CN116108592A - Evaluation method for building comprehensive hub area system model based on tod concept - Google Patents

Abstract

The invention belongs to the technical field of comprehensive junction sheet area assessment by utilizing a tod concept, in particular to an assessment method for building a comprehensive junction sheet area system model based on the tod concept. According to the method for evaluating the system model establishment of the comprehensive junction area based on the tod concept, the purposes of carrying out data fitting and demonstration on the track tod site evaluation system by multi-source data are achieved, when a user needs to go out, the user can obtain the destination of the user needing to go out through the traditional data analysis, the traditional traffic analysis research method is changed by combining the multi-source big data with the advanced classification model method, a set of multi-dimensional, multi-layer and all-dimensional evaluation system is established, and the adaptability of the track tod development of the comprehensive junction area can be deeply analyzed from the aspects of land, traffic, economy, population relations and the like.

Description

Evaluation method for building comprehensive hub area system model based on tod concept

Technical Field

The invention relates to the technical field of comprehensive junction sheet area assessment by utilizing a tod concept, in particular to an assessment method for building a comprehensive junction sheet area system model based on the tod concept.

Background

tod (Transit-Oriented-Development) is a "public transportation Oriented" Development model. The concept is proposed by the new city sense representing the character Biget-Karsol, and is a kind of walk urban area taking public transportation as the center and comprehensively developing to solve the unrestricted spread of the United states city after the second battle. The public transportation is mainly rail transportation such as subways, light rails and the like and bus trunk lines, and then urban areas integrating work, business, culture, education, living and the like are built by taking bus stops as centers and 400-800 m (5-10 min walking path) as radiuses. So as to realize an organic coordination mode of compact development of each city group. tod is an internationally representative urban community development model. Meanwhile, the method is also one of the most representative modes of newcastle disease. Is widely used in urban development, especially in areas where cities are not developed in pieces, public transportation is introduced by soliciting land for planning development areas at a lower price in advance, time difference for developing land price is formed, then 'prepared land' with perfect infrastructure is sold, and the government recovers the prior investment of public transportation from the return of land rise value.

The existing TOD concept is developed by public transportation commission, can not be combined with a mobile data network to change traditional user travel traffic analysis, and can not be used for deeply analyzing adaptability of TOD development of comprehensive hub area track sites from the aspects of land, traffic, economy, population relations and the like, so an evaluation method based on TOD concept for building a comprehensive hub area system model is needed.

Disclosure of Invention

Based on the prior art, the invention provides an evaluation method for building a comprehensive hub area system model based on the tod concept.

The invention provides an evaluation method for building a comprehensive hub area system model based on a tod concept, which comprises the following steps of firstly, collecting, counting and analyzing access data of a user by utilizing a network;

inputting the collected, counted and analyzed user data into a high-level classification model for classification;

training the statistical data;

and step four, storing the training data and waiting for calling.

Preferably, the method of collecting in the first step is to access data by a user existing in a mobile network, the data having a fixed travel route and a fixed travel time, get on from a fixed station, take on a high-speed rail, a vehicle with a fixed subway route, and get off at a fixed station;

taking a cell 1-a cell 5 used by a user around the existing traffic route as an example; frequent signaling interaction is needed between the eNB and the UE in the UE moving process, so that the problems of large signaling overhead, long transmission interruption time or low throughput period, excessive PDCCH scheduling times, PDCCH resource consumption and high UE power consumption are caused; thus, the cell to be used by the UE is preconfigured through the collection, statistics and analysis of the user behaviors by the mobile network.

Preferably, the method for collecting, counting and analyzing the user behavior by the network is divided into a centralized type and a distributed type, and an independent user information collecting module, a counting module and an analyzing module are added on the centralized type by utilizing the network side; the module is connected with a plurality of eNBs, each eNB transmits the collected user information to the module, and the module is responsible for comprehensively collecting the user information and carrying out statistics and analysis; the module obtains that the user regularly passes through the cells 1-5 in sequence through statistics and analysis; the statistical mode comprises the steps that a module counts the using times of the cell; the module notifies the relevant eNB of the statistics result and transmits the statistics result to the eNB1 and the eNB5; the module makes a pre-judgment on the behavior of the user, and when the user takes a high-speed rail vehicle, the running route of the user is fixed, so that the network makes a pre-judgment on a cell which the user taking the high-speed rail is about to arrive; the basis for the pre-determination includes the cell in which the user is located at this time, or a series of serving cells that have been previously passed.

Preferably, the distribution is through exchanging the history information of switching of UE and history information of cell configuration between eNBs; the eNB records a cell used by a user, and when the user changes the cell, the source eNB transmits switching history information of the UE or history information of the cell use to the target eNB; when the UE changes from cell 1 to cell 2, eNB1 will notify eNB2 of the history information that cell 1 was serving as the serving cell, and eNB2 retains the notified information; when a user moves to a cell 5, the information collected by the eNB5 is that the service cell history information of the UE is the cell 1-cell 5, at the moment, the eNB5 carries out statistics and analysis on the behavior information of the user, and if the user frequently appears the service cell of the cell 1-cell 5, the user is obtained to pass through the cell 1-cell 5 in sequence frequently; the eNB5 notifies the statistics result to the eNB1 or to all of the cells 1 to 4; further, in the case of the high-speed rail scenario, when the user moves from cell 1 to cell 3, the network side determines that the user is riding the high-speed rail, and the eNB3 notifies the eNB4 and eNB5 of the result of the determination.

Preferably, according to the cell management scenario analysis: when the UE passes through the cell 1-cell 5, the main cell is unchanged all the time, and the cell 0 and the cell 1-cell 5 are used as auxiliary cells of the UE, wherein the cell configuration information is configuration information related to the auxiliary cells in the RRC connection reconfiguration message; the broadcast message of eNB0 contains the cell configuration information of cells 1-5; when the UE enters the eNB0 to which the cell 0 belongs, the UE reads the cell configuration information of the cell 1-cell 5 from the broadcast of the eNB 0; thereafter, with the movement of the UE, when the UE triggers measurement reporting, the UE automatically starts to use the cell configuration of cell 2, and for the configuration of cell 1 that was used before, the UE selects to delete or continue to reserve; when the eNB receives a measurement report message reported by the UE, the network side starts to use the cell 0+cell 2 to carry out data transmission on the UE.

Preferably, according to the cell handover scenario analysis: the UE generates cell switching when passing through the cell 1-the cell 5; the cell configuration information is configuration information related to cell switching in the RRC connection reconfiguration message; eNB0 has grouped the UE with the same route in advance, when a certain number of the UE with the same grouping enters the eNB0 to which the cell 0 belongs, the eNB0 sends the cell configuration information of the cells 1-5 to the UE in a multicast mode, and the UE feeds back ACK/NACK information to the eNB; after that, along with the movement of the UE, when the UE triggers measurement report, the UE performs a random access process on the cell 2, accesses the cell 2, and selects to delete or keep reserving the configuration of the cell 1 used before; when the eNB1 receives a measurement report message reported by the UE, the eNB1 transmits the UE context information to the eNB2 and simultaneously transmits the data which is completed for transmission to the eNB2, and after the UE accesses the eNB2, the eNB2 starts to transmit the data to the UE.

Preferably, in the second step, the user data collected, counted and analyzed is input into the training set for classification, and the training set χ is assumed _training ＝{(x ₁ ,y ₁ ),···,(x _L ,y _L ) O in } contains L marked data items, wherein the first component x of the i-th element _i ＝(x _il ,···,x _ip ) For the P-dimensional array, representing the P-dimensional attribute of the ith training sample, the second component y _i E y represents element x _i Y= |y| is the number of label categories in the classification problem, a binary classification problem when y=2, and Y>2 hoursClassification problems for multiple classes; generalization capability detection of classifier constructed from training set is to use test set χ containing U unlabeled data items _test ＝{x _L +1,···,x _L +U }.

Preferably, training in the third step is performed by training a marked data set χ _trainning The data in (a) is obtained into a classifier; in the network-based model, the classifier is represented by a network containing man-conveying data and associated labels; we refer to the output network obtained in the training phase as the training network;

training phase at this stage, training data uses network formation techniques

Conversion to get a network->

Training network->

The method comprises V= |gamma|nodes and E= |epsilon|edges; each node represents training sample χ _trainning Wherein v=l;

training networks are built using e-radius and k-man nearest neighbor algorithms, i.e., both algorithms applied alone may produce densely connected networks or split nodes into unconnected components.

Preferably, the construction of the training network combines the epsilon-radius and k-person neighbor algorithms; training set node x _i The adjacent city of (2) is:

wherein y is _i Representing training samples x _i E-radius (x) _i ,y _i ) Return set { x _i ,j∈γ:d(x _i ,y _i )＜＝∈∧y _i ＝y _i The k-nearest neighbor algorithm returns in principle to x _i Having the same labelIs a set of k nearest nodes; the calculation process of the K-neighbor algorithm return set is as follows: the algorithm firstly regards nodes in the training set as and data item x _i Is ordered according to the degree of similarity of the sequences, the ordered sequence y (x _i )＝{x _i ⁽¹⁾ ,···,x _i ^(k-1) ,x _i ^k ,x _i ^(k+1) ,···,x _i ^(Y(xi)-1) (wherein Y (x) _i ) Is with x _i The number of data items having the same tag; in this sequence, x _i ⁽¹⁾ Representation and x _i The most similar data items are presented in the form of,

representation and x _i Data items with the largest differences; the algorithm tries to try x _i Data items most similar to k thereof, i.e. { x _i ⁽¹⁾ ,···,x _i ^{(k -1)} ,x _i ^(k) Connected together; if the data item x is obtained in this step _i If there is more than one graphic component with the same class label, then the algorithm will delete the least similar data item from the k most similar data items, namely delete data item x _i ^(k) Then try to get the next similar data item x _i ^(k+1) And x _i Are connected; this process recursively proceeds until x is found _i The algorithm prevents in this way the presence of multiple network components with the same type of tag, in connection with other data items.

Preferably, the e-radius technique is used for data item dense areas (|e-radius (x) _i ,y _i )|>k) And the k-nearest neighbor algorithm is used for calculating the sparse region of the data item; using this strategy, each type of label that results will be represented by a separate component.

The beneficial effects of the invention are as follows:

the method has the advantages that the data accessed by the user are collected, counted and analyzed through the mobile network, the data are input into the advanced classification model for automatic classification, and data training is carried out, so that the purpose that the multi-source data are subjected to data fitting and demonstration of a track tod site evaluation system is achieved, when the user needs to go out, the destination of the user needing to go out can be obtained through the conventional data analysis, the conventional traffic analysis research method is changed by combining the multi-source big data with the advanced classification model method, a set of multi-dimensional, multi-level and all-dimensional evaluation system is established, and the adaptability of the track tod of the comprehensive hub area can be developed from the aspects of land, traffic, economy, population relations and the like.

Drawings

FIG. 1 is a schematic diagram of an evaluation method for building a comprehensive hub plate area system model based on the tod concept;

FIG. 2 is a schematic diagram of a training network formed in a training stage of an evaluation method for building a comprehensive hub-area system model based on the tod concept;

FIG. 3 is a schematic diagram of reasoning formed in the training stage of the evaluation method for building the comprehensive hub-area system model based on the tod idea;

fig. 4 is a schematic diagram of cell configuration under a traffic network of an evaluation method established by a comprehensive hub-plate system model based on the tod concept;

fig. 5 is a schematic diagram of cell pre-configuration under a traffic network of an evaluation method established by a comprehensive hub-plate system model based on the tod concept;

fig. 6 is a schematic diagram of cell pre-configuration consistent with user behavior under a traffic network based on an evaluation method established by a tod concept comprehensive hub-plate system model.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments.

1-6, an evaluation method based on tod concept for building a comprehensive hub plate area system model; step one, collecting, counting and analyzing access data of a user by utilizing a network;

the method for collecting in the first step is that users access data through a mobile network, wherein the data has a fixed travel route and a fixed travel time, the users get on the bus from a fixed station, take vehicles with fixed high-speed rails and subway routes, and get off the bus at the fixed station;

as shown in fig. 4, the user is exemplified by a cell 1-cell 5 used around the existing traffic route; frequent signaling interaction is needed between the eNB and the UE in the UE moving process, so that the problems of large signaling overhead, long transmission interruption time or low throughput period, excessive PDCCH scheduling times, PDCCH resource consumption and high UE power consumption are caused; the method comprises the steps of collecting, counting and analyzing user behaviors through a mobile network, and pre-configuring a cell to be used by UE;

the method for collecting, counting and analyzing the user behaviors by the network is divided into a centralized type and a distributed type, and an independent user information collecting module, a counting module and an analyzing module are added on the centralized type by utilizing a network side; the module is connected with a plurality of eNBs, each eNB transmits the collected user information to the module, and the module is responsible for comprehensively collecting the user information and carrying out statistics and analysis; the module obtains that the user regularly passes through the cells 1-5 in sequence through statistics and analysis; the statistical mode comprises the steps that a module counts the using times of the cell; the module notifies the relevant eNB of the statistics result and transmits the statistics result to the eNB1 and the eNB5; the module makes a pre-judgment on the behavior of the user, and when the user takes a high-speed rail vehicle, the running route of the user is fixed, so that the network makes a pre-judgment on a cell which the user taking the high-speed rail is about to arrive; the basis of the prejudgment comprises a cell in which the user is positioned at the moment or a series of service cells which pass before;

the distributed mode is to interact the switching history information of the UE and the history information of the cell configuration among the eNBs; the eNB records a cell used by a user, and when the user changes the cell, the source eNB transmits switching history information of the UE or history information of the cell use to the target eNB; when the UE changes from cell 1 to cell 2, eNB1 will notify eNB2 of the history information that cell 1 was serving as the serving cell, and eNB2 retains the notified information; when a user moves to a cell 5, the information collected by the eNB5 is that the service cell history information of the UE is the cell 1-cell 5, at the moment, the eNB5 carries out statistics and analysis on the behavior information of the user, and if the user frequently appears the service cell of the cell 1-cell 5, the user is obtained to pass through the cell 1-cell 5 in sequence frequently; the eNB5 notifies the statistics result to the eNB1 or to all of the cells 1 to 4; further, for the high-speed rail scene, when a user runs from the cell 1 to the cell 3, the network side is utilized to judge that the user is riding the high-speed rail, and the eNB3 informs the eNB4 and the eNB5 of the pre-judging result;

as illustrated in fig. 5, cells may be preconfigured by the above analysis of user behavior. The eNB performs cell pre-configuration on the UE through RRC dedicated signaling. After the eNB1 obtains the statistics that a certain UE regularly passes through the cells 1-5, and when the cell 1 serves the UE again, the eNB1 to which the cell 1 belongs will send all the cell configurations of the cells 1-5 to the UE, and the UE retains the cell configurations after receiving the cell configurations.

According to the cell management scenario analysis: when the UE passes through the cell 1-cell 5, the main cell is unchanged all the time, and the cell 0 and the cell 1-cell 5 are used as auxiliary cells of the UE, wherein the cell configuration information is configuration information related to the auxiliary cells in the RRC connection reconfiguration message; the broadcast message of eNB0 contains the cell configuration information of cells 1-5; when the UE enters the eNB0 to which the cell 0 belongs, the UE reads the cell configuration information of the cell 1-cell 5 from the broadcast of the eNB 0; thereafter, with the movement of the UE, when the UE triggers measurement reporting, the UE automatically starts to use the cell configuration of cell 2, and for the configuration of cell 1 that was used before, the UE selects to delete or continue to reserve; when the eNB receives a measurement report message reported by the UE, the network side starts to use a cell 0+a cell 2 to carry out data transmission on the UE;

according to the cell handover scene analysis: the UE generates cell switching when passing through the cell 1-the cell 5; the cell configuration information is configuration information related to cell switching in the RRC connection reconfiguration message; eNB0 has grouped the UE with the same route in advance, when a certain number of the UE with the same grouping enters the eNB0 to which the cell 0 belongs, the eNB0 sends the cell configuration information of the cells 1-5 to the UE in a multicast mode, and the UE feeds back ACK/NACK information to the eNB; after that, along with the movement of the UE, when the UE triggers measurement report, the UE performs a random access process on the cell 2, accesses the cell 2, and selects to delete or keep reserving the configuration of the cell 1 used before; when eNB1 receives a measurement report message reported by UE, eNB1 transmits the UE context information to eNB2 and simultaneously transmits data which is completed for transmission to eNB2, and when the UE is accessed to eNB2, eNB2 starts to transmit data to the UE;

as shown in fig. 6, the eNB may perform cell pre-configuration on the UE by broadcasting or multicasting. When a large number of users exist and act consistently, a large number of users get on the same station and get off the same station, and the network side can perform cell pre-configuration on the corresponding UE in a broadcast or multicast mode.

Inputting the collected, counted and analyzed user data into a high-level classification model for classification;

step two, user data obtained through collection, statistics and analysis are input into a training set for classification, and the training set χ is assumed _training ＝{(x ₁ ,y ₁ ),···,(x _L ,y _L ) O in } contains L marked data items, wherein the first component x of the i-th element _i ＝(x _il ,···,x _ip ) For the P-dimensional array, representing the P-dimensional attribute of the ith training sample, the second component y _i E y represents element x _i Y= |y| is the number of label categories in the classification problem, a binary classification problem when y=2, and Y>2, classifying the problems in multiple classes; generalization capability detection of classifier constructed from training set is to use test set χ containing U unlabeled data items _test ＝{x _L +1,···,x _L +U };

training the statistical data; training through training marker dataset χ _trainning The data in (a) is obtained into a classifier; in the network-based model, the classifier is represented by a network containing man-conveying data and associated labels; we refer to the output network obtained in the training phase as the training network;

training phase at this stage, training data uses network formation techniques

Conversion to get a network->

Training network->

The method comprises V= |gamma|nodes and E= |epsilon|edges; each node represents training sample χ _trainning Wherein v=l;

training a network is constructed by using an epsilon-radius and k-person neighbor algorithm, namely, the two algorithms are singly applied to possibly generate a densely connected network or divide nodes into unconnected components;

the construction of the training network combines an epsilon-radius and a k-person neighbor algorithm; training set node x _i The adjacent city of (2) is:

wherein y is _i Representing training samples x _i E-radius (x) _i ,y _i ) Return set { x _i ,j∈γ:d(x _i ,y _i )＜＝∈∧y _i ＝y _i The k-nearest neighbor algorithm returns in principle to x _i K nearest node sets with the same label; the calculation process of the K-neighbor algorithm return set is as follows: the algorithm firstly regards nodes in the training set as and data item x _i Ranking the similarity of the sequences to obtain a ranking sequence

Wherein Y (x) _i ) Is with x _i The number of data items having the same tag; in this sequence, x _i ⁽¹⁾ Representation and x _i Most similar data item, +.>

Representation and x _i Data items with the largest differences; the algorithm tries to try x _i Data items most similar to k thereof, i.e. { x _i ⁽¹⁾ ,···,x _i ^(k-1) ,x _i ^(k) Connected together; if the data item x is obtained in this step _i If there is more than one graphic component with the same class label, then the algorithm will delete the least similar data item from the k most similar data items, namely delete data item x _i ^(k) Then try to get the next similar data item x _i ^(k+1) And x _i Are connected; this process recursively proceeds until x is found _i The algorithm prevents in this way the presence of multiple network components with the same type of tag, in connection with other data items;

the e-radius technique is used for data item dense areas (|e-radius (x) _i ,y _i )|>k) And the k-nearest neighbor algorithm is used for calculating the sparse region of the data item; using this strategy, each type of tag that results will be represented by a separate component;

as shown in FIG. 1, a scatter plot determines which nodes are adjacent to a centrally located dark node. Let k=2, e be the radius shown in the figure. Since the graph contains three nodes within an epsilon-radius, 3>k, the area within the circle in the graph is considered a dense area within which the epsilon-radius algorithm will be chosen to be employed. Thus, the centrally located dark node will connect with the other three dark nodes within the circle.

As shown in fig. 2-3, a multi-class classification problem diagram with y=3 is illustrated as the configuration state of the network at the end of the training phase. Wherein each class is a representative component, in the figure we denote these components by circles, respectively

And->

And step four, storing the training data and waiting for calling.

The method has the advantages that the data accessed by the user are collected, counted and analyzed through the mobile network, the data are input into the advanced classification model for automatic classification, and data training is carried out, so that the purpose that the multi-source data are subjected to data fitting and demonstration of a track tod site evaluation system is achieved, when the user needs to go out, the destination of the user needing to go out can be obtained through the conventional data analysis, the conventional traffic analysis research method is changed by combining the multi-source big data with the advanced classification model method, a set of multi-dimensional, multi-level and all-dimensional evaluation system is established, and the adaptability of the track tod of the comprehensive hub area can be developed from the aspects of land, traffic, economy, population relations and the like.

Working principle: the data information is accessed through a usual travel mobile network of a user, collected, counted and analyzed through the evaluation method, automatically classified, trained and stored in the advanced classification model, and when the user goes out again, the track tod site is evaluated by utilizing the multi-source data through the evaluation method, and the travel direction position of the user is analyzed and prejudged, so that the traditional traffic analysis method is changed.

The foregoing is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art, who is within the scope of the present invention, shall cover the same or different embodiments according to the technical solution and the inventive concept of the present invention.

Claims (10)

1. An evaluation method for building a comprehensive hub area system model based on a tod concept is characterized by comprising the following steps of: step one, collecting, counting and analyzing access data of a user by utilizing a network;

inputting the collected, counted and analyzed user data into a high-level classification model for classification;

training the statistical data;

and step four, storing the training data and waiting for calling.

2. The method for evaluating the system model establishment of the comprehensive junction piece region based on the tod idea according to claim 1, wherein the method comprises the following steps: the method for collecting in the first step is to access data through users existing in a mobile network, wherein the data has a fixed travel route and fixed travel time, get on the bus from a fixed station, take vehicles with fixed high-speed rails and subway routes, and get off the bus at the fixed station;

setting a cell used by a user around an existing traffic route as a cell 1-a cell 5; frequent signaling interaction is required between the eNB and the UE in the UE moving process; thus, the cell to be used by the UE is preconfigured through the collection, statistics and analysis of the user behaviors by the mobile network.

3. The method for evaluating the system model establishment of the comprehensive junction piece region based on the tod idea according to claim 2, wherein the method comprises the following steps: the method for collecting, counting and analyzing the user behaviors by the network is divided into a centralized type and a distributed type, and an independent user information collecting module, a counting module and an analyzing module are added on a centralized type utilization network side; the module is connected with a plurality of eNBs, each eNB transmits the collected user information to the module, and the module is responsible for comprehensively collecting the user information and carrying out statistics and analysis; the module obtains that the user regularly passes through the cells 1-5 in sequence through statistics and analysis; the statistical mode comprises the steps that a module counts the using times of the cell; the module notifies the relevant eNB of the statistics result and transmits the statistics result to the eNB1 and the eNB5; the module makes a pre-judgment on the behavior of the user, and when the user takes a high-speed rail vehicle, the running route of the user is fixed, so that the network makes a pre-judgment on a cell which the user taking the high-speed rail is about to arrive; the basis for the pre-determination includes the cell in which the user is located at this time, or a series of serving cells that have been previously passed.

4. The method for evaluating the system model establishment of the comprehensive junction piece region based on the tod idea according to claim 3, wherein the method comprises the following steps: the distributed mode is that the switching history information of the UE and the history information of the cell configuration are interacted between the eNBs; the eNB records a cell used by a user, and when the user changes the cell, the source eNB transmits switching history information of the UE or history information of the cell use to the target eNB; when the UE changes from cell 1 to cell 2, eNB1 will notify eNB2 of the history information that cell 1 was serving as the serving cell, and eNB2 retains the notified information; when a user moves to a cell 5, the information collected by the eNB5 is that the service cell history information of the UE is the cell 1-cell 5, at the moment, the eNB5 carries out statistics and analysis on the behavior information of the user, and if the user frequently appears the service cell of the cell 1-cell 5, the user is obtained to pass through the cell 1-cell 5 in sequence frequently; the eNB5 notifies the statistics result to the eNB1 or to all of the cells 1 to 4; further, in the case of the high-speed rail scenario, when the user moves from cell 1 to cell 3, the network side determines that the user is riding the high-speed rail, and the eNB3 notifies the eNB4 and eNB5 of the result of the determination.

5. The method for evaluating the system model establishment of the comprehensive junction piece region based on the tod idea, which is characterized in that: according to the cell management scenario analysis: when the UE passes through the cell 1-cell 5, the main cell is unchanged all the time, and the cell 0 and the cell 1-cell 5 are used as auxiliary cells of the UE, wherein the cell configuration information is configuration information related to the auxiliary cells in the RRC connection reconfiguration message; the broadcast message of eNB0 contains the cell configuration information of cells 1-5; when the UE enters the eNB0 to which the cell 0 belongs, the UE reads the cell configuration information of the cell 1-cell 5 from the broadcast of the eNB 0; thereafter, with the movement of the UE, when the UE triggers measurement reporting, the UE automatically starts to use the cell configuration of cell 2, and for the configuration of cell 1 that was used before, the UE selects to delete or continue to reserve; when the eNB receives a measurement report message reported by the UE, the network side starts to use the cell 0+cell 2 to carry out data transmission on the UE.

6. The method for evaluating the system model establishment of the comprehensive junction piece region based on the tod idea, which is characterized in that: according to the cell handover scene analysis: the UE generates cell switching when passing through the cell 1-the cell 5; the cell configuration information is configuration information related to cell switching in the RRC connection reconfiguration message; eNB0 has grouped the UE with the same route in advance, when a certain number of the UE with the same grouping enters the eNB0 to which the cell 0 belongs, the eNB0 sends the cell configuration information of the cells 1-5 to the UE in a multicast mode, and the UE feeds back ACK/NACK information to the eNB; after that, along with the movement of the UE, when the UE triggers measurement report, the UE performs a random access process on the cell 2, accesses the cell 2, and selects to delete or keep reserving the configuration of the cell 1 used before; when the eNB1 receives a measurement report message reported by the UE, the eNB1 transmits the UE context information to the eNB2 and simultaneously transmits the data which is completed for transmission to the eNB2, and after the UE accesses the eNB2, the eNB2 starts to transmit the data to the UE.

7. The method for evaluating the system model establishment of the comprehensive junction piece region based on the tod idea according to claim 1, wherein the method comprises the following steps: in the second step, the collected, counted and analyzed user data are input into a training set for classification, and the training set χ is assumed _training ＝{(x ₁ ,y ₁ ),···,(x _L ,y _L ) O in } contains L marked data items, wherein the first component x of the i-th element _i ＝(x _il ,···,x _ip ) For the P-dimensional array, representing the P-dimensional attribute of the ith training sample, the second component y _i E y represents element x _i Y= |y| is the number of label categories in the classification problem, a binary classification problem when y=2, and Y>2, classifying the problems in multiple classes; generalization capability detection of classifier constructed from training set is to use test set χ containing U unlabeled data items _test ＝{x _L +1,···,x _L +U }.

8. The method for evaluating the system model establishment of the comprehensive junction piece region based on the tod idea, which is characterized in that: training through training mark data set χ in the step three _trainning Data in (a) is obtainedA classifier; in the network-based model, the classifier is represented by a network containing man-conveying data and associated labels; we refer to the output network obtained in the training phase as the training network;

training phase at this stage, the training data uses network formation technique g:

converting to obtain network

Training network->

The method comprises V= |gamma|nodes and E= |epsilon|edges; each node represents training sample χ _trainning Wherein v=l;

training networks are built using e-radius and k-man nearest neighbor algorithms, i.e., both algorithms applied alone may produce densely connected networks or split nodes into unconnected components.

9. The method for evaluating the system model establishment of the comprehensive junction piece region based on the tod idea, which is characterized in that: the construction of the training network combines an epsilon-radius and a k-person neighbor algorithm; training set node x _i The adjacent city of (2) is:

wherein y is _i Representing training samples x _i E-radius (x) _i ,y _i ) Return set { x _i ,j∈γ:d(x _i ,y _i )＜＝∈∧y _i ＝y _i The k-nearest neighbor algorithm returns in principle to x _i K nearest node sets with the same label; the calculation process of the K-neighbor algorithm return set is as follows: the algorithm firstly regards nodes in the training set as and data item x _i Ranking the similarity of the sequences to obtain a ranking sequence

Wherein Y (x) _i ) Is with x _i The number of data items having the same tag; in this sequence, x _i ⁽¹⁾ Representation and x _i Most similar data item, +.>

Representation and x _i Data items with the largest differences; the algorithm tries to try x _i Data items most similar to k thereof, i.e. { x _i ⁽¹⁾ ,···,x _i ^(k-1) ,x _i ^(k) Connected together; if the data item x is obtained in this step _i If there is more than one graphic component with the same class label, then the algorithm will delete the least similar data item from the k most similar data items, namely delete data item x _i ^(k) Then try to get the next similar data item x _i ^(k+1) And x _i Are connected; this process recursively proceeds until x is found _i The algorithm prevents in this way the presence of multiple network components with the same type of tag, in connection with other data items.

10. The method for evaluating the system model establishment of the comprehensive junction piece region based on the tod idea according to claim 9, wherein the method comprises the following steps: the e-radius technique is used for data item dense areas (|e-radius (x) _i ,y _i )|>k) And the k-nearest neighbor algorithm is used for calculating the sparse region of the data item; using this strategy, each type of label that results will be represented by a separate component.

Images