CN111629319B - Position prediction method and device - Google Patents
Position prediction method and device Download PDFInfo
- Publication number
- CN111629319B CN111629319B CN201910151995.4A CN201910151995A CN111629319B CN 111629319 B CN111629319 B CN 111629319B CN 201910151995 A CN201910151995 A CN 201910151995A CN 111629319 B CN111629319 B CN 111629319B
- Authority
- CN
- China
- Prior art keywords
- user
- intelligent entity
- target
- database
- prediction model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/025—Services making use of location information using location based information parameters
- H04W4/027—Services making use of location information using location based information parameters using movement velocity, acceleration information
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/029—Location-based management or tracking services
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Abstract
The invention provides a position prediction method and device. In the invention, the mobility prediction model of each user classification is pre-trained in a mode of classifying all user historical tracks and movement characteristics, compared with a mode of predicting a single user by utilizing the track of the single user, the same class of users only need to maintain one mobility prediction model, and the calculation burden of model training can be greatly reduced; the dynamic updating process of the model can also solve the problem of inaccurate prediction caused by the change of the movement rule of the user. In addition, the invention can achieve better training effect under the condition of insufficient initial data volume, relieves the pressure of data collection in the early stage, and does not need to wait for the collection of historical data of a certain user, thereby realizing rapid position prediction. In addition, for a new user without a historical track, the embodiment of the invention can judge the motion mode of the user through the real-time track of the user and select a proper model for prediction.
Description
Technical Field
The embodiment of the invention relates to the technical field of communication, in particular to a position prediction method and device.
Background
In mobile internet services, location-based services are rapidly developed, bringing great convenience to the lives of people. For example, the method provides traffic road condition information in front of the path traveled by the user, helps the user to make travel route planning in advance, and avoids traffic jam; providing tourist route recommendations for the guest, etc. The application scenarios have a common problem to be solved, namely, the future movement track of the user needs to be accurately predicted.
With the rapid development of geographic information systems, mobile positioning technologies, wireless communication networks and intelligent terminal technologies, convenience is provided for collecting position data of users. Therefore, how to use the above-mentioned technology to quickly realize high-precision position prediction becomes a problem to be solved urgently.
Disclosure of Invention
An object of the embodiments of the present invention is to provide a method and an apparatus for position prediction, which can quickly implement high-precision position prediction.
The embodiment of the invention provides a position prediction method, which is applied to an edge intelligent entity and comprises the following steps:
the edge intelligent entity determines the user classification of the target user according to the current position range and/or the movement speed interval of the target user of the position to be predicted;
the edge intelligent entity selects a target mobility prediction model corresponding to the user classification to which the target user belongs according to the user classification to which the target user belongs, wherein the plurality of mobility prediction models respectively correspond to different user classifications, and the mobility prediction model corresponding to each user classification is obtained by the cloud intelligent entity according to position data of the user under the user classification in a database;
and the edge intelligent entity obtains the predicted position of the target user at the predicted time by using the target mobility prediction model according to the historical position and the current position of the target user.
Preferably, the method further comprises:
and the edge intelligent entity collects the user data in the region and sends the user data to the database for storage.
Preferably, the method further comprises:
the edge intelligent entity receives a first notification message sent by the cloud intelligent entity, the first notification message is sent after the cloud intelligent entity retrains according to user data stored in a database to obtain a mobility prediction model corresponding to each user classification, the retraining is executed periodically or is executed after the receiving frequency of update requests received by the cloud intelligent entity exceeds a preset threshold, and the update requests are requests sent by each edge intelligent entity for updating the mobility prediction model;
and the edge intelligent entity acquires a mobility prediction model corresponding to each user classification obtained by retraining the cloud intelligent entity stored in a database according to the first notification message.
Preferably, the method further comprises:
the edge intelligent entity obtains the actual position of the target user when the predicted time arrives;
judging whether the prediction accuracy of the target mobility prediction model meets a preset requirement or not according to the predicted positions and the actual positions of the plurality of predicted moments;
and when the prediction accuracy of the target mobility prediction model does not meet the preset requirement, sending an updating request for updating the target mobility prediction model to the cloud intelligent entity.
Preferably, in the above method, the step of obtaining the actual location of the target user includes:
when the target user is in a Radio Resource Control (RRC) activated state, acquiring an actual cell where the target user is located, and determining the actual position of the target user according to the received signal strength of the target user recorded by the actual cell, or determining the actual position of the target user according to the received signal strength of the target user, satellite positioning system information and other auxiliary information recorded by the actual cell;
and when the target user is in an RRC idle state, determining the actual position of the target user according to the satellite positioning system information of the target user, or taking the predicted position of the target user as the actual position of the target user.
Preferably, in the above method, after determining the actual location of the target user, the method further includes:
and updating the user data of the target user in the database according to the actual position of the target user.
Preferably, the method further comprises:
the edge intelligent entity receives a second notification message sent by the cloud intelligent entity after the target mobility prediction model is updated according to the updating request;
and the edge intelligent entity acquires the updated target mobility prediction model stored in a database by the cloud intelligent entity according to the second notification message.
The embodiment of the invention provides another position prediction method, which is applied to a cloud intelligent entity,
according to a preset period, user data of each edge intelligent entity in an area to which the edge intelligent entity belongs, wherein the edge intelligent entity is stored in a database, and the user data comprises position data and/or speed data of a user;
dividing the user data into a plurality of user classifications according to the position range and the speed interval to which the user data belongs;
according to the user data under each user classification, a mobility prediction model corresponding to the user classification is obtained through retraining respectively, the mobility prediction model obtained through retraining is stored in a database, and a first notification message used for indicating that the model is updated is sent to the edge intelligent entity.
Preferably, the method further comprises:
receiving an update request sent by a first edge intelligent entity for updating a first mobility prediction model of a first user class, wherein the update request is sent by the first edge intelligent entity when the prediction accuracy of the first mobility prediction model does not meet the preset requirement;
according to the updating request, user data stored in a database under the first user classification is obtained;
according to the user data under the first user classification, the first mobility prediction model is retrained or updated, the updated first mobility prediction model is stored in a database, and a second notification message used for indicating that the model is updated is sent to the first edge intelligent entity.
Preferably, the method further comprises:
judging whether the receiving frequency of the updating requests sent by each edge intelligent entity exceeds a preset threshold or not;
and when the receiving frequency exceeds the preset threshold, returning to the step of acquiring the user data in the region of each edge intelligent entity stored in the database.
Preferably, in the above method, after dividing the user data into a plurality of user classifications, the method further includes: and storing the user data obtained by dividing under each user classification into a database.
The embodiment of the present invention further provides an edge intelligent entity, including: a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the steps of the position prediction method as described above.
The embodiment of the present invention further provides a cloud intelligent entity, including: a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the steps of the position prediction method as described above.
An embodiment of the present invention further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the method are implemented as described above.
According to the position prediction method and the position prediction equipment provided by the embodiment of the invention, the mobility prediction model of each user classification is pre-trained in a mode of classifying all user historical tracks and movement characteristics, and compared with a mode of predicting a single user by utilizing the track of the single user, the same type of user only needs to maintain one mobility prediction model, so that the calculation burden of model training can be greatly reduced; the dynamic updating process of the model can also solve the problem of inaccurate prediction caused by the change of the user movement rule. In addition, the embodiment of the invention can achieve better training effect under the condition that the initial data volume is not enough, relieves the pressure of early-stage data collection, and does not need to wait for the collection of historical data of a certain user, thereby realizing rapid position prediction. In addition, for a new user without a historical track, the embodiment of the invention can also judge the user motion mode through the real-time track of the user and select a proper model for prediction.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
FIG. 1 is a schematic diagram illustrating a user location prediction method in the prior art;
FIG. 2 is a schematic diagram of a location prediction system according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating a location prediction method according to an embodiment of the present invention;
FIG. 4 is a schematic flow chart illustrating a location prediction method according to an embodiment of the present invention;
FIG. 5 is a schematic flow chart of a position prediction method according to an embodiment of the present invention
FIG. 6 is a diagram illustrating short-term model updates in a location prediction method according to an embodiment of the present invention;
FIG. 7 is a diagram illustrating long-term model update in a location prediction method according to an embodiment of the present invention;
FIG. 8 is a diagram illustrating an exemplary location prediction method according to an embodiment of the present invention;
FIG. 9 is a schematic structural diagram of an edge intelligent entity according to an embodiment of the present invention;
fig. 10 is a schematic structural diagram of a cloud-end smart entity according to an embodiment of the present invention;
fig. 11 is a schematic structural diagram of a communication device according to an embodiment of the present invention;
fig. 12 to 13 are simulation effect diagrams of the position prediction method according to the embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
The terms first, second and the like in the description and in the claims of the present application are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are, for example, capable of operation in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus. "and/or" in the specification and claims means at least one of the connected objects
The techniques described herein are not limited to Long Time Evolution (LTE)/LTE Evolution (LTE-Advanced) systems, and may also be used for various wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency-Division Multiple Access (SC-FDMA), and other systems. The terms "system" and "network" are often used interchangeably. CDMA systems may implement radio technologies such as CDMA2000, Universal Terrestrial Radio Access (UTRA), and so on. UTRA includes Wideband CDMA (Wideband code division Multiple Access, WCDMA) and other CDMA variants. TDMA systems may implement radio technologies such as Global System for Mobile communications (GSM). The OFDMA system may implement radio technologies such as Ultra Mobile Broadband (UMB), evolved-UTRA (E-UTRA), IEEE 802.11(Wi-Fi), IEEE 802.16(WiMAX), IEEE 802.20, Flash-OFDM, etc. UTRA and E-UTRA are parts of the Universal Mobile Telecommunications System (UMTS). LTE and higher LTE (e.g., LTE-A) are new UMTS releases that use E-UTRA. UTRA, E-UTRA, UMTS, LTE-A, and GSM are described in documents from an organization named "third Generation Partnership Project" (3 GPP). CDMA2000 and UMB are described in documents from an organization named "third generation partnership project 2" (3GPP 2). The techniques described herein may be used for both the above-mentioned systems and radio technologies, as well as for other systems and radio technologies. However, the following description describes the NR system for purposes of example, and NR terminology is used in much of the description below, although the techniques may also be applied to applications other than NR system applications.
The following description provides examples and does not limit the scope, applicability, or configuration set forth in the claims. Changes may be made in the function and arrangement of elements discussed without departing from the spirit and scope of the disclosure. Various examples may omit, substitute, or add various procedures or components as appropriate. For example, the described methods may be performed in an order different than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.
In this embodiment of the present invention, the first communication device or the second communication device may be an LTE Base Station 5G or a later-version Base Station (e.g., a gNB, a 5G NR NB, etc.), or a Base Station in other communication systems (e.g., an eNB, a WLAN access point, or other access points, etc.), where the Base Station may also be referred to as a node B, an evolved node B, an access point, a Base Transceiver Station (BTS), a radio Base Station, a radio Transceiver, a Basic Service Set (BSS), an Extended Service Set (ESS), a node B, an evolved node B (eNB), a home node B, a home evolved node B, a WLAN access point, a WiFi node, or some other suitable term in the field, as long as the same technical effect is achieved, and the Base Station is not limited to a specific technical vocabulary.
The existing user location prediction algorithm is mainly based on the flow shown in fig. 1:
(1) and training a mobility prediction model corresponding to a certain user by using historical position data of the user. The specific training method of the mobility prediction model can be realized by various methods. Sequence prediction is performed on which cells the user is likely to move in, for example, using a markov chain method; or a frequent item set mining technology is adopted to establish association rules; or comparing the historical movement track data of the user, and predicting the next cell moved by the user. With the rapid development of artificial intelligence technology, deep-loop neural network algorithms such as Long Short-Term Memory network (LSTM) and the like are widely applied to user mobility prediction.
(2) The latest position data of the user is obtained as input for predicting the position of the user at an unknown time. Specifically, assuming that the predicted time is D days, the input position may be a user position at a time t, which is a continuous time t-N · Δ t, … t-2 · Δ t, t- Δ t before the time t, or a user position at the same time t on D-N, …, D-1 days. Here, Δ t represents a preset data acquisition period.
(3) And outputting the predicted position at the moment t by using the latest position data of the user and combining the trained mobility prediction model.
The following problems exist in the prior art:
(1) in the early stage of prediction, due to the lack of sufficient historical geographic position information, the selection of the target geographic position of the mobile user is greatly limited, so that the accuracy of the target geographic position of the mobile user is low; under the condition that the historical track quantity of each user is not enough, the model of a single user cannot be well trained, so that the prediction accuracy is reduced, and the prediction cannot be performed on the user without the historical track. Therefore, it is not feasible to design a relatively independent user mobility prediction model for a single user, and a user mobility prediction method for a certain category should be designed relatively.
(2) On the other hand, the matching prediction method is high in calculation complexity through matching similarity with the historical track of the user, and due to the fact that the prediction granularity is limited by the accuracy of the historical track of the user, prediction cannot be performed on the ideal prediction granularity.
(3) Due to the influence of some external interference factors and the change of the user movement rule, the pre-trained model may not always maintain good prediction performance. Therefore, when there is an error in the prediction, a suitable mechanism is needed to improve the prediction model. Meanwhile, when the movement prediction is performed on one type of users, the movement data of different users should be supplemented with each other, so that the accuracy of the prediction model is improved.
Generally, a user mobility prediction model is a process which still needs to be updated and evolved continuously after training is completed, so that a user mobility rapid high-precision prediction method is required to be designed, interaction processes of network elements such as an edge (base station gNB, edge computing gateway and the like) intelligent entity, a cloud (such as a Data analysis unit (DAF) intelligent entity, a database (such as a Data storage unit (Data) replication (DR)) and the like are perfected, and the problem of mobile prediction model misalignment caused by user movement rule change and Data loss is solved.
Referring to fig. 2, a location prediction system provided in an embodiment of the present invention includes a cloud intelligent entity, an edge intelligent entity, and a database. Here, the cloud-side smart entity may be an entity such as a data analysis unit (DAF); the edge intelligent entity can be entities such as various base stations (such as gNB), edge computing gateways and the like; the database may be a Data storage unit (DR). It should be noted that, the system is divided into an edge smart entity and a cloud smart entity, mainly considering that the computing power of the edge smart entity is limited, and only local mobile data is visible. Therefore, the training of the mobility prediction model can be carried out by the cloud intelligent entity, and the edge intelligent entity carries out prediction by using the model trained by the cloud and feeds back the prediction result. The cloud intelligent entity and the database may belong to different network elements (for example, located in DAF and DR, respectively), or may belong to the same network element.
Fig. 2 also shows an example of the main functional modules of each network element in the location prediction system according to the embodiment of the present invention. The cloud intelligent entity finishes the training work of the mobility prediction model and is interactively matched with the database to update the mobility model in a long term and a short term; the edge intelligent entity carries out online prediction of a single user and tracks prediction accuracy rate information; the database stores the trained model and various user data. The edge intelligent entity, the cloud intelligent entity and the database realize the rapid high-precision prediction of the user mobility through the position prediction method provided by the embodiment of the invention.
The position prediction method according to the embodiment of the present invention will be described below from the edge smart entity and the cloud smart entity, respectively.
Referring to fig. 3, the position prediction method provided in the embodiment of the present invention, when applied to an edge intelligent entity, includes:
and step 31, the edge intelligent entity determines the user classification of the target user according to the current position range and/or the movement speed interval of the target user of the position to be predicted.
Here, the current position of the user may be divided into a plurality of non-overlapping position ranges in advance, and then the current position of the user is determined by a relevant position algorithm (e.g., according to the received signal strength, positioning data of a satellite positioning system, etc.), so as to determine a position range to which the current position of the user belongs. Similarly, the movement speed interval of the user may be divided into a plurality of non-overlapping movement speed intervals in advance, and then the movement speed interval to which the current speed of the user belongs may be determined.
And step 32, selecting, by the edge intelligent entity, a target mobility prediction model corresponding to the user classification to which the target user belongs according to the user classification to which the target user belongs, wherein the plurality of mobility prediction models respectively correspond to different user classifications, and the mobility prediction model corresponding to each user classification is obtained by the cloud intelligent entity according to user data under the user classification in a database, and the user data includes position data and/or speed data of the user.
Here, each piece of user data may correspond to a group of location sequences of the user, the group of location sequences is composed of a plurality of locations acquired successively, and the number of the plurality of locations may be preset. Thus, each user datum includes geographic location information and/or velocity information of the user at those locations.
In the embodiment of the present invention, a mobility prediction model is trained in advance for users under the same user classification, and in consideration of the operation performance, the cloud-side intelligent entity may perform the training, and store the trained model in the database, so that the edge intelligent entity may obtain the model from the database for the model selection in step 32.
And step 33, the edge intelligent entity obtains the predicted position of the target user at the prediction moment by using the target mobility prediction model according to the historical position and the current position of the target user.
Through the steps, the edge intelligent entity of the embodiment of the invention can predict and obtain the predicted position of the target user by utilizing the mobility prediction model under the same user classification obtained by pre-training, compared with a model for predicting a single user by utilizing the track of the single user, the same type of user of the embodiment of the invention only needs to maintain one mobility prediction model, and the calculation burden of model training can be greatly reduced. In addition, the embodiment of the invention can achieve better training effect under the condition of insufficient initial data volume, relieves the pressure of early-stage data collection, and does not need to wait for the collection of historical data of a certain user, thereby realizing rapid position prediction. In addition, for a new user without a historical track, the embodiment of the invention can also judge the user motion mode through the real-time track of the user and select a proper model for prediction.
In addition, in the method according to the embodiment of the present invention, each edge intelligent entity may collect user data in the area to which the edge intelligent entity belongs and send the user data to the database for storage, where the user data includes position data and/or speed data of the user. Therefore, the cloud intelligent entity can train or update the mobility prediction model according to the user data stored in the database.
After the predicted position of the target user is obtained through prediction, the edge intelligent entity can also obtain the actual position of the target user when the predicted time arrives, and then the actual position is matched with the predicted position to judge whether the position prediction is accurate. For example, a prediction may be considered accurate when the distance between the actual location and the predicted location is less than a predetermined distance, otherwise the prediction may be considered inaccurate. The predetermined distance may be set according to the cell size, the terminal speed, and other factors.
Specifically, there are various ways to obtain the actual location of the target user. For example, when the target user is in a radio resource control RRC activated state, an actual cell where the target user is located is obtained, and an actual location of the target user is determined according to the received signal strength of the target user recorded by the actual cell, or the actual location of the target user is determined according to the received signal strength of the target user, satellite positioning system information, and other auxiliary information recorded by the actual cell. And when the target user is in an RRC idle state, determining the actual position of the target user according to the satellite positioning system information of the target user, or taking the predicted position of the target user as the actual position of the target user.
In addition, after determining the actual position of the target user, the edge intelligent entity according to the embodiment of the present invention may further update the user data of the target user in the database according to the actual position of the target user.
Further, the edge intelligent entity may further determine whether the prediction accuracy of the target mobility prediction model meets a preset requirement according to the predicted positions and the actual positions of the plurality of predicted times. For example, the prediction accuracy of the model may be considered unsatisfactory when 6 prediction positions out of 10 predictions are inaccurate. Specifically, when the prediction accuracy of the target mobility prediction model does not meet the preset requirement, an update request for updating the target mobility prediction model may be sent to the cloud intelligent entity. In this way, the cloud intelligent entity can obtain the user data under the user classification to which the target user belongs from the database again according to the update request, retrain or update the model, obtain the updated target mobility prediction model of the user classification to which the target user belongs, and store the updated target mobility prediction model in the database. The cloud-side intelligent entity may further send a notification message to the edge intelligent entity, so that the edge intelligent entity may obtain the updated target mobility prediction model stored in the database by the cloud-side intelligent entity after receiving the notification message.
In the embodiment of the invention, the cloud intelligent entity can periodically retrain all models, and after the training is finished, the cloud intelligent entity can send another notification message to the edge intelligent entity, so that the edge intelligent entity can receive the another notification message sent by the cloud intelligent entity, and further obtain the mobility prediction models corresponding to the user classifications obtained by retraining, stored in the database by the cloud intelligent entity, according to the first notification message, so as to update the models.
Preferably, after the receiving frequency of the update request received by the cloud intelligent entity exceeds the predetermined threshold, the cloud intelligent entity may also perform retraining or updating of all models, and after obtaining a new model, store the new model at the database, and send another notification message to the edge intelligent entity, so that the edge intelligent entity may receive another notification message sent by the cloud intelligent entity, and according to the another notification message, obtain the new model at the database, thereby implementing updating of the model.
Through the steps, the embodiment of the invention can realize the dynamic update of the model and can solve the problem of prediction misalignment caused by the change of the movement rule of the user.
Referring to fig. 4, when the location prediction method provided in the embodiment of the present invention is applied to a cloud-end intelligent entity, the method includes:
And 42, dividing the user data into a plurality of user classifications according to the position range and the speed interval to which the user data belongs.
And 43, according to the user data under each user classification, retraining respectively to obtain a mobility prediction model corresponding to the user classification, storing the retrained mobility prediction model in a database, and sending a first notification message for indicating that the model is updated to the edge intelligent entity.
Through the following steps, the cloud intelligent entity can periodically update the mobility prediction model and inform the edge intelligent entity to acquire the updated model, so that the effectiveness of the model can be ensured, and the accuracy of position prediction can be improved.
The model update in steps 41 to 43 is based on a preset period, and the preset period can be set to be relatively long, so the embodiment of the present invention also refers to the update as a long-term update.
Preferably, in the method according to the embodiment of the present invention, the cloud-side intelligent entity may further receive an update request sent by the first edge intelligent entity for updating the first mobility prediction model of the first user category, where the update request is sent by the first edge intelligent entity when the prediction accuracy of the first mobility prediction model does not meet the preset requirement. And then, according to the updating request, obtaining the user data stored in the database under the first user classification. And then, according to the user data under the first user classification, the first mobility prediction model is retrained or updated, the updated first mobility prediction model is stored in a database, and a second notification message used for indicating that the model is updated is sent to the first edge intelligent entity. The model update is triggered based on one update request of the edge intelligent device, and the update is also called short-term update by the embodiment of the invention.
Preferably, in the method according to the embodiment of the present invention, the cloud intelligent entity may further determine whether a receiving frequency of the update request sent by each edge intelligent entity exceeds a predetermined threshold. And when the receiving frequency exceeds the preset threshold, returning to the step 41, executing the action of acquiring the user data of each edge intelligent entity stored in the area of the edge intelligent entity at the database, and executing the subsequent steps 42-43 to update the model. Since the model update herein is triggered based on multiple update requests by one or more edge smart devices, embodiments of the present invention also refer to the update as a long-term update.
Preferably, in the method according to the embodiment of the present invention, after the user data is divided into a plurality of user classifications, the cloud intelligent entity may further store the user data obtained by the division under each user classification into the database.
Through the steps, the embodiment of the invention can realize the long-term and short-term dynamic update of the model and can solve the problem of inaccurate prediction caused by the change of the movement rule of the user.
The position prediction method of the embodiment of the invention is introduced from the edge intelligent entity and the cloud intelligent entity respectively. To better help understanding the location prediction method of the embodiment of the present invention, the following will further describe the embodiment of the present invention through an interaction flow between various entities.
Referring to fig. 5, the position prediction method according to the embodiment of the present invention may include the following steps:
(1) the edge intelligent entity collects the position data of the users in the area and reports the position data to the database;
(2) the cloud intelligent entity extracts position data of different users from a database user data area, and analyzes, processes and classifies the positions; the classification criterion comprises a movement speed interval where the position sequence is located or a geographic position range where the position sequence is located.
(3) The cloud intelligent entity performs offline learning and long-term updating processes: pre-training a mobility prediction model of each category by using the position sequence of each category, and storing the trained mobility prediction model and the classified position sequence into a corresponding area of a database;
(4) and (3) the edge intelligent entity carries out online prediction: loading a corresponding mobility prediction model for a user needing prediction, performing real-time track prediction according to the current position and the historical position of the user, storing the current position into a database user data area, judging the performance of the model according to a certain judgment standard after multiple predictions, and informing a cloud intelligent entity to update the mobility model of the user if the performance of the model is poor;
(5) after receiving a model updating request of the edge intelligent entity, the cloud intelligent entity acquires the position data of the type of user to finely adjust the relevant model (namely, updating in a short term, see figure 4), and stores the updated model in a user model cache region of a database;
(6) the cloud intelligent entity periodically renews and offline learns the mobility prediction models of different motion modes of the whole system, and stores the updated models into the database classification model area (long-term updating, see figure 5);
(7) and after the cloud intelligent entity updates the model each time, the cloud intelligent entity informs the edge intelligent entity to download a new mobility data model for online prediction.
Referring to fig. 6, the short-term model updating process according to the embodiment of the present invention may include the following steps:
(1) the edge intelligent entity carries out user position prediction and evaluation: loading a corresponding mobility prediction model for a user needing prediction, performing real-time track prediction according to the current position and the historical position of the user, storing the current position into a database user data area, judging the performance of the model according to a certain judgment standard after multiple predictions, and informing a cloud intelligent entity to update the mobility model of the user if the performance of the model is poor;
(2) after receiving a model updating request of an edge intelligent entity, a cloud intelligent entity sends classified user data request information to a database;
(3) the database feeds back and provides classified user moving data to the cloud intelligent entity, wherein the data can comprise the user historical track information of the classification where the user is located;
(4) the cloud intelligent entity conducts retraining (or updating) of the mobility of the classified users;
(5) the cloud intelligent entity issues the updated model, stores the updated model in a user model cache region of the database, and informs the edge intelligent entity to use a new classification mobility model;
(6) the edge intelligent entity obtains a new classification mobility model.
Referring to fig. 7, the long-term model updating process according to the embodiment of the present invention may include the following steps:
(1) the cloud intelligent entity sends a long-term updating request to the database when one of the following conditions is met:
specifically, the long-term update request may be sent when the time from the last long-term update reaches a certain period T1, or the cloud-end intelligent entity receives short-term update requests of more than N categories within a certain time T2; here, T1, T2, and N are all preset values.
(2) The database sends the historical track data of the user to the cloud intelligent entity, and in addition, the historical track data of the user, the consumption habit data of the user, the package data and other attribute data can also be sent;
(3) the cloud intelligent entity reclassifies the users by using the historical track information of the users and retrains and updates the mobility prediction model of each classified user;
(4) the cloud intelligent entity issues the updated classification result and the mobility prediction model, stores the classification result and the mobility prediction model in a user model cache region of the database, and informs the edge intelligent entity to use a new classification mobility model;
(5) the edge intelligent entity obtains a new classification mobility model.
Referring to fig. 8, a specific example of a location prediction method according to an embodiment of the present invention is used to illustrate how an edge smart entity interacts with a cloud smart entity and a database. According to fig. 8, the whole interaction flow mainly includes the following steps:
(1) after detecting the user mobility prediction request, the edge intelligent entity checks whether a related mobility prediction model exists or not through the user identification or the user prediction data. The mobility prediction model refers to the LSTM deep cycle neural network corresponding to a certain class of users and the weights thereof.
(2) And if the edge intelligent entity has the relevant model, predicting and returning relevant data to the demand body. And if the relevant model does not exist, accessing the database to obtain the relevant model. The data used to predict time t may be the location p of the user at time t-1 (i.e., the current time)t-1Forming a sequence p ═ p together with the historical positions of the past length N-1t-N,pt-N+1,…,pt-1And the same-time data of the previous N days or historical N days sampled in a certain way can also be adopted.
(3) And after the edge intelligent entity finishes prediction, continuously tracking the actual position of the user. For an Active state user, acquiring a user network position or GPS data as a user actual position; for the Idle state user, if the GPS data can be obtained, the actual position of the user is taken as the actual position of the user, otherwise, the user is assumed to predict correctly.
(4) The edge intelligent entity detects the completed prediction times, if a certain threshold value T is reached1Then look at the prediction accuracy.
(5) If the edge intelligent entity finds that the prediction accuracy is lower than a certain threshold T2Then the database is checked for an updated mobility prediction model corresponding to that type of user. If so, acquiring, otherwise, notifying the cloud intelligent entity to retrain the model.
(6) And the cloud intelligent entity retrains the mobility prediction model of the user by using the database data. And after the training is finished, storing the relevant model to a database, and informing the relevant edge intelligent entity.
(7) And the cloud intelligent entity detects whether a long-term updating period is reached, and if the long-term updating period is reached, reclassification is carried out according to all data, and a user mobility data model is trained.
By the above description, it can be seen that the embodiments of the present invention provide:
an interactive process of an edge intelligent entity, a cloud intelligent entity and a database is used for classifying historical data according to motion modes, pre-training mobility prediction models of each class, selecting a proper mobility prediction model for a user to predict the position in real time, updating and improving the prediction model of each user through a short-term updating correction mechanism, and updating a unified prediction model of each class through a long-term updating mode.
Here, the short-term model updating process occurs when the edge smart entity evaluates the prediction result and determines that the prediction performance of the model is lower than a certain threshold, and then sends a short-term updating request to the cloud smart entity.
Here, the model long-term update process occurs when the cloud-end smart entity receives a short-term update process number that exceeds a certain threshold within a certain time or the cloud-end smart entity triggers periodically (e.g., a week or a month, as the case may be).
Here, the short-term update request may include a user classification flag, or may include information such as a user movement trajectory time stamp.
After receiving the short-term updating request, the cloud intelligent entity requests the database for classifying the user mobile data and updates the mobility prediction model; after the cloud intelligent entity finishes model training, storing the classified mobility prediction model into a database, and sending an update completion notification to the edge intelligent entity; and after receiving the update completion notification, the edge intelligent entity acquires the updated classification mobility prediction model from the database.
Here, after the cloud intelligent entity initiates a long-term update request, the database sends user mobile data to the cloud intelligent entity; the cloud intelligent entity reclassifies the mobile data and calculates a mobility prediction model for each type; after the cloud intelligent entity finishes model training, storing the classified mobility prediction model into a database, and sending an update completion notification to the edge intelligent entity; and after receiving the update completion notification, the edge intelligent entity acquires the updated classification mobility prediction model from the database.
Here, the selection of an appropriate mobility prediction model for a mobile user in the embodiment of the present invention means that the edge intelligent entity queries whether a prediction model of the user exists in a user model cache region of the database, and if so, selects the model; otherwise, loading the uniform mobility prediction model of the category from the database model area according to the category to which the user belongs.
Here, the mobility prediction model for each user classification is pre-trained in the embodiment of the present invention, an appropriate machine learning algorithm may be selected according to features of existing data, and the prediction model for each class is trained through the position sequence of the class in a supervised manner, and may be used for position prediction of any user in the class.
Here, the classifying according to the motion mode in the embodiment of the present invention refers to dividing the preprocessed track sequence into different categories according to a specific classification criterion, where the classification criterion includes a motion speed interval where the position sequence is located or a geographic position range where the position sequence is located.
Here, the evaluation of the prediction result in the embodiment of the present invention may refer to obtaining the actual position of the user, comparing the predicted position with the actual position, and if the distance error is smaller than a certain threshold, predicting correctly; otherwise the prediction is wrong.
Here, the prediction performance of the embodiment of the present invention refers to a ratio of the number of times of correct statistical prediction to the total number of times of prediction after reaching a specific number of times of prediction, that is, a prediction accuracy.
Here, the obtaining of the user real location according to the embodiment of the present invention may distinguish the state of the user equipment. If the user equipment is in a connected state (Active), a cell number where Active UE is located can be obtained through a network, and the position where the UE is located is judged according to the receiving intensity of a wireless signal; and mathematically obtains its true location using GPS information and other aiding information such as actual road information, building conditions, etc. If the user equipment is in an IDLE state, recording GPS information to determine the position of the user, and reporting and utilizing the GPS information; if no GPS signal is present, the predicted position is assumed to be the true position.
Here, the position sequence newly added to the database in the embodiment of the present invention refers to adding the actual position of the user acquired through the network or the GPS to the database user data area in real time in the estimation of the prediction result.
Various methods of embodiments of the present invention have been described above. An apparatus for carrying out the above method is further provided below.
An embodiment of the present invention provides an edge intelligent entity shown in fig. 9. Referring to fig. 9, an embodiment of the present invention provides an edge intelligent entity 90, including:
a first determining unit 91, configured to determine, according to a current position range and/or a motion speed interval of a target user of a to-be-predicted position, a user category to which the target user belongs;
a model selecting unit 92, configured to select, according to the user classification to which the target user belongs, a target mobility prediction model corresponding to the user classification to which the target user belongs, where the plurality of mobility prediction models respectively correspond to different user classifications, and a mobility prediction model corresponding to each user classification is obtained by a cloud-end intelligent entity through training according to user data in the user classification in a database, where the user data includes position data and/or speed data of the user;
and a location prediction unit 93, configured to obtain, according to the historical location and the current location of the target user, a predicted location of the target user at a prediction time by using the target mobility prediction model.
Preferably, the edge intelligent entity further comprises:
and the data collection unit is used for collecting the user data in the region and sending the user data to the database for storage.
Preferably, the edge intelligent entity further comprises:
a first receiving unit, configured to receive a first notification message sent by the cloud smart entity, where the first notification message is sent after the cloud smart entity retrains to obtain a mobility prediction model corresponding to each user classification according to user data stored in a database, where the retraining is performed periodically or is performed after a receiving frequency of an update request received by the cloud smart entity exceeds a predetermined threshold, and the update request is a request sent by each edge smart entity to update the mobility prediction model;
and the first obtaining unit is used for obtaining a mobility prediction model corresponding to each user classification obtained by retraining, which is stored in a database by the cloud intelligent entity, according to the first notification message.
Preferably, the edge intelligent entity further comprises:
the second acquisition unit is used for acquiring the actual position of the target user when the predicted time arrives;
the judging unit is used for judging whether the prediction accuracy of the target mobility prediction model meets a preset requirement or not according to the prediction positions and the actual positions of the plurality of prediction moments;
the first sending unit is used for sending an updating request for updating the target mobility prediction model to the cloud intelligent entity when the prediction accuracy of the target mobility prediction model does not meet the preset requirement.
Preferably, the second obtaining unit is further configured to:
when the target user is in a Radio Resource Control (RRC) activated state, acquiring an actual cell where the target user is located, and determining the actual position of the target user according to the received signal strength of the target user recorded by the actual cell, or determining the actual position of the target user according to the received signal strength of the target user, satellite positioning system information and other auxiliary information recorded by the actual cell;
and when the target user is in an RRC idle state, determining the actual position of the target user according to the satellite positioning system information of the target user, or taking the predicted position of the target user as the actual position of the target user.
Preferably, the edge intelligent entity further comprises:
and the position updating unit is used for updating the user data of the target user in the database according to the actual position of the target user after the actual position of the target user is determined.
Preferably, the edge intelligent entity further comprises:
a second receiving unit, configured to receive a second notification message sent by the cloud smart entity after updating the target mobility prediction model according to the update request;
a third obtaining unit, configured to obtain, according to the second notification message, the updated target mobility prediction model stored at a database by the cloud smart entity.
The edge intelligent entity provided by the embodiment of the invention can predict the position of the target user by utilizing a mobility prediction model of each user classification obtained by pre-training of the cloud intelligent entity. Compared with a model for predicting a single user by using the track of the single user, the embodiment of the invention only needs to maintain one mobility prediction model for the same type of users, thereby greatly reducing the calculation burden of model training; the dynamic updating process of the model can also solve the problem of inaccurate prediction caused by the change of the user movement rule. In addition, the embodiment of the invention can achieve better training effect under the condition that the initial data volume is not enough, relieves the pressure of early-stage data collection, and does not need to wait for the collection of historical data of a certain user, thereby realizing rapid position prediction. In addition, for a new user without a historical track, the embodiment of the invention can also judge the user motion mode through the real-time track of the user and select a proper model for prediction.
Referring to fig. 10, an embodiment of the present invention provides a structural schematic diagram of a cloud-end smart entity 100, and referring to fig. 10, an embodiment of the present invention provides a cloud-end smart entity 100, including:
a first obtaining unit 101, configured to obtain, according to a preset period, user data, stored in a database by each edge intelligent entity, in an area to which the edge intelligent entity belongs, where the user data includes position data and/or speed data of a user;
a classifying unit 102, configured to divide the user data into a plurality of user classifications according to a position range and a speed interval to which the user data belongs;
the first training unit 103 is configured to retrain the user data under each user category to obtain a mobility prediction model corresponding to the user category, store the mobility prediction model obtained through retraining in a database, and send a first notification message used for indicating that the model has been updated to the edge intelligent entity.
Preferably, the cloud intelligent entity further includes:
a first receiving unit, configured to receive an update request sent by a first edge intelligent entity for updating a first mobility prediction model of a first user class, where the update request is sent by the first edge intelligent entity when a prediction accuracy of the first mobility prediction model does not meet the preset requirement;
a second obtaining unit, configured to obtain, according to the update request, user data in the first user category stored in a database;
and the second training unit is used for retraining or updating the first mobility prediction model according to the user data under the first user classification, storing the updated first mobility prediction model in a database, and sending a second notification message for indicating that the model is updated to the first edge intelligent entity.
Preferably, the cloud intelligent entity further includes:
the judging unit is used for judging whether the receiving frequency of the updating requests sent by each edge intelligent entity exceeds a preset threshold or not; and when the receiving frequency exceeds the preset threshold, triggering the first acquisition unit to execute the step of acquiring the user data of the edge intelligent entity in the region of the edge intelligent entity stored at the database.
Preferably, the cloud intelligent entity further includes:
and the data storage unit is used for storing the user data under each user classification obtained by the division into a database after the user data is divided into a plurality of user classifications.
The cloud intelligent entity provided by the embodiment of the invention obtains the mobility prediction model of each user classification through pre-training, and then provides the mobility prediction model for the edge intelligent entity to predict the position of the target user. Compared with a model for predicting a single user by using the track of the single user, the embodiment of the invention only needs to maintain one mobility prediction model for the same type of users, thereby greatly reducing the calculation burden of model training; the dynamic updating process of the model can also solve the problem of inaccurate prediction caused by the change of the user movement rule. In addition, the embodiment of the invention can achieve better training effect under the condition that the initial data volume is not enough, relieves the pressure of early-stage data collection, and does not need to wait for the collection of historical data of a certain user, thereby realizing rapid position prediction. In addition, for a new user without a historical track, the embodiment of the invention can also judge the user motion mode through the real-time track of the user and select a proper model for prediction.
Referring to fig. 11, an embodiment of the present invention provides a structural diagram of a communication device 1100, including: a processor 1101, a transceiver 1102, a memory 1103, and a bus interface, wherein:
the processor 1101 is configured to read a program in the memory and execute the steps in the position prediction method shown in fig. 3 or fig. 4.
In fig. 11, the bus architecture may include any number of interconnected buses and bridges, with one or more processors, represented by processor 1101, and various circuits, represented by memory 1103, linked together. The bus architecture may also link together various other circuits such as peripherals, voltage regulators, power management circuits, and the like, which are well known in the art, and therefore, will not be described any further herein. The bus interface provides an interface. The transceiver 1102, which may be a plurality of elements including a transmitter and a receiver, provides a means for communicating with various other apparatus over a transmission medium.
The processor 1101 is responsible for managing the bus architecture and general processing, and the memory 1103 may store data used by the processor 1101 in performing operations.
Finally, the effectiveness of the location prediction method of an embodiment of the present invention is further illustrated by a simulation example.
Accuracy of rate characterization: the velocities were classified into three categories of 1m/s 4m/s 8m/s, yielding 50 different users at each velocity, and a 100-day trajectory was generated by the model for these 150 users. The position sequence was randomly intercepted from the daily trajectory in units of 1h (time interval 1min), thus giving a total of 15000 trajectory sequences of 60 length, of which 13500 were used for training and 1500 were used for testing. A Support Vector Machine (SVM) was used for the classification experiments, and the classification accuracy of the test set is shown in table 1. As can be seen from Table 1, the SVM can determine the category of the user with an accuracy of 95%.
| At a ratio of 1m/s | Is judged as a ratio of 4m/s | The ratio of 8m/s | |
| Practically 1m/s | 99.39％ | 0.61％ | 0 |
| Practically 4m/s | 0.80％ | 99.00％ | 0.20％ |
| Practically 8m/ |
0 | 0 | 100％ |
Table 1: user moving speed classification accuracy table
Tracks of 15 users lasting for 100 days are generated through a mobility prediction model, 3000 position sequences (time intervals are 1 minute) are obtained by randomly intercepting in 1h and are stored in a database user data area. The 3000 position sequences are classified into three categories of low speed, medium speed and high speed according to the speed, each category is pre-trained in a long-short term memory (LSTM) mode, and the trained models are stored in a memory category model area.
Two low speed users (user1, user2) in Active state are predicted. The method comprises the steps of loading a unified prediction model of the category from a database category model area, predicting the positions of two users in real time by using the model, comparing the predicted position with the real position when a network can acquire the real position of the user at the prediction time, and considering that the prediction is correct if the error is within a certain range (such as 30m, the error can be comprehensively determined according to the size of a cell and the required prediction precision), or else, judging that the prediction is wrong. And simultaneously saving the real position to a user data area of the database.
Calculating the average prediction accuracy (the accuracy is defined as the ratio of the correct prediction times to the actual prediction times) every 20 times of prediction, and if the prediction accuracy is lower than 60%, updating the model: and acquiring all position sequences of the user from a database, and continuing training on the basis of the existing prediction model by using an LSTM mode. And caching the updated model into a database user model area, and continuing to predict in real time by using the new model.
FIGS. 12-13 show the prediction accuracy before and after short-term updating for two users (user1 and user2), where the horizontal axis 0 represents the initial state, i.e., prediction using the unified class model; 1 denotes the prediction after a short-term update, 2, 3, 4 are similar; the vertical axis represents the statistical prediction accuracy after 20 real-time predictions. The strip blocks filled with diagonal line shading indicate the prediction accuracy using the initial category model, and the strip blocks filled with dotted shading indicate the prediction accuracy using the updated user model. It can be seen that the accuracy of the prediction of the positions of the different users by using the unified category model can basically reach more than 55%, and the prediction accuracy is improved to a certain extent after the model is dynamically updated for each user.
The simulation results fully show the feasibility and effectiveness of the pre-training and model updating mechanism according to the categories.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiments of the present invention.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U disk, a removable hard disk, a ROM, a RAM, a magnetic disk or an optical disk, and various media capable of storing program codes.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (12)
1. A position prediction method is applied to an edge intelligent entity, and is characterized by comprising the following steps:
the edge intelligent entity determines the user classification of the target user according to the current position range and/or the movement speed interval of the target user of the position to be predicted;
the edge intelligent entity selects a target mobility prediction model corresponding to the user classification to which the target user belongs according to the user classification to which the target user belongs, wherein a plurality of mobility prediction models respectively correspond to different user classifications, the mobility prediction model corresponding to each user classification is obtained by the cloud intelligent entity according to user data in the database under the user classification, and the user data comprises position data and/or speed data of the user;
the edge intelligent entity obtains the predicted position of the target user at the predicted time by using the target mobility prediction model according to the historical position and the current position of the target user;
the method further comprises the following steps:
the edge intelligent entity obtains the actual position of the target user when the predicted time arrives;
judging whether the prediction accuracy of the target mobility prediction model meets a preset requirement or not according to the predicted positions and the actual positions of the plurality of predicted moments;
and when the prediction accuracy of the target mobility prediction model does not meet the preset requirement, sending an updating request for updating the target mobility prediction model to the cloud intelligent entity.
2. The method of claim 1, further comprising:
and the edge intelligent entity collects the user data in the region and sends the user data to the database for storage.
3. The method of claim 2, further comprising:
the edge intelligent entity receives a first notification message sent by the cloud intelligent entity, the first notification message is sent after the cloud intelligent entity retrains according to user data stored in a database to obtain a mobility prediction model corresponding to each user classification, the retraining is executed periodically or is executed after the receiving frequency of update requests received by the cloud intelligent entity exceeds a preset threshold, and the update requests are requests sent by each edge intelligent entity for updating the mobility prediction model;
and the edge intelligent entity acquires a mobility prediction model corresponding to each user classification obtained by retraining the cloud intelligent entity stored in a database according to the first notification message.
4. The method of claim 1, wherein the step of obtaining the actual location of the target user comprises:
when the target user is in a Radio Resource Control (RRC) activated state, acquiring an actual cell where the target user is located, and determining the actual position of the target user according to the received signal strength of the target user recorded by the actual cell, or determining the actual position of the target user according to the received signal strength of the target user, satellite positioning system information and other auxiliary information recorded by the actual cell;
and when the target user is in an RRC idle state, determining the actual position of the target user according to the satellite positioning system information of the target user, or taking the predicted position of the target user as the actual position of the target user.
5. The method of claim 4, wherein after determining the actual location of the target user, the method further comprises:
and updating the user data of the target user in the database according to the actual position of the target user.
6. The method of claim 3, further comprising:
the edge intelligent entity receives a second notification message sent by the cloud intelligent entity after the target mobility prediction model is updated according to the updating request;
and the edge intelligent entity acquires the updated target mobility prediction model stored in a database by the cloud intelligent entity according to the second notification message.
7. A position prediction method is applied to a cloud intelligent entity and is characterized by comprising the following steps:
according to a preset period, user data of each edge intelligent entity in an area to which the edge intelligent entity belongs, wherein the edge intelligent entity is stored in a database, and the user data comprises position data and/or speed data of a user;
dividing the user data into a plurality of user classifications according to the position range and the speed interval to which the user data belongs;
according to user data under each user classification, respectively retraining to obtain a mobility prediction model corresponding to the user classification, storing the retrained mobility prediction model in a database, and sending a first notification message for indicating that the model is updated to the edge intelligent entity;
the method further comprises the following steps:
receiving an update request sent by a first edge intelligent entity for updating a first mobility prediction model of a first user class, wherein the update request is sent by the first edge intelligent entity when the prediction accuracy of the first mobility prediction model does not meet the preset requirement;
according to the updating request, user data stored in a database under the first user classification is obtained;
according to the user data under the first user classification, the first mobility prediction model is retrained or updated, the updated first mobility prediction model is stored in a database, and a second notification message used for indicating that the model is updated is sent to the first edge intelligent entity.
8. The method of claim 7, further comprising:
judging whether the receiving frequency of the updating requests sent by each edge intelligent entity exceeds a preset threshold or not;
and when the receiving frequency exceeds the preset threshold, returning to the step of acquiring the user data in the region of each edge intelligent entity stored in the database.
9. The method of claim 7, wherein after dividing the user data into a plurality of user classifications, the method further comprises: and storing the user data obtained by dividing under each user classification into a database.
10. An edge smart entity, comprising: memory, processor and computer program stored on the memory and executable on the processor, which computer program, when executed by the processor, carries out the steps of the position prediction method according to any one of claims 1 to 6.
11. A cloud-based smart entity, comprising: memory, processor and computer program stored on the memory and executable on the processor, which computer program, when executed by the processor, carries out the steps of the position prediction method according to any one of claims 7 to 9.
12. A computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, which computer program, when being executed by a processor, carries out the steps of the position prediction method according to any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910151995.4A CN111629319B (en) | 2019-02-28 | 2019-02-28 | Position prediction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910151995.4A CN111629319B (en) | 2019-02-28 | 2019-02-28 | Position prediction method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111629319A CN111629319A (en) | 2020-09-04 |
| CN111629319B true CN111629319B (en) | 2022-05-31 |
Family
ID=72272466
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910151995.4A Active CN111629319B (en) | 2019-02-28 | 2019-02-28 | Position prediction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111629319B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112255628A (en) * | 2020-10-09 | 2021-01-22 | 新石器慧义知行智驰(北京)科技有限公司 | Obstacle trajectory prediction method, apparatus, device, and medium |
| CN112702194B (en) * | 2020-12-16 | 2023-04-07 | 中国联合网络通信集团有限公司 | Indoor cell fault positioning method and device and electronic equipment |
| CN112714398B (en) * | 2021-01-26 | 2024-03-29 | 上海明略人工智能(集团)有限公司 | Method, device and equipment for correcting positioning coordinate drift of indoor positioning system |
| KR20220150691A (en) * | 2021-05-04 | 2022-11-11 | 현대자동차주식회사 | Apparatus for correcting position of vehicle and method thereof |
| CN116776986A (en) * | 2022-03-07 | 2023-09-19 | 维沃移动通信有限公司 | Model accuracy determining method and device and network side equipment |
| WO2023169402A1 (en) * | 2022-03-07 | 2023-09-14 | 维沃移动通信有限公司 | Model accuracy determination method and apparatus, and network side device |
| CN116776985A (en) * | 2022-03-07 | 2023-09-19 | 维沃移动通信有限公司 | Model accuracy determining method and device and network side equipment |
| CN116776987A (en) * | 2022-03-07 | 2023-09-19 | 维沃移动通信有限公司 | Model accuracy determining method and device and network side equipment |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102567391A (en) * | 2010-12-20 | 2012-07-11 | 中国移动通信集团广东有限公司 | Method and device for building classification forecasting mixed model |
| CN104348855A (en) * | 2013-07-29 | 2015-02-11 | 华为技术有限公司 | User information processing method, mobile terminal and server |
| CN104838708A (en) * | 2012-12-14 | 2015-08-12 | 华为技术有限公司 | Systems and methods for user equipment mobility prediction |
| CN105718750A (en) * | 2016-01-29 | 2016-06-29 | 长沙理工大学 | Prediction method and system for vehicle travelling track |
| CN107194412A (en) * | 2017-04-20 | 2017-09-22 | 百度在线网络技术（北京）有限公司 | A kind of method of processing data, device, equipment and computer-readable storage medium |
| EP3260349A1 (en) * | 2011-03-18 | 2017-12-27 | The Raymond Corporation | Dynamic stability control method for lift trucks |
| CN107563429A (en) * | 2017-07-27 | 2018-01-09 | 国家计算机网络与信息安全管理中心 | A kind of sorting technique and device of network user colony |
| WO2018022313A1 (en) * | 2016-07-29 | 2018-02-01 | Apple Inc. | Systems, devices, and methods for dynamically providing user interface controls at a touch-sensitive secondary display |
| CN108427698A (en) * | 2017-08-29 | 2018-08-21 | 平安科技（深圳）有限公司 | Updating device, method and the computer readable storage medium of prediction model |
| CN108616816A (en) * | 2018-05-07 | 2018-10-02 | 重庆邮电大学 | A kind of track of vehicle prediction and MEC application moving methods based on limit study |
| CN108761509A (en) * | 2018-04-09 | 2018-11-06 | 吉林大学 | A kind of traval trace and mileage prediction technique based on historical data |
| CN109005223A (en) * | 2018-07-26 | 2018-12-14 | 南京邮电大学 | Internet of Things resource regulating method and system, computer readable storage medium and terminal |
-
2019
- 2019-02-28 CN CN201910151995.4A patent/CN111629319B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102567391A (en) * | 2010-12-20 | 2012-07-11 | 中国移动通信集团广东有限公司 | Method and device for building classification forecasting mixed model |
| EP3260349A1 (en) * | 2011-03-18 | 2017-12-27 | The Raymond Corporation | Dynamic stability control method for lift trucks |
| CN104838708A (en) * | 2012-12-14 | 2015-08-12 | 华为技术有限公司 | Systems and methods for user equipment mobility prediction |
| CN104348855A (en) * | 2013-07-29 | 2015-02-11 | 华为技术有限公司 | User information processing method, mobile terminal and server |
| CN105718750A (en) * | 2016-01-29 | 2016-06-29 | 长沙理工大学 | Prediction method and system for vehicle travelling track |
| WO2018022313A1 (en) * | 2016-07-29 | 2018-02-01 | Apple Inc. | Systems, devices, and methods for dynamically providing user interface controls at a touch-sensitive secondary display |
| CN107194412A (en) * | 2017-04-20 | 2017-09-22 | 百度在线网络技术（北京）有限公司 | A kind of method of processing data, device, equipment and computer-readable storage medium |
| CN107563429A (en) * | 2017-07-27 | 2018-01-09 | 国家计算机网络与信息安全管理中心 | A kind of sorting technique and device of network user colony |
| CN108427698A (en) * | 2017-08-29 | 2018-08-21 | 平安科技（深圳）有限公司 | Updating device, method and the computer readable storage medium of prediction model |
| CN108761509A (en) * | 2018-04-09 | 2018-11-06 | 吉林大学 | A kind of traval trace and mileage prediction technique based on historical data |
| CN108616816A (en) * | 2018-05-07 | 2018-10-02 | 重庆邮电大学 | A kind of track of vehicle prediction and MEC application moving methods based on limit study |
| CN109005223A (en) * | 2018-07-26 | 2018-12-14 | 南京邮电大学 | Internet of Things resource regulating method and system, computer readable storage medium and terminal |
Non-Patent Citations (4)
| Title |
|---|
| 《Cluster-Aided Mobility Predictions》;Jaeseong Jeong;《IEEE INFOCOM 2016》;20161231;全文 * |
| 《Mobile crowd location Prediction with hybrid features using ensemble learning》;Zhongliang Zhao;《ELSEVIER》;20180622;全文 * |
| 室内用户语义位置模式挖掘研究――以商场为例;张心悦等;《测绘与空间地理信息》;20160225(第02期);全文 * |
| 时空数据语义理解:技术与应用;姚迪等;《软件学报》;20180416(第07期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111629319A (en) | 2020-09-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111629319B (en) | Position prediction method and device | |
| CN107333232B (en) | Terminal positioning method and network equipment | |
| US8634853B2 (en) | Method for enhancing location identity through incorporation of shorter-range communication and sensing (nearlocate) | |
| US20160227367A1 (en) | Method of providing positioning data to mobile device | |
| US20160205509A1 (en) | Mechanism for determining location history via multiple historical predictors | |
| WO2020215783A1 (en) | Locating method and apparatus, and storage medium | |
| EP3045001B1 (en) | Methods, wireless device and network node for managing positioning method based on prediction | |
| CN108668224A (en) | Base station location determines method, apparatus, server and storage medium | |
| US10963917B2 (en) | Method and system for determining fact of visit of user to point of interest | |
| US11856552B2 (en) | Devices and methods for automatically labelling high-accuracy indoor localization and determining location information | |
| US10885532B2 (en) | Facilitating demographic assessment of information using targeted location oversampling | |
| US20160192155A1 (en) | Facilitating estimation of mobile device presence inside a defined region | |
| JP2013197704A (en) | Traffic amount prediction processing device and computer program | |
| CN113316087A (en) | Dynamic paging method based on terminal position prediction in LTE (Long term evolution) system | |
| US20140378160A1 (en) | Method of correcting global position error | |
| Zhang et al. | Gaussian process based radio map construction for LTE localization | |
| JP6433059B2 (en) | Apparatus, program and method for positioning mobile terminal based on communication record | |
| EP4104324A1 (en) | Method and system for estimating indoor radio transmitter count | |
| Iwata et al. | A stayed location estimation method for sparse GPS positioning information | |
| EP4250181A1 (en) | Utilizing machine learning models to estimate user device spatiotemporal behavior | |
| US11914065B2 (en) | Mobile signal based building footprints | |
| EP4322637A1 (en) | Machine learning model validation for ue positioning based on reference device information for wireless networks | |
| US20240135247A1 (en) | Method and Apparatus for Selecting Machine Learning Model for Execution in a Resource Constraint Environment | |
| Xia et al. | A generative learning approach for spatio-temporal modeling in connected vehicular network | |
| CN107172637A (en) | A kind of method and apparatus classified to calling |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |