KR100944655B1 - Apparatus and method for constructing data and apparatus and method for retrieving data including non-spatial information - Google Patents

Abstract

PURPOSE: An apparatus and a method for constructing data and an apparatus and a method for retrieving data including non-spatial information are provided to search documents quickly by generating an index tree and an index bucket including data bucket and address information of data bucket which includes position information and aspatial information of a data object. CONSTITUTION: A data bucket generating unit(110) creates data bucket including the aspatial information. An index tree generating unit(120) creates an index tree by hierarchically arranging a terminal node and a nonterminal node. An index bucket generating unit(130) creates the index bucket including the terminal node information and the index bucket including the nonterminal node information and the address information. A multiplexer(140) creates encoded data. A time information output unit(150) changes the address information of the index bucket and data bucket based on the transfer rate of the encoded data into the time information.

Description

Apparatus and method for constructing data and apparatus and method for retrieving data including non-spatial information}

The present invention relates to a data generating apparatus and method and a data retrieval apparatus and method including non-spatial information, and more particularly, to a data object having position information and non-spatial information to be transmitted through a wireless broadcasting channel. An apparatus and method for generating a network and an apparatus and method for retrieving location information and non-spatial information about a data object from a data stream received through a wireless broadcasting channel.

As wireless networks become widespread and positioning technologies such as global positioning systems (GPS) emerge, location based services (LBSs) are emerging as one of the promising applications in mobile terminal environments. It became. This is a service that utilizes information about the location of a mobile terminal, so that information corresponding to a spatial query, which is a query related to the current location of the mobile terminal, can be provided.

The spatial queries that are the core of location-based services include range queries and k-nearest neighbor (k-NN) queries. The range query retrieves all target data that falls within q.R, the given spatial query range. For example, a range query can be expressed in the form of "search all restaurants located within a five-mile radius of the current location." In contrast, the most recent query retrieves one or more target data nearest to the given spatial query point q.P. The closest query can be expressed, for example, in the form of 'search for the three restaurants closest to the current location'.

As a method for efficiently providing target data corresponding to the spatial query to a user of a mobile terminal, a wireless data broadcasting method is commonly used. This method can accommodate multiple mobile terminals at the same cost and at the same cost in mobile terminal environment such as PDA or mobile phone, and it is effective to provide location based service because of its excellent bandwidth efficiency and scalability. It is a way.

In the wireless data broadcasting method, the server periodically spreads a plurality of data in a predetermined order through the wireless broadcasting channel. In addition, each mobile terminal accesses a broadcasting channel and continuously scans a broadcast data stream to search for target data. In this case, access time, which is a time for a mobile terminal to remain connected to a broadcasting channel to retrieve target data, and tuning time, which is a time required to download the retrieved target data from a data stream, The longer the energy consumption increases. In order to reduce such energy consumption, a method of storing a query and a location corresponding to the query in a data stream for searching the data is stored in an index structure and then included in the data stream for broadcasting. . According to this method, the mobile terminal can retrieve a query for retrieving data from an index, obtain a position on the data stream of data corresponding to the query, and obtain data at the corresponding position of the data stream.

A tree structure is used to create an index that is broadcast with the data stream. An R-tree structure is the most widely used index structure that includes location-related data. The R-tree is an index structure for storing multidimensional spatial data, and divides and stores the multidimensional space into a plurality of minimum bounding rectangles (MBRs). The mobile station accesses the broadcasting channel and retrieves data corresponding to the spatial query from the data stream including the index of the R-tree structure generated in the above manner.

This process is to create an index and search for the target data using only the location information of each spatial data. However, there may be cases where such spatial queries specify additional restrictions on non-spatial attribute values of the target data. For example, the latest contact is displayed in the form of 'search for the two closest French restaurants with an average price of $ 30 to $ 40'. Hereinafter, this case is defined as a generalized spatial query. The index generation method of the R-tree structure or the method of retrieving the target data from the generated index is difficult to apply in this generalized space query. Therefore, a need arises for a method of generating an R-tree index that can include non-spatial matters and retrieving target data corresponding to a generalized spatial query.

An object of the present invention is to provide a data generating apparatus and method including spatial information that enables the mobile terminal to efficiently retrieve data in terms of standby time and energy consumption through a wireless broadcasting channel.

Another technical problem to be solved by the present invention is to provide a program for executing a data generation method on a computer including spatial information that enables the mobile terminal to efficiently retrieve data in terms of latency and energy consumption through a wireless broadcasting channel. To provide a computer-readable recording medium for recording.

Another technical problem to be solved by the present invention is to reduce the time for accessing the wireless broadcasting channel and the time for downloading the data in retrieving the data stream received through the wireless broadcasting channel. It is to provide a data retrieval apparatus and method comprising a.

Another technical problem to be solved by the present invention is to search for a data stream received through a wireless broadcasting channel, thereby reducing the time for connecting to the wireless broadcasting channel and downloading the data, thereby reducing the latency and energy consumption. The present invention provides a computer-readable recording medium having recorded thereon a program for executing a data retrieval method including information.

In order to achieve the above technical problem, a data generating apparatus including spatial information according to the present invention includes data for generating a data bucket including non-spatial information representing characteristics of the data object corresponding to each of a plurality of data objects. Bucket generation unit; Logical sum of a terminal node comprising a data code encoding non-spatial information of the data objects and terminal node information including address information of data buckets corresponding to the data objects, and a data code included in a lower child node. An index tree generation unit for generating an index tree by hierarchically arranging non-terminal nodes including non-terminal node information including a node code calculated as a result; For each terminal node of the index tree, an index bucket including terminal node information corresponding to each terminal node is generated, and for each non-terminal node of the index tree, non-terminal information of each non-terminal node and each terminal An index bucket generator configured to generate an index bucket including address information of an index bucket corresponding to a child node of a horse node; A multiplexer configured to sequentially arrange the index bucket and the data bucket to generate data to be encoded; And a time information calculator for converting address information of the data bucket and the index bucket into time information based on the data amount of the data bucket and the index bucket and the transmission speed of the data to be encoded generated by the multiplexer. The data bucket generator and the index bucket generator change address information of the data bucket and the index bucket corresponding to the time information calculated by the time information calculator.

According to another aspect of the present invention, there is provided a data generation method including spatial information, the method comprising: generating a data bucket including non-spatial information representing characteristics of the data object corresponding to each of a plurality of data objects. Creating a data bucket; Logical sum of a terminal node comprising a data code encoding non-spatial information of the data objects and terminal node information including address information of data buckets corresponding to the data objects, and a data code included in a lower child node. An index tree generation step of generating an index tree by hierarchically arranging non-terminal nodes formed of non-terminal node information including a node code calculated as; For each terminal node of the index tree, an index bucket including terminal node information corresponding to each terminal node is generated, and for each non-terminal node of the index tree, non-terminal information of each non-terminal node and each terminal An index bucket generation step of generating an index bucket consisting of address information of an index bucket corresponding to a child node of a horse node; Multiplexing the data to be encoded by sequentially placing the index bucket and the data bucket; And a time information calculating step of converting address information of the data bucket and the index bucket into time information based on the data amount of the data bucket and the index bucket and the transmission speed of the data to be encoded generated by the multiplexer. In the data bucket generation step and the index bucket generation step, address information of the data bucket and the index bucket corresponding to the data bucket is changed to the time information calculated by the time information calculator.

In order to achieve the above technical problem, a data retrieval apparatus including spatial information according to the present invention includes position information indicating positions of a plurality of data objects in a predetermined coordinate space and non-space indicating characteristics of the data objects. A data receiver configured to receive a data stream including a data bucket including information and an index bucket generated corresponding to each node of the index tree including address information of the data bucket, from a transmission channel; The data bucket is extracted at a reception point determined by address information of the data bucket included in the index bucket extracted from the data stream, and if the index bucket corresponds to a non-terminal node of the index tree, the data bucket is extracted to the index bucket. A bucket extracting unit for extracting an index bucket corresponding to the child node at a reception point identified by address information of an index bucket corresponding to a child node of the non-terminal node; And a data extracting unit extracting positional information and non-spatial information of the data object from the data bucket, wherein the index buckets generated corresponding to the terminal nodes of the index tree are adjacent to each other in the coordinate space. And terminal node information including location information, a data code encoding non-spatial information of the adjacent data objects, and address information of data buckets corresponding to the adjacent data objects, and the terminal node information including non-terminal nodes of the index tree. The corresponding index bucket is calculated by a logical sum of the location information of the minimum bounding rectangle configured to include the data objects included in the terminal node existing in the lower layer in the coordinate space and the data codes included in the child node located below. Consisting of non-terminal node information including the node code All.

According to another aspect of the present invention, there is provided a data retrieval method including spatial information, comprising: location information indicating positions of a plurality of data objects in a predetermined coordinate space and a ratio indicating characteristics of the data objects. A data receiving step of receiving a data stream including a data bucket including spatial information and an index bucket generated corresponding to each node of the index tree including address information of the data bucket from a transmission channel; The data bucket is extracted at a reception point determined by address information of the data bucket included in the index bucket extracted from the data stream, and if the index bucket corresponds to a non-terminal node of the index tree, the data bucket is extracted to the index bucket. A bucket extraction step of extracting an index bucket corresponding to the child node at a reception point identified by address information of an index bucket corresponding to a child node of the non-terminal node; And a data extraction step of extracting location information and non-spatial information of the data object from the data bucket, wherein the index buckets generated corresponding to the terminal nodes of the index tree are adjacent to each other in the coordinate space. And terminal node information including location information, a data code encoding non-spatial information of the adjacent data objects, and address information of data buckets corresponding to the adjacent data objects, and the terminal node information including non-terminal nodes of the index tree. The corresponding index bucket is calculated by a logical sum of the location information of the minimum bounding rectangle configured to include the data objects included in the terminal node existing in the lower layer in the coordinate space and the data codes included in the child node located below. It consists of non-terminal node information including node code. Lose.

According to an apparatus and method for generating data including spatial information according to the present invention, a data bucket including location information and non-spatial information of a data object in a process of generating data to be encoded and transmitted to a mobile terminal through a wireless broadcasting channel. And by creating an index tree and the index bucket including the address information of the data bucket, it is possible to quickly search for documents containing the desired search words without downloading all the data from the mobile terminal. In addition, by copying and inserting the index bucket several times in the data to be encoded, the index bucket can be downloaded quickly without waiting until the next unit data reception time. In addition, according to the data retrieval apparatus and method including the spatial information according to the present invention, when retrieving the target data corresponding to the generalized spatial query from the data stream, the index bucket and the data bucket to the position information and non-spatial information of the data object By selectively extracting on the basis of this, it is possible to reduce the time for the mobile terminal to operate in the active mode and to reduce the energy consumption of the mobile terminal.

Hereinafter, with reference to the accompanying drawings will be described in detail a preferred embodiment of a data generating apparatus and method, and a data retrieval apparatus and method including spatial information according to the present invention.

1 is a block diagram showing the configuration of a preferred embodiment of a data generating device including spatial information according to the present invention.

Referring to FIG. 1, the data generating apparatus according to the present invention includes a data bucket generator 110, an index tree generator 120, an index bucket generator 130, a multiplexer 140, and a time information calculator ( 150).

The data bucket generation unit 110 generates a data bucket including non-spatial information representing characteristics of the data object corresponding to each of the plurality of data objects. In addition, the data bucket generated by the data bucket generator 110 may further include location information indicating the location of the data object in a predetermined coordinate space.

In order to enable the user terminal to search the target data by spatial query, the server providing the data should provide the location information of the data objects to be searched. In addition, it is necessary to provide non-spatial information of data objects in order to search the target data corresponding to the generalized spatial query in which limitations on non-spatial attribute values are specified when searching the target data. Therefore, the data bucket generator 110 generates a data bucket that can represent both spatial information and non-spatial information of the data object.

The data bucket is represented in the form of (dP, dA). Here, dP represents location information of a data object in a coordinate space, and can be expressed in a two-dimensional coordinate format or a three-dimensional coordinate format such as (dx, dy). In addition, dA represents non-spatial information indicating the characteristics of the data object and is represented by a group of non-spatial information expressed by da _i (1 ≦ _i ≦ m).

Next, the index tree generation unit 120 includes a terminal node including terminal node information including data codes encoding non-spatial information of data objects and address information of data buckets corresponding to adjacent data objects, and a lower node. A non-terminal node composed of non-terminal node information including a node code calculated by a logical sum of data codes included in a child node is hierarchically arranged to generate an index tree.

The terminal node information constituting the terminal node of the index tree may further include location information of data objects adjacent to each other in the coordinate space, and the non-terminal node information constituting the non-terminal node of the index tree is a lower layer in the coordinate space. It may further include location information of the minimum bounding rectangle set to include data objects included in the terminal node existing in the. Here, the position information of the minimum boundary rectangle may be expressed as coordinates of two vertices positioned on the diagonal of the minimum boundary rectangle in the coordinate space.

As described above, an index structure may be used to facilitate retrieval of data when the server transmits a data object through a wireless broadcasting channel so that a user can retrieve the target data. In addition, the R-tree structure is most widely used as an index structure including spatial information of a data object. Hereinafter, a process of generating an R-tree index structure using only spatial information excluding non-spatial information of a data object will be described.

FIG. 2 is a graph showing position information of a plurality of data objects located in a two-dimensional space, and FIG. 3 is a diagram illustrating an R-tree structure generated from the data objects shown in the graph of FIG.

Referring to FIG. 2, r ₁ to r ₈ are data objects each having two-dimensional coordinates, and minimum boundary squares N ₁ to N ₆ are generated based on their location information. In addition, each minimum boundary rectangle corresponds to each node of the R-tree shown in FIG. 3. The number of entries included in each node of the R-tree depends on the user's setting. In the following, it is assumed that each node of the R-tree has two entries.

First, two spatial data located at the closest distance among the spatial data of r ₁ to r ₈ belong to one minimum boundary rectangle to generate a leaf node of the R-tree. That is, is included in r ₁ and r ₂ is the terminal node information forming one minimum bounding rectangle N ₃ to the position information of N ₃ constituting the terminal node N ₃ of the R- tree, as illustrated in 2, r ₃ and r ₄ is formed in another one of the minimum bounding rectangle is included in the leaf node N ₄ and the information is position information of N ₄ constituting the leaf node N ₄ of R- tree. Terminal node information constituting the terminal nodes N ₅ and N ₆ of the R-tree is generated by the same method.

Next, to form two non-terminal nodes in the R-tree by four terminal nodes, four minimum boundary squares shown in the graph of FIG. 2 are combined two to form two minimum boundary squares. That is, the minimum bounding rectangle corresponding to N ₃ and N ₄ are included in a non-single end node information to form a minimum bounding rectangle N ₁ position information of N ₁ constituting the non-single end node N ₁ of the R- tree, N the minimum bounding rectangle for the ₅ and N ₆ forming the other of the minimum bounding rectangle of the N ₂ to be included in the non-single end node, the information is position information of N ₂ forming the other non-single end node N ₂ of the R- tree. Finally, since the root node of the R-tree may have two entries, the position information of the minimum boundary rectangles N ₁ and N ₂ finally formed is included in the non-terminal node information forming the root node of the R-tree.

The index tree generation unit 120 generates an index tree that may include non-spatial information of the data object based on the R-tree structure described above. In the present invention, the modified R-tree structure including both the location information and the non-spatial information is defined as a BSR-tree (Bucket-based Signature R-tree). Each node of the index tree having the BSR-tree structure includes non-spatial information of the data object in addition to the information contained in each node of the R-tree generated by the method described above.

First, a terminal node of an index tree includes terminal information of location information of adjacent data objects in a coordinate space, a data code encoding non-spatial information of adjacent data objects, and address information of data buckets corresponding to adjacent data objects. It consists of node information. At this time, the location information of the data object corresponds to d.P in the form of (d.P, d.A) representing a data bucket. In other words, it represents the coordinates of the data object in the coordinate space. In addition, the address information of the data bucket indicates a position in the data stream transmitted through the wireless broadcasting channel, and is converted by the time information calculator 150 to a transmission point through the broadcasting channel.

The data code is obtained by encoding dA according to a predetermined criterion. Non-spatial information of an object such as the type of restaurant or the average price point mentioned in the previous example is usually represented by a character, unlike location information that can be represented by simple coordinates, so a method of displaying the object is needed. As an example of a method of generating a data code, there is a method of generating using a hash function. That is, using the hash function, when m non-spatial information of da ₁ to da _m exists for a data object, the data code encoding non-spatial information dA is h ₁ (da ₁ ) + h ₂ (da ₂ ) + … + h _m (da _m ) This will be described in detail through specific examples.

4 is a diagram illustrating non-spatial information of a data object having location information in a coordinate space. Referring to FIG. 4, the data objects of r ₁ to r ₈ each represent a restaurant having location information, and _three non-spatial information included in each data object are da ₁ to da ₃ . da ₁ is a unique identification number for each restaurant, da ₂ is the type of restaurant, and da ₃ is the average price range. The hash function applied when generating a data code based on each of these non-spatial information is as follows.

Among the data objects showing the above hash function in Figure 4 on the basis of applying a hash function to the non-spatial information of the r _3, because it is the first identification of r ₃ No. 3 h ₁ is 01, r ₃ is so French restaurant h ₂ is 01, Finally, the average price of r ₃ is $ 35, so h ₃ is 100. Therefore, the data code generated for the non-spatial information of r ₃ becomes '01 + 01 + 100 = 0101100 '. In the same way, if a data code is generated for the non-spatial information of r ₈ , since h1 is 11, h2 is 01, and h3 is 100, it becomes '1101100'.

The non-terminal node of the index tree is generated based on the information of the child node located below instead of the information about the data bucket. Therefore, the non-terminal node of the index tree is configured to include data objects included in the terminal node existing in the lower layer in the coordinate space. Non-terminal node information including a node code calculated by the logical sum of the location information and the data codes included in the child nodes located below.

As described above, the minimum bounding rectangle refers to an area including a plurality of data objects in a coordinate space and corresponds to each node of the index tree. Since the method of generating the minimum bounding rectangle is the same as that described in the method of generating the R-tree with reference to FIGS. 2 and 3, a detailed description thereof will be omitted. However, the minimum boundary rectangle may not be generated based on the coordinates of the data objects as in the existing R-tree, but may be generated based on the non-spatial information of the data objects. In this case, the configuration of the information about the data objects included in the terminal node of the index tree may vary.

In addition, the node code is a code integrating all the non-spatial information of the data objects included in the child node located below the non-terminal node. For example, if the data codes included in the child nodes are '0101100' and '1101100', respectively, the node code calculated by the logical sum of them is '1101100'. FIG. 5 illustrates node codes generated for the minimum boundary rectangles N ₁ to N ₆ including data codes and data objects generated for the data objects r1 to r8 based on the non-spatial information of the data object shown in FIG. 4. .

As described above, an algorithm for generating an index tree based on the information included in the terminal node and the non-terminal node is shown in FIG. 6. Referring to FIG. 6, first, fanout f of each node constituting the index tree should be set. f should be set to a value much smaller than | D | which is the total number of data objects, and may be set by Equation 1 below.

Here, 'bucket capacity' refers to the data capacity of the data bucket and the index bucket, the capacity of all buckets is set equal. 'size of header information' refers to the data capacity of the header of the bucket, and the header of the bucket includes information indicating whether the bucket is a data bucket or an index bucket, the data capacity of the bucket, and the like. 'Size of an entry' refers to the data capacity of one entry included in one node of the index tree.

The index tree of the BSR-tree structure is formed sequentially from the terminal node to the root node in the same manner as in the case of the R-tree. The index tree generator 120 first generates a list (dlist) of data buckets generated by the data bucket generator 110 (rows 1 to 5), calculates a fanout f (row 6), and then index tree Run the PACK function that creates each node of the line (line 7).

개의 그룹으로 분류되는데, 이는 인덱스 트리의 단말노드의 개수와 동일하다(PACK의 1~2행). 또한 동일한 그룹에 속하는 데이터 버킷들은 인덱스 트리의 동일한 단말노드에 속하게 된다(PACK의 3~10행). 다음으로 인덱스 트리 생성부(120)는 단말노드정보, 즉 좌표공간에서의 최소경계사각형에 대한 위치정보들의 리스트(ilist)를 생성하고 다시 PACK 함수를 실행시켜(PACK의 19행) 단말노드의 상위에 위치하는 비단말노드들이 생성되도록 한다(PACK의 11~15행). 이러한 과정은 최종적으로 루트노드가 생성될 때까지 반복된다(PACK의 20~24행). 또한 각각의 노드에 포함될 수 있는 엔트리의 개수는 팬아웃 f보다 적게 설정되는 것이 바람직하다. 이러한 알고리즘을 사용한 일 예로, f를 2로 설정하고 각 노드의 엔트리 역시 두 개가 되도록 하였을 때 생성 된 인덱스 트리의 형태는 도 3에 도시된 R-트리의 구조와 동일하고, 다만 각 노드에 데이터 객체의 비공간정보인 데이터부호 또는 노드부호가 더 포함된다는 점이 상이하다.When the PACK function is executed, first the entire data bucket

It is classified into three groups, which is equal to the number of terminal nodes in the index tree (rows 1 and 2 of PACK). In addition, data buckets belonging to the same group belong to the same terminal node of the index tree (rows 3 to 10 of PACK). Next, the index tree generation unit 120 generates an ilist of terminal node information, that is, location information of the minimum boundary rectangle in the coordinate space, and executes the PACK function again (row 19 of the PACK) to the upper level of the terminal node. Causes non-terminal nodes to be created (lines 11-15 of the PACK). This process is repeated until the root node is finally created (lines 20-24 of the PACK). It is also desirable that the number of entries that can be included in each node is set less than the fanout f. As an example of using this algorithm, when f is set to 2 and each node has two entries, the shape of the index tree generated is the same as that of the R-tree shown in FIG. The difference is that the non-spatial information of the data code or node code is further included.

The index bucket generation unit 130 generates an index bucket including terminal node information corresponding to each terminal node for each terminal node of the index tree, and each non-terminal node for each non-terminal node of the index tree. An index bucket is created that includes end node information and address information of an index bucket corresponding to a child node of each non-terminal node.

The index bucket is created so that terminal node information or non-terminal node information included in each node of the index tree is included in the data stream transmitted through the wireless broadcasting channel. When generating the index tree, the terminal node is created first, and then the root node is finally generated by moving to the upper node. However, the transmission order of the created index buckets is reversed to the order of creation of each node of the index tree, and the index buckets generated from each node by the breadth-first search method are transmitted starting from the root node. Therefore, when the mobile station accesses the transport channel and retrieves data, the mobile station sequentially searches from the index bucket corresponding to the root node. To this end, the index bucket generated from the non-terminal node of the index tree should further include address information of the index bucket corresponding to the child node located below each non-terminal node in addition to the non-terminal node information. By doing so, it is possible to move directly to the node containing the target data without sequentially searching all nodes of the index tree, thereby reducing the search time of the data. Similar to the address information of the data bucket, the address information of the index bucket is also converted by the time information calculator 150 to a transmission point through the wireless broadcasting channel.

The multiplexer 140 sequentially generates an index bucket and a data bucket to generate data to be encoded.

The data buckets and index buckets created in the manner as described above are sorted and encoded in a certain order and then form a data stream to be sent to the wireless broadcasting channel. At this time, the user terminal receiving the data stream first extracts the index bucket to find out the address information of the data bucket including the target data corresponding to the spatial query, and then extracts the corresponding data bucket to obtain the target data therefrom. Therefore, when generating the data to be encoded and transmitted in the data generating apparatus according to the present invention, it is preferable to generate the data arranged in the order of the index bucket and the data bucket in consideration of the searching process on the user terminal side.

FIG. 7 illustrates data to be encoded generated by the multiplexer 140 using the data bucket and the index bucket based on the information of the data objects shown in FIG. 4. In FIG. 7, 'Root' is an index bucket corresponding to the root node of the index tree generated by the index tree generator 120, and N ₁ to N ₆ are terminal nodes and non-terminal nodes except for the root node of the index tree, respectively. Index buckets corresponding to and r ₁ to r ₈ represent data buckets generated from the data objects shown in FIG. 4, respectively. The sorting criteria of the index buckets may be configured to traverse each node of the index tree by a specific traversal method, for example, a breadth-first search method, so that the index buckets generated from each node may be sequentially sorted. In addition, in the case of the data bucket, it can be arranged in accordance with any sorting criteria, for example, according to the identification number assigned to the data object corresponding to each data bucket. As described above, one data to be encoded is generated by sequentially arranging all index buckets and data buckets by the multiplexer 140 as unit data. This unit data is encoded and repeatedly displayed on a data stream transmitted through a wireless broadcasting channel.

As described above, in order to receive data generated and encoded by the multiplexer 140 and transmitted through a wireless broadcasting channel at the user terminal and to search for target data corresponding to a spatial query, an index bucket must first be extracted. . However, when the index bucket is included only once in one unit data, the longest time to wait to extract the index bucket is a part of the index bucket and the time taken to transfer the data bucket. Thus, there is a need for an efficient data generation and retrieval system that reduces such latency.

One way to reduce the latency for extracting the index bucket is to insert the index bucket multiple times in the unit data. The unit data generated by this method is in the form of unit data shown in FIG. Referring to FIG. 7, it can be seen that an index bucket is inserted twice in one unit data. Looking at the specific process, first, the data bucket generator 110 generates a plurality of partial data buckets by dividing the data buckets by a preset index replication number, and the index bucket generator 130 generates an index bucket according to the number of index replications. Duplicate Finally, the multiplexer 140 sequentially and repeatedly arranges the index bucket and the partial data bucket to generate data to be encoded. Referring to FIG. 7, for example, when the index replication count is 2, the entire data bucket is divided into two groups, r1 to r4 and r5 to r8. Next, the entire index buckets of Root to N ₆ are replicated to create the same index buckets of Root 'to N ₆ '. The multiplexer 140 first sequentially arranges the index buckets of Root to N ₆ , and then arranges the data buckets corresponding to r1 to r4. Next, index buckets of Root 'through N ₆ ' are sequentially arranged, followed by data buckets corresponding to r5 through r8. By doing so, the mobile terminal can extract the index bucket from the middle portion of the unit data when retrieving the data, thereby reducing the waiting time. However, if the number of index replications is too large, the access time of the mobile terminal increases, so it is important to determine the optimal number of replications.

The time information calculator 150 converts address information of the data bucket and the index bucket into time information based on the data amount of the data bucket and the index bucket and the transmission speed of the data to be encoded generated by the multiplexer 140.

Among the index buckets generated by the index bucket generating unit 130, an index bucket corresponding to a terminal node of the index tree corresponds to address information of a data bucket, and an index bucket corresponding to a non-terminal node of the index tree corresponds to a child node. It contains the bucket's address information. This address information corresponds to location information based on the data amount of each bucket. If the address information included in the index bucket is composed of unit data by the multiplexing unit 140 by maintaining the state of the location information as it is, and encoded and transmitted, the mobile terminal directly searches the address information included in the buckets when searching for data. Since it is necessary to find the reception time of each bucket by converting the information, there is a problem that compatibility between the mobile terminal and the data generating apparatus according to the present invention is difficult to use. Therefore, when the server transmitting the data converts the address information into time information in advance, the mobile terminal can efficiently extract only the time information included in the transmitted bucket and download desired target data accordingly.

The time information calculator 150 first receives data amounts of the data bucket and the index bucket from the data bucket generator 110 and the index bucket generator 130, respectively. Next, since information on the number of times of index replication must be present, the amount of data of the entire unit data and the location of each bucket can be grasped so that the multiplexer 140 receives the number of indexes for configuring data to be encoded. In addition, since the data stream transmitted through the wireless broadcasting channel is transmitted at a predetermined constant transmission rate, address information may be converted into time information using these information.

When the time information of each bucket is calculated by the time information calculator 150, the data bucket generator 110 and the index bucket generator 130 input the calculated time information again to time the address information included in the bucket. Change to information. Therefore, the address information of the data bucket and the index bucket described above are all changed to time information and output. The buckets located in the data to be encoded generated by the multiplexing unit 140 include time information.

Since the data to be encoded includes time information of each bucket, the mobile terminal can know when the target data is received. Therefore, it is only necessary to download the data by switching the state of the mobile terminal to the active mode only when the target data is received. Accordingly, the time for operating the mobile terminal in the active mode is reduced, thereby reducing the time for accessing the wireless broadcasting channel and reducing the energy consumption required for data retrieval.

8 is a flowchart illustrating a process of performing a preferred embodiment of the data generation method according to the present invention.

Referring to FIG. 8, the data bucket generator 110 generates a data bucket including non-spatial information representing characteristics of the data object in response to each of the plurality of data objects (S810). In this case, the data bucket may further include location information indicating the location of the data object in a predetermined coordinate space. Next, the index tree generation unit 120 includes a terminal node including terminal node information including data codes encoding non-spatial information of adjacent data objects and address information of data buckets corresponding to the adjacent data objects, and a lower node. An index tree is generated by hierarchically arranging non-terminal nodes including non-terminal node information including a node code calculated by a logical sum of data codes included in a child node located at (S820). In addition, the terminal node information constituting the terminal node of the index tree may further include location information of data objects adjacent to each other in the coordinate space, and the non-terminal node information constituting the non-terminal node of the index tree may be located in a lower layer in the coordinate space. The location information of the minimum boundary rectangle set to include the data objects included in the existing terminal node may be further included. The generation of the index tree is performed by the method described above using the algorithm shown in FIG.

The index bucket generation unit 130 generates an index bucket including terminal node information corresponding to each terminal node for each terminal node of the index tree, and non-terminal node of the non-terminal node for each non-terminal node of the index tree. In operation S830, an index bucket including information and address information of an index bucket corresponding to a child node of each non-terminal node is generated.

The multiplexer 140 sequentially generates an index bucket and a data bucket to generate data to be encoded (S840). Finally, the time information calculator 150 converts the address information of the data bucket and the index bucket into time information based on the data amount of the data bucket and the index bucket and the transmission speed of the data to be encoded generated in the multiplexer 140. (S850). Accordingly, the data bucket generator 110 and the index bucket generator 130 change the address information of the data bucket and the index bucket corresponding to the time information calculated by the time information calculator 150. In addition, the data generating method according to the present invention may be performed by the data generating apparatus according to the present invention.

9 is a block diagram showing the configuration of a preferred embodiment of a data retrieval apparatus and method including spatial information according to the present invention, Figure 10 is a data retrieval apparatus and method including spatial information according to the present invention A block diagram showing the configuration of another preferred embodiment.

9, a data retrieval apparatus according to the present invention includes a data receiver 910, a bucket extractor 920, and a data extractor 930. 10, the data retrieval apparatus according to the present invention further includes a controller 940 and a data bucket storage unit 950 in addition to the data receiver 910, the bucket extractor 920, and the data extractor 930. do.

The data receiving unit 910 includes a data bucket including position information indicating positions of a plurality of data objects in a predetermined coordinate space and non-spatial information indicating characteristics of the data object, and an index tree including address information of the data bucket. Receives a data stream from the transport channel, the data stream including the index bucket generated corresponding to each node of the. The data stream received by the data receiver 910 may be a data stream generated and encoded by the data generating apparatus according to the present invention and transmitted through a wireless broadcasting channel.

The bucket extractor 920 extracts the data bucket at a reception point identified by the address information of the data bucket included in the index bucket extracted from the data stream, but when the index bucket corresponds to a non-terminal node of the index tree, the index bucket The index bucket corresponding to the child node is extracted at the reception point identified by the address information of the index bucket corresponding to the child node of the non-terminal node included in the.

As described above, an R-tree structure is mainly used as an index structure for data objects including location information. Therefore, when searching for the target data corresponding to the spatial query in the mobile terminal, a method of searching the data from the index of the R-tree structure is used. Hereinafter, a process of retrieving data using an index of an R-tree structure to process a spatial query including only spatial information excluding non-spatial information will be described.

11 is a diagram for searching for target data corresponding to a middle range query of a spatial query using an index of an R-tree structure. A range qR of spatial quality is indicated in FIG. 11, and the mobile terminal searches in turn from the root node of the R-tree to obtain target data, which is all spatial data included in qR. That is, since only the minimum boundary rectangle corresponding to N ₁ of the root node's entry overlaps with qR, the node moves to node N ₁ including the location information of minimum boundary rectangle N ₁ for the next search. Similarly, since the minimum boundary rectangles corresponding to N ₃ and N ₄ overlap with qR among the entries of node N ₁ , the mobile station moves to nodes N ₃ and N ₄ in order and performs the next search. Since nodes N ₃ and N ₄ are terminal nodes, an entry of a node becomes a pointer indicating a transmission time of spatial data belonging to a data stream. First, only spatial data r ₂ belongs to qR among the entries of N ₃ , and only spatial data r ₄ belongs to qR among the entries of N ₄ . Therefore, the target data corresponding to the range query becomes r ₂ and r ₄ , and the mobile terminal downloads the target data r ₂ and r ₄ from the data stream based on the transmission time of r ₂ and r ₄ obtained from the R-tree. .

Next, there are two methods of retrieving the target data corresponding to the nearest neighbor query using the R-tree: a best-first algorithm and an appearance-first algorithm. 12 is a diagram for searching for target data corresponding to a nearest neighbor query using an index of an R-tree structure. The mindist shown in the graph of FIG. 12 represents the shortest distance from the spatial query point qP to each minimum boundary rectangle. The highest priority search algorithm uses the mindist to search for the target data corresponding to the nearest query. For example, when the number of target data to be searched in the most recent query is 2, first, the mobile terminal puts all entries of the root node of the R-tree into the priority queue. At this time, the mindist between the minimum bounding square and qP corresponding to each entry also enters the priority queue. Next, the mobile station removes the node whose mindist from qP is the smallest among the entry of the root node stored in the priority queue from the priority queue, and moves to the node. That is, in the case of FIG. 12, since the distance to the minimum boundary square corresponding to N ₂ is the minimum, it moves to the node N ₂ including the location information of the minimum boundary square N ₂ , and next, all entries of N ₂ are priority queues. Add to Thus, the nodes currently stored in the priority queue are N ₁ , N ₅ and N ₆ . Node mindist is the minimum from the qP from the minimum bounding rectangle corresponding to the node N _1, so the mobile terminal must first remove the N ₁ from the priority queue, and the node N ₁ to yidonghan next entry of N ₃ the N ₁ N Add ₄ to the priority queue. Accordingly, the nodes stored in the priority queue are N ₃ , N ₄ , N _5, and N ₆ . Among them, because it is a minimum bounding rectangle for at least a mindist from qP corresponding to N ₄ the mobile station N _4, first removed from the priority queue and moves to node N ₄ first the entry of r ₃ and r ₄ of the N ₄ Rank To the queue. Thereto that by priority stored in the priority queue, _{N 3, r 3, r 4} , N 5 and N ₆ is in the mindist at least from qP may obtain r ₃ of r ₃ because one of the two target data. In order to acquire one more target data, the mobile station moves to N ₆ , which has a minimum mindist from qP among the components of the priority queue excluding r3. The entries of N ₆ are r ₇ and r ₈ , and among these, the minimum mindist from qP is r _7, so that another target data r ₇ can be obtained.

This first priority search algorithm has a problem that can increase the access time of the mobile terminal. FIG. 13 is a diagram illustrating a data stream to which a highest priority search algorithm for a nearest neighbor is applied. The index of the R-tree structure is broadcasted sequentially by breadth-first order for each node of the R-tree when broadcast over the channel. That is, as shown in FIG. 13, the order in which the index of the R-tree structure of FIG. 12 is broadcast is root node-N ₁ -N ₂ -N ₃ -N ₄ -N ₅ -N ₆ . Therefore, while waiting to be after the mobile terminal in the present example, the movement from the root node to the first N ₂ from above to move to the N ₁ are connected to the broadcast channel until the data stream in the following order to be broadcast. This raises the problem of increased access time. The situation-first search algorithm to solve the problems that may occur as each node of the R-tree is broadcasted sequentially searches for the target data considering not only the shortest distance but also the longest distance from qP to each minimum boundary rectangle. It is a way.

The above methods show a data retrieval method based only on location information without excluding non-spatial information of the data object. The data retrieval apparatus according to the present invention can utilize not only the positional information of the data object but also non-spatial information for the retrieval of the target data. First, a method of processing a generalized range query among spatial queries by the data retrieval apparatus according to the present invention will be described.

The generalized range query received from the user is expressed in the form (qR, qV). qR represents a query range in a coordinate space, and qV represents non-spatial information of target data to be searched. Corresponds to dA, which is non-spatial information of a data object expressed in the form of (dP, dA), and is represented by qv ₁ to qv _m (m is the number of non-spatial information) according to the number of non-spatial information that is a search condition. . When using the existing R-tree index tree (hereinafter, referred to as BR-tree) that does not include non-spatial information to search for target data corresponding to such a range query, it first refers to the query range qR. After extracting the index bucket and the data bucket included in the data stream, a method of selecting data buckets whose non-spatial information matches qV from the extracted data buckets may be used. However, this method has the disadvantage of increasing the synchronization time of the mobile terminal.

Since each node of the index tree, that is, the BSR-tree, generated by the index tree generating unit 120 of the data generating apparatus according to the present invention includes non-spatial information of a data object, an index bucket and a data bucket are extracted from the data stream. It is desirable to consider nonspatial information together during the extraction process.

In the data generating apparatus according to the present invention, the non-spatial information of the data object is encoded to generate a data code. The data searching apparatus according to the present invention receives a query code generated by encoding qV, which is non-spatial information, from a user. Can be. The query code is generated by the same method as the data code generation method using the hash function. That is, the query code is h ₁ (qv ₁ ) + h ₂ (qv ₂ ) +. h _m (qv _m ) According to the same example as in the method of generating the data code described above, when the query code is expressed in the form of '** 01100', '**' means that the identification number of the data object is not included in the search condition of the target data. '01' means that the target data is a French restaurant among the data objects. '100' also means that the data object with an average price range of $ 30 to $ 40 is the target data. In this way, the query code must be generated by the same method as the data code so that the number corresponding to each digit can be compared one-to-one, so that the target data can be easily retrieved.

The bucket extractor 920 sequentially reviews and extracts the index bucket and the data bucket included in the data stream received by the data receiver 910. The index bucket and the data bucket are sequentially arranged in the data stream as shown in FIG. 7, and the bucket extractor 920 starts searching for the target data from the index bucket corresponding to the root node. The first portion of an index bucket stream in which a plurality of index buckets are sequentially arranged, in which data transmitted when a data retrieval apparatus according to the present invention accesses a wireless broadcasting channel is traversed in a width-first manner to each node of an index tree. If not, wait until the next index bucket stream received. That is, the bucket extractor 920 always first extracts the index bucket corresponding to the root node of the index tree. The index bucket corresponding to the root node includes the location information of the minimum boundary rectangle, the node code calculated by the logical sum of the data codes included in the child node of the root node, and the address information of the index bucket corresponding to the child node. The bucket extractor 920 extracts the next index bucket using the location information of the minimum bounding rectangle and the node code.

First, the bucket extractor 920 compares the query range qR and the location information of the minimum bounding rectangle included in the index bucket, and when the minimum bounding crosses the querying range, an index generated from the node of the index tree corresponding to the minimum bounding rectangle. View the bucket as an extraction target. However, in addition to the non-spatial information, it is determined whether to extract the index bucket according to the comparison result of the query code and the node code. In the comparison between the query code and the node code, if the node code corresponding to the position indicated by 1 in the query code is set to 1, the index bucket corresponding to the node code is extracted. That is, if the logical sum of the query code and the node code matches the node code, the index bucket corresponding to the node code is extracted. For example, if the query code is '** 11011' and the node code is '0111111', each position of the node code corresponding to 1 of the query code is displayed as 1, so that '0111111', which is the logical sum of the query code and the node code, is Since the node code matches the node code, the query code and the node code have a partially-matched condition to extract the index bucket corresponding to the node code. However, if the node code is '0111100', the query code is displayed as 1 but the node code is 0 at the end of the node code. Therefore, the index bucket corresponding to the node code is not extracted.

In this way, the bucket extractor 920 extracts an index bucket by comparing the query range, the location information of the least bounded rectangle, the query code and the logical code, and then extracts the index bucket corresponding to the terminal node of the index tree. In the case of a bucket, the data bucket is extracted in the following order.

When extracting the data bucket, it is determined whether to extract the data bucket by comparing the query range, the location information in the coordinate space of the data object, and the query code and the data code. First, when the query range is compared with the location information of the data object, when the location information of the data object, that is, the coordinates of the data object in the coordinate space is included in the query range, a data bucket including the location information is extracted. However, if the query code and the data code are additionally compared and the two match, the bucket extractor 920 extracts the data bucket corresponding to the data code. For example, if the query code is '** 11011' and the data code is '0111011', since * is a don't care, the two are exactly the same. Therefore, the bucket extractor 920 extracts a data bucket corresponding to the corresponding data code. However, if the query code is '** 11011' and the data code is '0111111', the two do not match. However, even if the location information of the data object is included in the query range, the bucket extractor 920 stores the data corresponding to the data code. Do not extract the bucket.

In addition, the bucket extractor 920 may use a search list (search_list) and a result list (result_list) that store the extraction result of the bucket to extract the index bucket and the data bucket. 14 is a diagram illustrating an algorithm for processing a generalized range query using a search list and a result list. Referring to FIG. 14, first, a mobile terminal accesses a wireless broadcasting channel, and a bucket extractor 920 extracts an index bucket corresponding to a root node from a data stream received by the data receiver 910 to enter an entry of a root node. To the search list (line 2). That is, the location information of the minimum bounding rectangle included in the index bucket corresponding to the root node is added to the search list. Next, the bucket extractor 920 sequentially examines the index buckets transmitted through the data stream, and extracts the corresponding index buckets and determines whether to add information included in the index buckets to the search list. In addition, the information in the search list is deleted when the bucket extraction unit 920 examines the corresponding index bucket, and the bucket extraction process is repeated until all the information in the search list is deleted.

If the index bucket to be examined is an index bucket created from a non-terminal node of the index tree, the bucket extractor 920 determines that the location information of the minimum boundary rectangle included in the index bucket under review intersects the query range, and the node code and query If the logical sum of the codes coincides with the node code, the index bucket corresponding to the node code is extracted and the information about it is added to the search list (lines 11-16). In addition, when the index bucket to be reviewed is an index bucket created from a terminal node of the index tree, the bucket extractor 920 includes location information of a data object included in the index bucket under review in the query range, and the data code corresponds to the query code. If there is a match, the data bucket corresponding to the data code is extracted and the information is added to the result list (lines 17-22).

Finally, the data extractor 930 extracts a data object from the extracted data bucket. In the case of using the algorithm shown in FIG. 14, a data object is extracted from a data bucket stored in a result list. That is, location information and additional non-spatial information on the coordinate plane of the data object included in the data bucket are extracted. The extracted data object is the target data corresponding to the generalized range query to be searched in the mobile terminal.

Hereinafter, the above process will be described in detail with reference to the data objects shown in FIG. 5 and the data stream shown in FIG. 7. The index tree generated based on the data objects shown in FIG. 5 has the same form as that shown in FIG. 3. It is assumed that the query range is set equal to qR shown in FIG. 11 and that the query code is '** 11011'. First, the bucket extractor 920 extracts an index bucket corresponding to the root node of the index tree, and adds information of the minimum boundary rectangles, ie, N ₁ and N ₂ , included in the extracted index bucket to the search list. Next, the bucket extractor 920 examines N ₁ and N ₂ in order to determine whether to extract the index bucket corresponding thereto. The minimum bounding rectangle corresponding to N ₁ intersects the query range, and the logical sum of '0111111', which is the node code of N ₁ , and the query code, coincide with the node code. Thus bucket extraction unit 920 extracts a bucket index corresponding to N _1, and adds an entry of information of N ₃ and N ₄ N ₁ of the search list. In the case of N ₂ , since the minimum bounding rectangle does not intersect the query range, the index bucket corresponding to N ₂ is not extracted. The search list now contains N ₃ and N ₄ . The minimum bounding rectangle for the N ₃ is a cross-range and quality, but logical node code is "0010010" and the quality of the code N ₃ is inconsistent with the code node. Therefore, the bucket extractor 920 does not extract the index bucket corresponding to N ₃ . In case of N ₄ , the minimum bounding rectangle intersects the query range, and the logical sum of the node code '0111111' and the query code coincides with the node code. Accordingly, the bucket extractor 920 extracts the index bucket corresponding to N ₄ and examines the data buckets corresponding to the data objects r ₃ and r ₄ , which are the entries. Since the location information corresponding to r ₃ among r ₃ and r ₄ is included in the query range, '0111011', which is the data code of r ₃ , matches the query code, so that the bucket extractor 920 stores the data bucket corresponding to r ₃ . Extract and add information about it to the result list. Finally, the data extractor 930 extracts the data object from the data bucket stored in the result list.

In another form of generalized spatial query, a method different from generalized range query is used to retrieve the target data corresponding to the most recent query. To this end, the data retrieval apparatus according to the present invention further includes a control unit 940 and a data bucket storage unit 950. In addition, the most recent query is expressed in the form of (q.P, q.V), where q.P represents a query point and q.V represents a condition for nonspatial information of the target data.

The data bucket storage unit 950 stores address information of the same number of data buckets as the number of queries input from the user in correspondence with the distance between the data object included in each data bucket in the coordinate space and the query points input from the user. do. In addition, the controller 940 determines that the distance between the data object and the query point in the coordinate space is smaller than the maximum value between the data object and the query point stored in the data bucket storage unit 950, and the query code received from the user If it matches, the address information of the data bucket corresponding to the data code is stored in the data bucket storage unit 950.

In the process of searching for the target data corresponding to the generalized recent query by the data searching apparatus according to the present invention, the data receiving unit 910 is the data bucket and the index as in the search for the target data corresponding to the generalized range query. Receive a data stream containing a bucket from a transport channel.

The bucket extractor 920 also extracts the data bucket at the reception point identified by the address information of the data bucket included in the index bucket extracted from the data stream. However, when the data retrieval apparatus according to the present invention is used for the processing of the closest query, the data bucket storage unit 950 is further provided so that the bucket extraction unit 920 extracts the data buckets. ) Is extracted by the address information of the data bucket stored in). This will be described in detail below.

In the data stream received by the data receiving unit 910, the index bucket and the data bucket generated by the data generating apparatus according to the present invention are sequentially arranged. The bucket extractor 920 first extracts the index bucket corresponding to the root node of the index tree as in the case of the generalized range query, and sequentially examines the index buckets transmitted later. At this time, the first portion of the index bucket stream in which a plurality of index buckets sequentially generated when data transmitted when the data retrieval apparatus according to the present invention accesses wireless broadcasting is traversed in a width-first manner in each node of the index tree. If not, wait until the next index bucket stream received.

When extracting the index bucket from the bucket extractor 920, unlike the case of the generalized range query, the location information is not considered. Therefore, if the logical sum of the node code and the query code matches the node code, the index bucket corresponding to the node code is extracted. The method of comparing node code with query code is the same as that of generalized range query. After extracting the index bucket corresponding to the terminal node of the index tree, the data bucket is extracted. In this case, the control unit 940 operates to store address information of the data bucket to be extracted in the data bucket storage unit 950.

If the distance between the data object and the query point in the coordinate space is smaller than the maximum value of the distance between the data object and the query point stored in the data bucket storage unit 950 and the data code matches the query code, the data code Address information of the data bucket corresponding to the data bucket storage unit 950 is stored. Since the distance between the data object and the query point is both represented by the coordinates of a specific point in the coordinate space, the distance between the two points can be calculated. In addition, when the number of address information of the data bucket stored in the data bucket storage unit 950 is less than the number of queries, the maximum value of the distance between the data object and the query point is reported as infinity (∞) to correspond to the data code that has been examined. The address information of the data bucket is stored in the data bucket storage unit 950. The bucket extractor 920 extracts the data bucket at a reception point identified by the address information of the data bucket stored in the data bucket storage unit 950.

The reason for using a separate device for storing address information of the data bucket in the case of the generalized recent contact is that of the recent contact. The nearest query is of the form "search for k data objects that are located at the shortest distance from the query point". Therefore, it is not possible to determine whether to extract the data bucket according to the address information of the data bucket included in the index bucket until all index buckets corresponding to the terminal node of the index tree are examined. The data bucket storage unit 950 stores address information of the same number of data buckets as the number of queries so that the extraction target is completed when the bucket extraction unit 920 finishes examining all index buckets corresponding to the terminal node of the index tree. Ensure that the data bucket to be established is

In the case of generalized nearest neighbor query, search list can be used when processing query like generalized range query, and candidate list which is a list of address information of data bucket stored in data bucket storage unit 950 instead of result list. You can use (candidate_list). 15 is a diagram illustrating an algorithm for processing a generalized nearest query using a search list and a candidate list. Referring to the algorithm of FIG. 15, the generalized processing of the recent contact by the data retrieval apparatus according to the present invention will be described in detail.

First, when the data receiver 910 receives a data stream from the wireless broadcasting channel, the bucket extractor 920 extracts an index bucket corresponding to the root node of the index tree from the data stream, and then applies the minimum corresponding to the entry of the root node. Add bounding rectangle information to the search list (line 2). This is identical to the generalized range processing. Next, the bucket extractor 920 sequentially determines the index bucket corresponding to the node of the index tree indicated by the minimum bounding rectangle stored in the search list, and determines whether to extract the index bucket. Whether the index bucket is extracted or not is determined by comparing the node code and the query code, and if the logical sum of the node code and the query code does not match the node code, the bucket extractor 920 informs the minimum boundary square information corresponding to the node code. Delete from the search list (paragraphs 7-9). If the logical sum of the node code and the query code corresponding to the minimum bounding rectangle stored in the search list matches the node code, the process of extracting the index bucket corresponding to the node code and adding the entry of the node to the search list is repeated. Line 11).

In detail, when the index bucket corresponds to a non-terminal node of the index tree, as described above, the bucket extractor 920 compares the node code and the query code, and if the logical sum of the two matches the node code, the corresponding node. Extract the index bucket corresponding to the sign (lines 12-17). However, when the index bucket corresponds to a terminal node of the index tree, since the index bucket includes information about the data bucket, a review process for the index bucket is performed by the controller 940. That is, the distance between the data object and the query point in the coordinate space based on the location information of the data object included in the index bucket is smaller than the maximum value of the distance between the data object and the query point stored in the data bucket storage. If and the query code match, the address information of the data bucket corresponding to the data code is stored in the data bucket storage unit 950, that is, the candidate list (lines 18 to 24). At this time, the number of address information of the data bucket that can be stored in the candidate list is equal to the number of queries.

As in the case of the generalized range query, the bucket extraction unit 920 or the control unit 940 deletes the information stored in the search list for the index bucket that has been examined. Therefore, when all the information stored in the search list is deleted, all index buckets have been reviewed. Therefore, the candidate list, that is, the address information of the data bucket stored in the data bucket storage unit 950, is stored in the address information of the target data. Corresponding. Therefore, the bucket extractor 920 extracts the data bucket according to the address information of the data bucket stored in the data bucket storage 950.

Hereinafter, a process of the generalized recent concatenation described above by the data objects shown in FIG. 5 and the data stream shown in FIG. 7 will be described in detail. The index tree generated based on the data objects shown in FIG. 5 has the same form as that shown in FIG. 3. It is assumed that the query point is set equal to qP shown in FIG. 12, and the query code is '** 01100' and k = 2, that is, the number of questions is two. The bucket extractor 920 first extracts an index bucket corresponding to the root node of the index tree from the data stream received by the data receiver 910. Then add the root node's entries, N ₁ and N _2, to the search list. Node code corresponding to N ₁ is "0111111", and since the logical sum of the query code matches the node marks bucket extraction unit 920 to the node of the index tree for extracting index bucket corresponding to N _1, and corresponds to the N ₁ Add the included entries N ₃ and N ₄ to the search list. At the same time N ₁ is deleted from the search list. Node code for the N ₂ is '1111111', and also stored in nodes bucket extracting unit 920 because the logical OR of the query code matches the node code is extracted index bucket corresponding to N _2, and corresponds to the N ₂ Add the entries N ₅ and N ₆ to the search list. Therefore, the index bucket currently stored in the search list is {N ₃ , N ₄ , N ₅ , N ₆ }. Next, in the case of N ₃ , since the node code is '0010010', the logical sum with the control code does not match the node code. Also, in the case of N ₅ , since the node code is '1010101', the logical sum with the control code does not match the node code. Therefore, the bucket extractor 920 does not extract index buckets corresponding to N ₃ and N ₅ . Since the node code corresponding to N ₄ is '0111111' and the node code corresponding to N ₆ is '1101110', both the logical sum of the control code matches the node code. Therefore, the bucket extractor 920 extracts index buckets corresponding to N ₄ and N ₆ .

Next, since the index buckets corresponding to N ₄ and N ₆ are those corresponding to the terminal nodes of the index tree, the control unit 940 examines them. First, the data code of r ₄ among the entries of N ₄ nodes does not match the query code, but only the data code of r ₃ matches the query code. In addition, the address information of the currently stored data bucket does not exist in the data bucket storage unit 950. Therefore, the controller 940 stores the address information of the data bucket corresponding to r ₃ in the data bucket storage unit 950 without considering the distance between the data object and the query point in the coordinate space. Next, only '1101100', which is the node code of r ₈ , among the entries of the N ₆ nodes matches the query code, and since the address information of the data object stored in the data bucket storage unit 950 is less than 2, which is the number of queries, the controller 940. Adds address information of the data bucket corresponding to r ₈ to the data bucket storage unit 950.

The bucket extractor 920 extracts the data buckets corresponding to r ₃ and r ₈ at a reception point determined by the address information of the data buckets corresponding to r ₃ and r ₈ stored in the data bucket storage unit 950. The data extractor 930 extracts a data object from the extracted data bucket. The extracted data object becomes the target data corresponding to the generalized nearest query.

16 is a flowchart illustrating a process of performing a preferred embodiment of the data retrieval method according to the present invention.

Referring to FIG. 16, the data receiver 910 may include address information of a data bucket and data bucket including location information indicating positions of a plurality of data objects and non-spatial information indicating characteristics of the data object in a predetermined coordinate space. A data stream including an index bucket generated corresponding to each node of the included index tree is received from the transport channel (S1610). Next, the bucket extractor 920 extracts the data bucket at the reception point identified by the address information of the data bucket included in the index bucket extracted from the data stream to perform a generalized range query having a format of (qR, qV). Extract. In this case, when the index bucket corresponds to the non-terminal node of the index tree, the index bucket corresponding to the child node is extracted at the reception point identified by the address information of the index bucket corresponding to the child node of the non-terminal node included in the index bucket. do. When the index bucket is extracted, if the minimum boundary rectangle formed in the coordinate space intersects the query range received from the user and the logical sum of the node code and the query code matches the node code, the index bucket corresponding to the node code is extracted (S1620). . When extracting the data bucket, if the location information of the data object is included in the query range and the data code and the query code match, the data bucket corresponding to the data code is extracted (S1630). Finally, the data extractor 930 extracts a data object from the extracted data bucket (S1640). Such generalized quality processing may be performed by the data retrieval apparatus according to the present invention.

17 is a flowchart illustrating a process of performing another preferred embodiment of the data retrieval method according to the present invention.

Referring to FIG. 17, the data receiver 910 may include address information of a data bucket and data bucket including location information indicating positions of a plurality of data objects and non-spatial information indicating characteristics of the data object in a predetermined coordinate space. A data stream including an index bucket generated corresponding to each node of the included index tree is received from the transport channel (S1710). Next, the bucket extractor 920 receives the data bucket at the reception point identified by the address information of the data bucket included in the index bucket extracted from the data stream to perform a generalized recent query having a format (qP, qV). Extract In this case, when the index bucket corresponds to the non-terminal node of the index tree, the index bucket corresponding to the child node is extracted at the reception point identified by the address information of the index bucket corresponding to the child node of the non-terminal node included in the index bucket. do. When extracting the index bucket, if the logical sum of the node code and the query code matches the node code, the index bucket corresponding to the node code is extracted (S1720). Before extracting the data bucket, the controller 940 determines that the distance between the data object and the query point in the coordinate space is smaller than the maximum value between the data object and the query point stored in the data bucket storage, and the data code and the query code match. In operation S1730, address information of the data bucket corresponding to the data code is stored in the data bucket storage unit 950. The bucket extracting unit 920 extracts the data bucket at the reception point determined by the address information of the data bucket stored in the data bucket storage unit 950 (S1740). Finally, the data extractor 930 extracts a data object from the extracted data bucket (S1750). This generalized process of recent contact can also be performed by the data retrieval apparatus according to the present invention.

와 같이 정의된다. 이때 DATA_size는 전체 데이터 객체들의 데이터량이고, INDEX_size는 인덱스 트리의 데이터량이다. 데이터량의 단위는 버킷으로 나타내어진다.Experiments were conducted to confirm the performance by applying the data generating apparatus and method, and the data searching apparatus and method according to the present invention to a real system. The system is implemented to include servers, mobile terminals, and broadcasting channels, a UNIFORM dataset consisting of 6000 data objects uniformly distributed in coordinate space, and a REAL dataset consisting of 5922 data objects based on a real restaurant. Each was experimented using. Non-spatial information assigned to each data object is a restaurant identification number (a ₁ ), a restaurant type (a ₂ ), and an average price range (a ₃ ). There are two generalized forms of spatial query for these data objects. The first is the form of defining only a1 and a2 (indicated by q ₁ in the graph), and the second is the form of defining both a ₁ to a ₃ (indicated by q ₂ in the graph). Also, in the generalized range query, the ratio of the width of the query range to the width of the actual coordinate space is defined as SizeRatio. The data amount for each minimum bounding rectangle is set to 16 bytes, and the data amount of the data code, node code, and query code is set to 2 bytes for the address bucket and the address information of the data bucket. M, the number of index replications to reduce the access time of the mobile terminal,

Is defined as: DATA _size is the data amount of all data objects, and INDEX _size is the data amount of the index tree. The unit of data amount is represented by a bucket.

First, we review the results of the generalized range query. FIG. 18 is a graph showing the length of the tuning time for the SizeRatio measured for the UNIFORM dataset and the REAL dataset according to the format of the spatial query and the shape of the index tree. FIG. 19 is a diagram for the UNIFORM dataset and the REAL dataset. This graph shows the length of the access time for the SizeRatio according to the spatial query type and the index tree type. Units of tuning time and access time are expressed in number of buckets. The BSR-tree is a case where nonspatial information of a data object is included in an index tree and the BR-tree is a case where nonspatial information is not included in an index tree. 18 and 19, the data amount of the bucket is fixed at 256 bytes, and the SizeRatio is changed from 0.05 to 0.2. In both the UNIFORM and REAL datasets, the tuning time and access time increase as the SizeRatio increases.

FIG. 20 is a graph showing the lengths of the tuning time and the access time according to the data amount of the bucket measured for the UNIFORM data set. Although not shown in FIG. 20, the same results are obtained for the REAL data set. Referring to Fig. 20, the SizeRatio is fixed at 0.1 and the bucket data amount is changed from 64 to 512. In this case, regardless of the generalized range query format (q ₁ or q ₂ ) and the structure of the index tree, as the amount of data in the bucket increases, the tuning time and access time decrease, resulting in better data retrieval performance.

In addition, since the tuning time and the access time are shorter in the case of using the BSR-tree index than in the case of the BR-tree index in FIGS. I found out that it is good to use it.

Next, review the experimental results for the generalized recent cognate. FIG. 21 is a graph showing the length of the tuning time for the number of queries (k) measured for the UNIFORM data set and the REAL data set according to the format of the spatial query and the form of the index tree. FIG. 22 is a UNIFORM data set and the REAL. This is a graph showing the length of access time for the number of queries (k) measured for a dataset according to the form of spatial query and index tree. 21 and 22, the amount of data in the bucket is fixed to 256 bytes, and the number of queries is increased from 1 to 9. In the graph of FIG. 21, the use of the index tree of the BSR-tree structure shows a much shorter tuning time than the case of using the BR-tree. This is because, when the BR-tree is used, non-spatial information is not included in the index tree and an index bucket corresponding to all nodes must be extracted without comparing the node code and the query code. In FIG. 22, the case where the index tree of the BR-tree structure is used is excluded. This is because the use of the BR-tree shows a significantly longer access time, for example, an access time of about 34,000 buckets when the number of queries is 1 compared with the use of the BSR-tree.

FIG. 23 is a graph showing the tuning time and the access time length according to the data amount of the bucket measured for the UNIFORM data set. As in the case of Fig. 20, the same result is obtained for the REAL data set. Referring to Fig. 23, the number of queries k is fixed at 5, and the data amount of the bucket is changed from 64 to 512 bytes. As a result of the experiment, it can be seen that as the data volume of the bucket increases, the tuning time and access time decrease, thereby improving the performance of data retrieval. Also, the use of BSR-tree shows shorter tuning time and access time than that of BR-tree. In particular, in the case of using the BR-tree, when the bucket size is 512 bytes, the access time of about 20,000 buckets is shown, which is not illustrated in FIG. 23.

Referring to these experimental results, it can be seen that the data generating apparatus and method including the spatial information according to the present invention, and the data searching apparatus and method are more efficient in terms of standby time and energy consumption than the conventional methods. .

The invention can also be embodied as computer readable code on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, and the like, and may also be implemented in the form of a carrier wave (for example, transmission over the Internet). Include. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

Although the preferred embodiments of the present invention have been shown and described above, the present invention is not limited to the specific preferred embodiments described above, and the present invention belongs to the present invention without departing from the gist of the present invention as claimed in the claims. Various modifications can be made by those skilled in the art, and such changes are within the scope of the claims.

1 is a block diagram showing the configuration of a preferred embodiment of a data generating apparatus including spatial information according to the present invention;

2 is a graph showing a plurality of spatial data located in a two-dimensional space;

3 is a diagram illustrating an R-tree structure generated using spatial data shown in the graph of FIG. 2;

4 is a diagram illustrating non-spatial information of a data object having location information in a coordinate space;

5 is a diagram illustrating a data code generated for r1 to r8 and a node code generated for N1 to N6 based on the non-spatial information of the data object shown in FIG. 4;

6 is a diagram illustrating an algorithm for generating an index tree based on information included in a terminal node and a non-terminal node;

7 is a diagram illustrating data to be encoded generated by a multiplexer using a data bucket and an index bucket based on information of data objects shown in FIG. 4;

8 is a flowchart illustrating a process of performing a preferred embodiment of the data generation method according to the present invention;

9 is a block diagram showing the configuration of a preferred embodiment of a data retrieval apparatus and method including spatial information according to the present invention;

10 is a block diagram showing the configuration of another preferred embodiment of a data retrieval apparatus and method including spatial information according to the present invention;

11 is a diagram for searching for target data corresponding to a range query using an index of an R-tree structure;

12 is a diagram for searching for target data corresponding to a nearest neighbor query using an index of an R-tree structure;

FIG. 13 is a diagram showing a data stream to which the highest priority search algorithm for the nearest neighbor is applied; FIG.

14 is a diagram illustrating an algorithm for processing a generalized range query using a search list and a result list.

15 is a diagram illustrating an algorithm for processing a generalized nearest query using a search list and a candidate list;

16 is a flowchart illustrating a process of performing a preferred embodiment of a data retrieval method according to the present invention;

17 is a flowchart illustrating a process of performing another preferred embodiment of a data retrieval method according to the present invention;

18 is a graph showing the length of the tuning time for the SizeRatio measured for the UNIFORM and REAL datasets according to the format of the spatial query and the shape of the index tree;

19 is a graph showing the length of the access time for the SizeRatio measured for the UNIFORM and REAL datasets according to the format of the spatial query and the shape of the index tree;

20 is a graph showing the tuning time and the access time length according to the data amount of the bucket measured for the UNIFORM data set;

FIG. 21 is a graph showing the length of synchronization time for the number of queries (k) measured for UNIFORM and REAL datasets according to the form of spatial query and index tree;

22 is a graph showing the length of the access time for the number of queries (k) measured for the UNIFORM data set and the REAL data set according to the format of the spatial query and the index tree;

FIG. 23 is a graph showing the tuning time and the access time length according to the data amount of the bucket measured for the UNIFORM data set.

Claims (18)

A data bucket generation unit for generating a data bucket including non-spatial information representing characteristics of the data object corresponding to each of a plurality of data objects; Logical sum of a terminal node comprising a data code encoding non-spatial information of the data objects and terminal node information including address information of data buckets corresponding to the data objects, and a data code included in a lower child node. An index tree generation unit for generating an index tree by hierarchically arranging non-terminal nodes including non-terminal node information including a node code calculated as a result; For each terminal node of the index tree, an index bucket including terminal node information corresponding to each terminal node is generated, and for each non-terminal node of the index tree, non-terminal information of each non-terminal node and each terminal An index bucket generator configured to generate an index bucket including address information of an index bucket corresponding to a child node of a horse node; A multiplexer configured to sequentially arrange the index bucket and the data bucket to generate data to be encoded; And And a time information calculator for converting address information of the data bucket and the index bucket into time information based on the data amount of the data bucket and the index bucket and the transmission speed of the data to be encoded generated by the multiplexer. And the data bucket generator and the index bucket generator change address information of the data bucket and the index bucket corresponding to the time information calculated by the time information calculator. The method of claim 1, The data bucket generator generates a data bucket further including location information indicating a location of the data object in a predetermined coordinate space. The index tree generation unit is configured to include a terminal node including terminal node information further including location information of data objects adjacent to each other in the coordinate space, and data objects included in terminal nodes existing in a lower layer in the coordinate space. And generating an index tree by hierarchically arranging non-terminal nodes formed of non-terminal node information including position information of a minimum boundary rectangle. The method according to claim 1 or 2, The data code and the node code is characterized in that the binary representation. The method according to claim 1 or 2, The data bucket generation unit generates a plurality of partial data buckets by dividing the data buckets by a preset index replication number, The index bucket generation unit replicates the index bucket according to the number of times of index replication, And the multiplexing unit sequentially and repeatedly arranges each of the index buckets and the partial data buckets to generate data to be encoded. Generating a data bucket corresponding to each of a plurality of data objects, the data bucket including non-spatial information representing characteristics of the data objects; Logical sum of a terminal node comprising a data code encoding non-spatial information of the data objects and terminal node information including address information of data buckets corresponding to the data objects, and a data code included in a lower child node. An index tree generation step of generating an index tree by hierarchically arranging non-terminal nodes formed of non-terminal node information including a node code calculated as; For each terminal node of the index tree, an index bucket including terminal node information corresponding to each terminal node is generated, and for each non-terminal node of the index tree, non-terminal information of each non-terminal node and each terminal An index bucket generation step of generating an index bucket consisting of address information of an index bucket corresponding to a child node of a horse node; Multiplexing the data to be encoded by sequentially placing the index bucket and the data bucket; And And a time information calculating step of converting address information of the data bucket and the index bucket into time information based on the data amount of the data bucket and the index bucket and the transmission speed of the data to be encoded generated in the multiplexing step. The data bucket generation step and the index bucket generation step, characterized in that for changing the address information of the data bucket and the index bucket corresponding to each of the time information calculated by the time information calculator. The method of claim 5, In the data bucket generation step, generating a data bucket further comprising location information indicating the location of the data object in a predetermined coordinate space, In the index tree generation step, a terminal node including terminal node information further including position information of data objects adjacent to each other in the coordinate space, and data objects included in terminal nodes existing in a lower layer in the coordinate space. And generating an index tree by hierarchically arranging non-terminal nodes formed of non-terminal node information including position information of a minimum boundary rectangle. The method according to claim 5 or 6, The data code and the node code is characterized in that the binary representation. The method according to claim 5 or 6, In the data bucket generation step, the data bucket is divided by a preset index replication number to generate a plurality of partial data buckets, In the index bucket generation step, the index bucket is replicated according to the number of times of index replication, In the multiplexing step, the data generation method comprising generating the data to be encoded by sequentially placing each of the index buckets and the partial data buckets sequentially. Corresponding to each node of the index tree including location information indicating the location of each of the plurality of data objects in the predetermined coordinate space and non-spatial information indicating the characteristics of the data object and address information of the data bucket. A data receiver configured to receive a data stream including a generated index bucket from a transport channel; The data bucket is extracted at a reception point determined by address information of the data bucket included in the index bucket extracted from the data stream, and if the index bucket corresponds to a non-terminal node of the index tree, the data bucket is extracted to the index bucket. A bucket extracting unit for extracting an index bucket corresponding to the child node at a reception point identified by address information of an index bucket corresponding to a child node of the non-terminal node; And And a data extractor configured to extract positional information and non-spatial information of the data object from the data bucket. The index bucket generated corresponding to the terminal node of the index tree corresponds to position information of data objects adjacent to each other in the coordinate space, a data code encoding non-spatial information of the adjacent data objects, and the adjacent data objects. Terminal node information including address information of data buckets The index bucket generated corresponding to the non-terminal node of the index tree is included in the location information of the minimum boundary rectangle and the child node located below the minimum boundary rectangle configured to include data objects included in the terminal node existing in the lower layer in the coordinate space. And non-terminal node information including a node code calculated by a logical sum of the obtained data codes. The method of claim 9, The bucket extractor indexes a bucket corresponding to the node code when the minimum boundary rectangle formed in the coordinate space intersects the query range received from the user and the logical sum of the node code and the query code input from the user matches the node code. And extracting a data bucket corresponding to the data code when the location information of the data object is included in the query range and the data code matches the query code. The method of claim 9, Data bucket storage unit for storing the address information of the data bucket of the same number of queries input from the user corresponding to the distance between the data object included in each data bucket in the coordinate space and the query point received from the user ; And When the distance between the data object and the query point in the coordinate space is smaller than the maximum value between the data object and the query point stored in the data bucket storage unit and the data code matches the query code input from the user And a controller configured to store address information of the data bucket corresponding to the data code, in the data bucket storage unit. The bucket extractor extracts an index bucket corresponding to the node code when the logical sum of the node code and the query code matches the node code, and is received by the address information of the data bucket stored in the data bucket storage unit. And extracting the data bucket at a point in time. The method according to any one of claims 9 to 11, Receiving the next index bucket stream when the index bucket does not correspond to the first portion of the index bucket stream in which a plurality of index buckets are generated by traversing each node of the index tree in a breadth first manner. A data retrieval apparatus, characterized by waiting until a point in time. Corresponding to each node of the index tree including location information indicating the location of each of the plurality of data objects in the predetermined coordinate space and non-spatial information indicating the characteristics of the data object and address information of the data bucket. Receiving a data stream including a created index bucket from a transmission channel; The data bucket is extracted at a reception point determined by address information of the data bucket included in the index bucket extracted from the data stream, and if the index bucket corresponds to a non-terminal node of the index tree, the data bucket is extracted to the index bucket. A bucket extraction step of extracting an index bucket corresponding to the child node at a reception point identified by address information of an index bucket corresponding to a child node of the non-terminal node; And And a data extraction step of extracting location information and non-spatial information of the data object from the data bucket. The index bucket generated corresponding to the terminal node of the index tree corresponds to position information of data objects adjacent to each other in the coordinate space, a data code encoding non-spatial information of the adjacent data objects, and the adjacent data objects. Terminal node information including address information of data buckets The index bucket generated corresponding to the non-terminal node of the index tree is included in the location information of the minimum boundary rectangle and the child node positioned below the minimum boundary rectangle configured to include data objects included in the terminal node existing in the lower layer in the coordinate space. And non-terminal node information including a node code calculated by a logical sum of the data codes. The method of claim 13, In the bucket extracting step, if the minimum boundary rectangle formed in the coordinate space intersects the query range input from the user and the logical sum of the node code and the query code input from the user coincides with the node code, the node code corresponds to the node code. Extracting an index bucket, and extracting a data bucket corresponding to the data code when the location information of the data object is included in the query range and the data code and the query code match. The method of claim 13, Address information of the same number of data buckets as the number of queries input from the user is stored in the data bucket storage corresponding to the distance between the data object included in each data bucket in the coordinate space and the query point received from the user. A storage step; and When the distance between the data object and the query point in the coordinate space is smaller than the maximum value between the data object and the query point stored in the data bucket storage unit and the data code matches the query code input from the user And storing the address information of the data bucket corresponding to the data code in the data bucket storage unit. In the bucket extracting step, if the logical sum of the node code and the query code matches the node code, the index bucket corresponding to the node code is extracted and identified by the address information of the data bucket stored in the data bucket storage unit. And extracting the data bucket at the time of reception. The method according to any one of claims 13 to 15, Receiving the next index bucket stream when the index bucket does not correspond to the first portion of the index bucket stream in which a plurality of index buckets are generated by traversing each node of the index tree in a breadth first manner. A data retrieval method comprising waiting until a point in time. A computer-readable recording medium having recorded thereon a program for executing the data generating method according to claim 5 or 6. A computer-readable recording medium having recorded thereon a program for executing the data retrieval method according to any one of claims 13 to 15.

Images