data, a UDF-based trajectory index technique, And to provide a UDF-based moving object locus data processing method.

According to an aspect of the present invention, there is provided a method of processing UDF-based trajectory data of a UDF-based trajectory data system, including: a mobject trajectory table for storing trajectory data; A trajectory segment table for storing trajectory segments generated by the trajectory segments, and a trajectory column for storing meta information of the trajectories, and stores the generated trajectory data in a predetermined number of trajectory segment sets Establishing a database for the database; Generating a GiST (Generalized Search Tree) index dataset from a Trajectory Segment Table; Searching a candidate record set satisfying a query condition in the generalized search tree (GiST) index data set and checking whether there is an actual record through a refine step; And returning a true / fulse value according to the presence or absence of a matching record.

The UDF-based trajectory data processing method of the UDF-based trajectory data system according to another embodiment of the present invention includes a trajectory table for storing trajectory data, a trajectory segment table for storing sequentially generated trajectory segments Constructing a database for storing generated trajectory data as a set number of trajectory segments using a trajectory column capable of storing meta information of each trajectory; Generating a temporary table (MT) with mpid (Moving Point IDentify) of a Trajectory Segment Table satisfying a query condition when detecting the occurrence of a user query event; Generating a generalized search tree (GiST) index data set from a trajectory segment table and searching for a candidate record set satisfying a query condition in the generalized search tree (GiST) index data set; And a refine step to check the presence / absence of the actual record, and returning a true / fulse value according to the presence / absence of the matching record.

In addition, in the present invention, the locus segment set is generated in the form of a Tpoint (Time Point) type array composed of point points at which a moving object moves and times at the instant.

In the present invention, the GiST index data set includes a Consistent (E, q) function for checking a query condition and a Union (P) function for merging input entries, a penalty function for determining a node position of an inserted entry E, E2), a PickSplit (P) function that determines node partitioning, a Compress (E) function that changes the input entry into a storable format, and a Decompress (E) function that is an inverse of the Compress .

A UDF-based trajectory data processing method of a UDF-based trajectory data system according to another embodiment of the present invention includes a trajectory table for storing trajectory data, a trajectory segment table for storing sequentially generated trajectory segments A database for storing generated trajectory data as a set number of trajectory segments using a trajectory column capable of storing meta information of each trajectory; Executing a selected UDF function module according to a user request; And the UDF function module scans a Trajectory Segment Table to retrieve a record whose mood of the object table matches the query condition cond, And returning a true / fulse value.

According to the UDF-based trajectory data processing method of the UDF-based trajectory data system of the present invention, it is possible to store and manage the trajectory of the moving object through a more efficient UDF-based trajectory indexing method and a query line instantiation table technique, The effect can be obtained.

FIG. 1 is a block diagram of a trajectory table for storing trajectory data (mobject_trajtable), a trajectory segment table for storing trajectory segments sequentially generated, and a metacat table (trajectory_column) for storing meta information of each trajectory FIG.
FIG. 2 is a diagram showing a query processing process of a simple locus UDF.
3 is a diagram illustrating a UDF-based trajectory index query process.
4 is a diagram showing a query processing process of a query line substantiation strategy.
5 is a diagram showing the execution time of each query when the size of the data set increases.
6 is a diagram showing a query execution time according to the query window size.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

In the UDF-based trajectory data processing method of the UDF-based trajectory data system of the present invention, a method of generating a 3D R-tree index based on a UDF using an indexing method of large volume trajectory data and a temporary view table are generated to improve the query performance, Is implemented. User-Defined Functions (UDFs) used in the present invention are user-defined functions, which are created by a user using SQL or by C-level code such as C language, And the functions that are installed and built in. In other words, it refers to a method in which a user operates as an API (application programming interface) capable of extending a DBMS without touching the DBMS source code. These UDF functions can be used within the DBMS as standard SQL commands.

UDF-based trajectory system model

1 shows a trajectory table (mject_trajtable) for storing trajectory data, a trajectory segment table (trajectory segment table) for storing sequentially generated trajectory segments, and a metagatable (trajectory_column) for storing meta information of each trajectory have.

The locus table (mobject_trajtable) is a user-defined table in which the user can inquire about the locus information. In fact, the user can inquire about the locus information using only this user defined table.

Trajectory Segment Table (Trajectory Segment Table) is a table that is automatically generated when a user defined table is created. This table stores the trajectory of moving objects. Each moving object has its own Moving Point IDentify (mpid). Since a large number of data is inserted into each mpid, a segid can be created to manage it. One segid will store the position information of the number specified in the addTrajectory Column (size of tpseg). rect is information to represent the rectangle surrounding locations in a segid. You can also create next_segid and before_segid to manage each segid like a double linked list.

The meta table (trajectory_column) is a meta table that contains overall information about each table, including the name of the trajectory_column specified by the user and the name of the automatically generated trajectory_segtable. It has the unique moid of the object and the oid of the table to store the trajectory of the object.

The simplest way to store the locus data is to store one position coordinate and one time value in a row. However, storing data in this way can cause some problems. In other words, inserting the trajectory data periodically generated one by one in a row unnecessarily increases the capacity of stored data, and when the trajectory data is inquired, the set of data returned as a query result becomes larger, And may cause performance degradation. Therefore, in the present invention, it is possible to store the trajectory data as a set of the set of trajectory segments when the trajectory data is stored. The trajectory segment set is created as a Tpoint (Time Point) type Array composed of the point at which the moving object moves and the time at that moment, thereby reducing the number of query results in the trajectory data query, do.

The following table shows how to create a table creation query and a trajectory column for this trajectory system model and a user query on the trajectory data.

CREATE TABLE taxi

(taxi_id int, taxi_number text,

taxi_model text, taxi_driver text);

SELECT addTrajectoryColumn

('public', 'taxi', 'traj', 4326, 'MOVINGPOINT', 2, 150);

The above table creation query creates an object table that can store the meta information of the moving object. Then, the next query adds a trajectory column to the generated object table, and creates a trajectory segment table to store and manage actual trajectory data. In this case, the parameters of the function include the scheme of the database, the name of the object table to which the trajectory column is added, the name of the trajectory column, the SRID of the spatial data, the data type, the dimension of the data type, Indicates the number.

SELECT taxi_id, taxi_number FROM taxi

WHERE TJ_Passes (traj, TJ_BOX (116.35, 39.93, 116.22, 40.14,

PERIODS (' 2008-02-02 13:30:44', '2008-02-02 15:54:46')));

The user query retrieves the ID and number information of the moving objects that passed through the specified time period. In this way, the user can query the space-time locus data.

Simple trajectory UDF technique (Naive)

The simple trajectory UDF (User-Defined Functions) query checks whether a record satisfies the query condition in the condition of the trajectory query. For the simple trajectory UDF query, it is implemented as functions that can express the phase relation of space time trajectory data by extending the phase relation functions of two dimensional spatial data. TJ_Passes () function to check the trajectories passing through a specific space and TJ_Inside () function to check the existence of trajectories staying in a specific space are the functions that are used to express the phase relation of the trajectories.

In the trajectory system model according to the present invention described above, a query for a moving object is made by joining an object table storing meta information of the moving object and a trajectory segment table storing the trajectory information of the moving object. At this time, the user queries the trajectory UDF function provided to the user without directly joining the two tables. The UDF function scans the Trajectory Segment Table to retrieve records that match the moid of the Object Table and the query condition (cond), and returns a true / false value depending on whether the matching record exists or not.

FIG. 2 shows a query processing process of a simple locus UDF. The simple trajectory UDF query has a disadvantage in that it costs a lot of query processing because it scans the entire recordset. Accordingly, the present invention provides a UDF-based trajectory index scheme and a query line substantiation table scheme for reducing the query processing cost.

UDF-based Trajectory Indexing Technique (GiST-R-NT)

The locus data is basically three-dimensional data in which two-dimensional spatial data and one-dimensional time data are combined. In the present invention, a 3D R-tree index that is an extension of the GiST (Generalized Search Tree) index is implemented as an index algorithm for the locus data, and the query processing time is compared using the implemented 3D R-tree index. GiST is a template tree that supports easy implementation of B-tree and R-tree or its transformation tree. In other words, to implement an R-tree index, the engineer can only implement at least six key functions. Union (P) function to merge input entries with Consistent (E, q) function to check query condition, Penalty (E1, E2) function necessary to determine node position of inserted entry, A function called PickSplit (P), a Compress (E) function that changes the input entry into a storable format, and a Decompress (E) function that is an inverse of the Compress (E) function.

3 shows a UDF-based trajectory index query process. The UDF-based trajectory index query uses the trajectory phase function used in the simple trajectory UDF query described above. In the query process, the candidate record set satisfying the query condition is searched in the GiST index data set generated from the trajectory segment table, and the presence / absence of the actual record is checked through a refine step. It returns a true / false value like a simple trajectory UDF query. PostgreSQL / PostGIS not only supports GiST-based 2-D R-tree, but also provides functions that can be extended to n-dimensional R-tree. In the present invention, a 3D R-tree is implemented by extending the n-dimensional R-tree support functions of the GiST index supported by PostgreSQL / PostGIS.

Query line substantiation table technique (GiST-R-MT)

In order to improve the retrieval performance of a large amount of data, generally, an index is created in a data set to be searched. In the present invention, a query line substantiation table technique may be implemented as another method for improving the search performance.

4 shows a query processing process of a query line substantiation strategy.

A query materialization table technique is a technique known as a materialized view. It is a method of physically generating a temporary view table when a query is first requested, and maintaining a temporary view table generated on the assumption that other queries will follow the view. In order to keep the contents of the view up-to-date, an efficient way to automatically update the view table when the base table is updated should be considered. The concept of incremental update was developed for this purpose, which determines which tuples should be inserted, deleted, and updated in the materialized view table when the change is applied to one of the defined base tables. Normally, this view is maintained for the duration of the query. If there is no query in the view for a certain period of time, the system automatically deletes the physical view table, and then reconstructs the view table from the beginning when there is a query referring to that view.

In the query processing method of the query line substantiation strategy, when there is a user query, a temporary table (MT) is created with the mpids of the trajectory segment table satisfying the query condition. Then, it checks whether the mpid of the records that refine the candidate record set satisfying the query condition exists in the generated temporary table like the UDF based trajectory index query, and then returns true / false according to the result of the check.

Experiment environment

As a UDF-based trajectory index, experiments on query processing performance were performed using 3DR-tree. In order to improve the query performance of the present invention, the performance of each technique is compared and analyzed by using the proposed query materialization table technique. The experiment was performed on a computer with an Intel Core i5 3.2GHz CPU and 8GB memory, and Linux version Ubuntu 14.04 was used as the operating system. The data used in the experiments are GPS locus data collected from taxis in Beijing, China to make a service to guide the route with fast, accurate, fast and safe route using large amount of GPS data collected from taxis. Were collected. This data is collected about six months from about 30,000 taxis.

Parameter Setting Page size 4k Node capacity 50, 100, 150 , 200, 250 Query window size 2%, 4%, 6% , 8%, 10% Number of queries 200 Dataset size 100K, 300k, 500K , 700k, 1M

Table 1 shows the parameters used in the experiments to measure query performance. Experiments were carried out on the user trajectory query for the three query methods described in Section 4.

Performance evaluation

Figure 5 shows the execution time of each query when the size of the dataset increases. In the experiment, the simple trajectory UDF query (Naive) increases the query execution time linearly as the data set increases, while the UDF-based trajectory index query method (GiST-R-NT) ) Can see that the query execution time increases slowly as the data set increases. In particular, it can be seen that the query substantiation technique shows a performance improvement of about 1.2 times the execution time of the UDF-based trajectory index query.

6 shows a query execution time according to the query window size. In this experiment, all three queries show a slow execution time as the query window size increases. When the query window size increases from 2% to 4%, the execution time variation is the largest because the number of data rows retrieved when the window size is 2% is relatively small and the execution time is also relatively small.

As described above, the present invention provides a UDF-based trajectory index technique and a query line instantiation table technique that can store and manage the trajectory of a moving object and quickly query large-capacity moving object data. In the performance evaluation, Performance is the result of the experiment.

The UDF-based trajectory data processing method according to the present invention may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions recorded on the medium may be those specially designed and configured for the present invention or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

While the invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the equivalents of the claims, as well as the claims.

Claims (6)

A trajectory table for storing trajectory data, a trajectory segment table for storing sequentially generated trajectory segments, and a trajectory column for storing meta information of each trajectory, Constructing a database for storing generated trajectory data as a set number of trajectory segments;
Generating a GiST (Generalized Search Tree) index dataset from a Trajectory Segment Table;
Searching candidate record sets satisfying a query condition in the GiST (Generalized Search Tree) index data set and checking whether there is an actual record through a refine step; And
Steps to return a true / fulse value depending on the presence or absence of matching records
Wherein the UDF-based trajectory data processing method comprises the steps of: A trajectory table for storing trajectory data, a trajectory segment table for storing sequentially generated trajectory segments, and a trajectory column for storing meta information of each trajectory, Constructing a database for storing generated trajectory data as a set number of trajectory segments;
Generating a temporary table (MT) with mpid (Moving Point IDentify) of a Trajectory Segment Table satisfying a query condition when detecting the occurrence of a user query event;
Generating a generalized search tree (GiST) index data set from a trajectory segment table and searching for a candidate record set satisfying a query condition in the generalized search tree (GiST) index data set; And
A step of refining the true record through the refine step and returning a true / fulse value according to the presence / absence of a matching record
Wherein the UDF-based trajectory data processing method comprises the steps of: 4. The method according to any one of claims 1 to 3,
Wherein the trajectory segment set is generated in a form of a Tpoint (Time Point) type array consisting of a point at which the moving object has moved and a time at the instant of time. The UDF-based trajectory data processing of the UDF- Way. 4. The method according to any one of claims 1 to 3,
The GiST index data set includes a Union (P) function for merging input entries with a Consistent (E, q) function for checking query conditions, a Penalty (E1, E2) function for determining a node position of an inserted entry, (E) function, which is an inverse function of a Compress (E) function, and a PickSplit (P) function for determining a node partition, a Compress (E) function for changing an input entry into a storable format, and a Decompress UDF - based trajectory data processing method based on trajectory data system. A trajectory table for storing trajectory data, a trajectory segment table for storing sequentially generated trajectory segments, and a trajectory column for storing meta information of each trajectory, Maintaining a database storing the generated trajectory data as a predetermined number of trajectory segment sets;
Executing a selected UDF function module according to a user request; And
The UDF function module scans the Trajectory Segment Table to retrieve records whose moids match the query condition (cond) of the object table, and if true, / Steps that return a true / fulse value
Wherein the UDF-based trajectory data processing method comprises the steps of: delete

Images