WO2021129634A1

WO2021129634A1 - Network positioning method and system

Info

Publication number: WO2021129634A1
Application number: PCT/CN2020/138463
Authority: WO
Inventors: 束纬寰; 林宇; 尹卜一; 冯朝阳; 马利; 石立臣; 柴华
Original assignee: 北京嘀嘀无限科技发展有限公司
Priority date: 2019-12-24
Filing date: 2020-12-23
Publication date: 2021-07-01
Also published as: CN111836358A; CN111836358B

Abstract

Disclosed is a network positioning method. The method comprises: acquiring a target positioning request; acquiring first information associated with the target positioning request, wherein the first information at least comprises a signal feature associated with the target positioning request; determining, on the basis of the first information, a preset point associated with the target positioning request; generating at least one first feature graph on the basis of the preset point; determining at least one second feature graph on the basis of the at least one first feature graph; and processing the at least one second feature graph to acquire a target position, wherein the processing at least comprises processing the at least one second feature graph on the basis of a convolution kernel.

Description

Network positioning method and system

cross reference

This application claims the priority of the Chinese application number 201911350794.3 filed on December 24, 2019, the entire content of which is incorporated herein by reference.

Technical field

This manual relates to the computer field, and in particular to a network positioning method and system.

Background technique

With the development of the mobile Internet, a large number of location-based services (for example, online car-hailing services, food delivery services) have emerged. In such services, the accuracy of locating target devices, terminals, or locations is crucial. However, if GPS/GNSS, NLP (Network Location Provider) and other methods are used for positioning, the signal strength may be unstable and the signal is unavailable; if the machine learning model is combined with the network signal characteristics for positioning, the network signal characteristics are It may cause a large loss during processing, which makes it difficult to ensure the accuracy of positioning.

Therefore, it is desirable to propose a network positioning method and system to obtain accurate target positioning.

Summary of the invention

One of the embodiments of this specification provides a network positioning method. The method includes: acquiring a target positioning request; acquiring first information associated with the target positioning request, the first information including at least: signal characteristics associated with the target positioning request; based on the first information , Determining a preset point associated with the target positioning request; generating at least one first feature map based on the preset point; determining at least one second feature map based on the at least one first feature map; and, The at least one second feature map is processed to obtain a target location, and the processing at least includes: processing the at least one second feature image based on a convolution kernel.

One of the embodiments of this specification provides a network positioning system, the system includes: at least one storage medium, the storage medium includes an instruction set for network positioning; at least one processor, the at least one processor and the At least one storage medium communication, wherein, when the instruction set is executed, the at least one processor is configured to: obtain a target positioning request; obtain first information associated with the target positioning request, the first information At least including: signal characteristics associated with the target positioning request; determining a preset point associated with the target positioning request based on the first information; generating at least one first characteristic map based on the preset point; Determine at least one second feature map based on the at least one first feature map; and process the at least one second feature map to obtain target positioning, and the processing at least includes: checking the at least one second feature map based on a convolution check. The second feature image is processed.

One of the embodiments of this specification provides a network positioning system. The system includes: a first acquisition module for acquiring a target positioning request; a second acquisition module for acquiring first information associated with the target positioning request , The first information includes at least: a signal characteristic associated with the target positioning request; a first determining module, configured to determine a preset point associated with the target positioning request based on the first information; generating A module, configured to generate at least one first feature map based on the preset point; a second determining module, configured to determine at least one second feature map based on the at least one first feature map; and, a third acquiring module, It is used to process the at least one second feature map to obtain target positioning, and the processing at least includes: processing the at least one second feature image based on a convolution kernel.

One of the embodiments of this specification provides a computer-readable storage medium that stores computer instructions for network positioning. When the computer reads the computer-executed instructions for network positioning in the storage medium, the computer executes the above-mentioned technology. The method described in the scheme.

Description of the drawings

This specification will be further described in the form of exemplary embodiments, and these exemplary embodiments will be described in detail with the accompanying drawings. These embodiments are not restrictive. In these embodiments, the same number represents the same structure, in which:

Fig. 1 is a schematic diagram of an application scenario of a network positioning system according to some embodiments of this specification;

Fig. 2 is a schematic diagram of exemplary hardware and/or software of an exemplary computing device according to some embodiments of the present specification;

Fig. 3 is a schematic diagram of exemplary hardware and/or software of an exemplary mobile device according to some embodiments of the present specification;

Fig. 4 is a block diagram of an exemplary processing device according to some embodiments of this specification;

Fig. 5 is a flowchart of an exemplary process of a network positioning method according to some embodiments of this specification;

Fig. 6 is a flowchart of an exemplary process of determining a preset point according to some embodiments of the present specification;

Fig. 7 is a flowchart of an exemplary process of determining similar grids according to some embodiments of the present specification;

FIG. 8 is a schematic diagram of a machine learning model of related technologies;

Fig. 9 is a flowchart of an exemplary process of a network positioning method according to some embodiments of this specification;

FIG. 10 is a flowchart of an exemplary process of determining a center point according to some embodiments of the present specification;

Fig. 11 is a flowchart of an exemplary process of determining an input grid according to some embodiments of the present specification;

Fig. 12 is a flowchart of another exemplary process of determining an input grid according to some embodiments of the present specification;

Fig. 13 is a schematic diagram of a characteristic diagram according to some embodiments of the present specification;

FIG. 14 is a flowchart of an exemplary process of training a convolutional neural network model according to some embodiments of this specification;

FIG. 15 is a flowchart of an exemplary process of training a convolutional neural network model according to some embodiments of this specification;

FIG. 16 is a schematic diagram of an exemplary process of determining target location and confidence based on a convolutional neural network model according to some embodiments of this specification;

FIG. 17 is a schematic diagram of an exemplary process of determining target location based on a convolutional neural network model according to some embodiments of the present specification;

Fig. 18 is a block diagram of a positioning device according to some embodiments of this specification.

Detailed ways

In order to more clearly describe the technical solutions of the embodiments of the present specification, the following will briefly introduce the accompanying drawings used in the description of the embodiments. Obviously, the drawings in the following description are only some examples or embodiments of this specification. For those of ordinary skill in the art, without creative work, this specification can also be applied to these drawings. Other similar scenarios. Unless it is obvious from the language environment or otherwise stated, the same reference numerals in the figures represent the same structure or operation.

It should be understood that the “system”, “device”, “unit” and/or “module” used herein is a method for distinguishing different components, elements, parts, parts, or assemblies of different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

As shown in this specification and claims, unless the context clearly indicates exceptions, the words "a", "an", "an" and/or "the" do not specifically refer to the singular, but may also include the plural. Generally speaking, the terms "include" and "include" only suggest that the clearly identified steps and elements are included, and these steps and elements do not constitute an exclusive list, and the method or device may also include other steps or elements.

The term "network positioning" refers to a fixed location based on the communication network infrastructure of the bottom, which does not rely on the satellite positioning system, and achieves more accurate positioning based on the signals of the communication network infrastructure. There are two ways of network positioning. One is WIFI positioning, which is based on the location of the WIFI router. This positioning accuracy is relatively high (because the range of a WIFI cell is only tens of meters), but it is not reliable because there is no way to record The location of each router on the earth, so from time to time there is the phenomenon of positioning to other places, even other provinces and cities. Another method of network positioning is base station positioning. This positioning is reliable, but the error is large, because this positioning method depends on the distribution density of base stations. The positioning accuracy of urban areas in developed areas will be relatively high, and currently the highest can reach within a few tens of meters to within a hundred meters. However, when the base stations in remote areas are distributed with a relatively large distance, the error will be large, sometimes even more than several kilometers. The common feature of network positioning is fast speed. As long as it is connected to the Internet, it can be located in an instant, and any mobile phone can be determined in seconds.

The term "fingerprint" refers to the characteristics of the wireless network signal obtained during network positioning (for example, the mac address and signal strength of the WIFI).

The term "convolutional neural network (CNN)" refers to a feed-forward neural network, which is composed of several convolutional layers and pooling layers, and has achieved great success in the field of computer vision. It has the characteristics of local area connection, weight sharing, down-sampling and so on. CNN reduces the number of weights that need to be trained through weight sharing, reduces the computational complexity of the network, and at the same time makes the network have certain invariance to the local transformation of the input, such as translation invariance, scaling invariance, etc., which improves the network The generalization ability. CNN can automatically input raw data directly into the network, and then implicitly learn from the training data, avoiding manual feature extraction.

In this specification, a flowchart is used to illustrate the operations performed by the system according to the embodiment of this specification. It should be understood that the preceding or following operations are not necessarily performed exactly in order. Instead, the steps can be processed in reverse order or at the same time. At the same time, other operations can be added to these processes, or a certain step or several operations can be removed from these processes.

Fig. 1 is a schematic diagram of an application scenario of a network positioning system according to some embodiments of this specification. The network positioning system 100 can determine the location of the user. In some embodiments, the network positioning system 100 may include a server 110, a network 120, a user terminal 130, and a storage device 140.

The server 110 may process data and/or information from at least one component of the network positioning system 100. For example, the user terminal 130 may receive the user's location request and send it to the server 110, and the server 110 processes the user's location request to obtain the user's location.

In some embodiments, the server 110 may be a single processing device or a group of processing devices. The processing device group may be a centralized processing device group connected to the network 120 via an access point, or a distributed processing device group respectively connected to the network 120 via at least one access point. In some embodiments, the server 110 may be locally connected to the network 120 or remotely connected to the network 120. For example, the server 110 may access information and/or data stored in the user terminal 130 and/or the storage device 140 via the network 120. For another example, the storage device 140 may be used as a back-end data storage of the server 110. In some embodiments, the server 110 may be implemented on a cloud platform. For example only, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-layer cloud, etc., or any combination thereof.

In some embodiments, the server 110 may include a processing device 112. The processing device 112 may process information and/or data related to at least one function described in this application. In some embodiments, the processing device 112 may perform the main functions of the network positioning system 100. In some embodiments, the processing device 112 can obtain the location of the user according to the location request of the user. In some embodiments, the processing device 112 may perform other functions related to the methods and systems described in this application. In some embodiments, the processing device 112 may include at least one processing unit (for example, a single-core processing device or a multi-core processing device). For example only, the processing device 112 includes a central processing unit (CPU), an application specific integrated circuit (ASIC), an application specific instruction set processor (ASIP), a graphics processing unit (GPU), a physical processing unit (PPU), and a digital signal processor. (DSP), Field Programmable Gate Array (FPGA), Programmable Logic Device (PLD), Controller, Microcontroller Unit, Reduced Instruction Set Computer (RISC), Microprocessor, etc., or any combination thereof.

The network 120 may facilitate the exchange of information and/or data. In some embodiments, at least one component in the network positioning system 100 (for example, the server 110, the user terminal 130, and the storage device 140) may send information and/or data to other components in the network positioning system 100 via the network 120. For example, the processing device 112 may obtain the first information related to the target positioning request from the storage device 140 via the network 120.

In some embodiments, the network 120 may be any form of wired or wireless network, or any combination thereof. For example only, the network 120 may include a cable network, a wired network, an optical fiber network, a telecommunication network, an internal network, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network ( MAN), public switched telephone network (PSTN), Bluetooth network, ZigBee network, near field communication (NFC) network, etc. or any combination thereof. In some embodiments, the network 120 may include at least one network access point. For example, the network 120 may include wired or wireless network access points, such as base stations and/or Internet exchange points 120-1, 120-2, ..., and at least one component of the network positioning system 100 may be connected to the network 120 to exchange data. And/or information.

The user terminal 130 may obtain the user's target positioning request and the first information associated with the target positioning request. In some embodiments, the user's positioning request is a network-based positioning request, and the first information includes at least signal characteristics associated with the target positioning request. In some embodiments, the user can actively initiate a target positioning request through the user terminal 130, and the active initiation methods include, but are not limited to, clicking a positioning button, touching a positioning button, checking a positioning option, voice inputting a positioning request, and so on. In some embodiments, the user terminal 130 may automatically initiate a positioning request for the user. For example, the user navigates through the navigation software on the user terminal 130, and the navigation software can automatically initiate a positioning request during the navigation process.

In some embodiments, in some embodiments, the user terminal 130 may include a mobile device 130-1, a tablet computer 130-2, a laptop computer 130-3, etc., or any combination thereof. In some embodiments, the mobile device 130-1 may include a smart home device, a wearable device, a smart mobile device, a virtual reality device, an augmented reality device, etc., or any combination thereof. In some embodiments, the smart home equipment may include smart lighting equipment, smart electrical appliance control devices, smart monitoring equipment, smart TVs, smart cameras, walkie-talkies, etc., or any combination thereof. In some embodiments, the wearable device may include smart bracelets, smart footwear, smart glasses, smart helmets, smart watches, smart clothes, smart backpacks, smart accessories, etc., or any combination thereof. In some embodiments, the smart mobile device may include a smart phone, a personal digital assistant (PDA), a gaming device, a navigation device, a point of sale (POS), etc., or any combination thereof. In some embodiments, the virtual reality device and/or augmented virtual reality device may include a virtual reality helmet, virtual reality glasses, virtual reality goggles, augmented reality helmets, augmented reality glasses, augmented reality goggles, etc., or any combination thereof. For example, the virtual reality device and/or the augmented reality device may include Google Glass ^(TM) , Oculus Rift ^(TM) , Hololens ^(TM) or Gear VR ^(TM), etc. In some embodiments, the in-vehicle device 130-4 may include an in-vehicle computer, an in-vehicle TV, and the like. In some embodiments, the user terminal 130 may be a device with positioning technology for locating the location of the service requester and/or the user terminal 130.

The storage device 140 may store data and/or instructions. For example, first information, second information, preset grids, etc. can be stored. In some embodiments, the storage device 140 may store data and/or instructions executable by the processing device 112, and the server 110 may execute or use the data and/or instructions to implement the exemplary methods described in this application. In some embodiments, the storage device 140 may include mass storage, removable storage, volatile read-write storage, read-only storage (ROM), etc., or any combination thereof. Exemplary mass storage devices may include magnetic disks, optical disks, solid state disks, and the like. Exemplary removable storage may include flash drives, floppy disks, optical disks, memory cards, compact disks, magnetic tapes, and the like. An exemplary volatile read-write memory may include random access memory (RAM). Exemplary RAM may include dynamic random access memory (DRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), static random access memory (SRAM), thyristor random access memory (T-RAM), and zero capacitance Random access memory (Z-RAM), etc. Exemplary read-only memory may include mask-type read-only memory (MROM), programmable read-only memory (PROM), erasable programmable read-only memory (PEROM), electrically erasable programmable read-only memory (EEPROM), CD-ROM and digital versatile disk read-only memory, etc. In some embodiments, the storage device 140 may be implemented on a cloud platform. For example only, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-layer cloud, etc., or any combination thereof.

Fig. 2 is a schematic diagram of exemplary hardware and/or software of an exemplary computing device according to some embodiments of the present specification. In some embodiments, the server 110 or the user terminal 130 may be implemented on the computing device 200. For example, the processing device 112 may implement and execute the functions of the processing device 112 disclosed in this specification on the computing device 200. As shown in FIG. 2, the computing device 200 may include a bus 210, a processor 220, a read-only memory 230, a random access memory 240, a communication port 250, an input/output 260, and a hard disk 270.

The processor 220 can execute calculation instructions (program code) and perform the functions of the network positioning system 100 described in this specification. The calculation instructions may include programs, objects, components, data structures, procedures, modules, functions (the functions refer to the specific functions described in this specification), and the like. For example, the processor 220 may process image or text data obtained from any other components of the network positioning system 100. In some embodiments, the processor 220 may include a microcontroller, a microprocessor, a reduced instruction set computer (RISC), an application specific integrated circuit (ASIC), an application specific instruction set processor (ASIP), a central processing unit (CPU) , Graphics processing unit (GPU), physical processing unit (PPU), microcontroller unit, digital signal processor (DSP), field programmable gate array (FPGA), advanced RISC machine (ARM), programmable logic device, and Any circuits and processors that perform one or more functions, etc., or any combination thereof. For illustration only, the computing device 200 in FIG. 2 only describes one processor, but it should be noted that the computing device 200 in this specification may also include multiple processors.

The memory of the computing device 200 (for example, read only memory (ROM) 230, random access memory (RAM) 240, hard disk 270, etc.) may store data/information acquired from any other components of the network positioning system 100. Exemplary ROMs may include mask ROM (MROM), programmable ROM (PROM), erasable programmable ROM (PEROM), electrically erasable programmable ROM (EEPROM), compact disk ROM (CD-ROM), and digital Universal disk ROM, etc. Exemplary RAM may include dynamic RAM (DRAM), double rate synchronous dynamic RAM (DDR SDRAM), static RAM (SRAM), thyristor RAM (T-RAM), zero capacitance (Z-RAM), and the like.

The input/output 260 may be used to input or output signals, data or information. In some embodiments, the input/output 260 may include an input device and an output device. Exemplary input devices may include a keyboard, a mouse, a touch screen, a microphone, etc., or any combination thereof. Exemplary output devices may include display devices, speakers, printers, projectors, etc., or any combination thereof. Exemplary display devices may include liquid crystal displays (LCD), light emitting diode (LED) based displays, flat panel displays, curved displays, television equipment, cathode ray tubes (CRT), etc., or any combination thereof.

The communication port 250 can be connected to a network for data communication. The connection can be a wired connection, a wireless connection, or a combination of both. Wired connections can include cables, optical cables, telephone lines, etc., or any combination thereof. The wireless connection may include Bluetooth, WIFI, WiMax, WLAN, ZigBee, mobile networks (for example, 3G, 4G, or 5G, etc.), etc., or any combination thereof. In some embodiments, the communication port 250 may be a standardized port, such as RS232, RS485, and so on. In some embodiments, the communication port 250 may be a specially designed port.

Fig. 3 is a schematic diagram of exemplary hardware and/or software of an exemplary mobile device according to some embodiments of the present specification. As shown in FIG. 3, the mobile device 300 may include a communication unit 310, a display unit 320, a graphics processing unit (GPU) 330, a central processing unit (CPU) 340, an input/output unit 350, a memory 360, a storage unit 370, and the like. In some embodiments, the mobile device 300 may also include any other suitable components, including but not limited to a system bus or a controller (not shown in the figure). In some embodiments, the operating system 361 (for example, iOS, Android, Windows Phone, etc.) and the application program 362 may be loaded from the storage unit 370 into the memory 360 so as to be executed by the CPU 340. The application program 362 may include a browser or an application program for receiving text, image, audio, or other related information from the network positioning system 100. The user interaction of the information flow may be implemented through the input/output unit 350 and provided to the processing device 112 and/or other components of the network positioning system 100 through the network 120.

In order to realize the various modules, units and functions described in this specification, a computing device or a mobile device can be used as a hardware platform for one or more components described in this specification. The hardware components, operating systems, and programming languages of these computers or mobile devices are conventional in nature, and those skilled in the art can adapt these technologies to the system described in this specification after being familiar with these technologies. A computer with user interface elements can be used to implement a personal computer (PC) or other types of workstations or terminal devices, and if properly programmed, the computer can also act as a server.

Fig. 4 is a block diagram of an exemplary processing device according to some embodiments of the present specification. As shown in FIG. 4, the processing device 112 of the network positioning system 100 may include a first acquiring module 410, a second acquiring module 420, a first determining module 430, a generating module 440, a second determining module 450, a third acquiring module 460, and The third determining module 470.

The first obtaining module 410 may be used to obtain a target positioning request. For more details about the target location request, please refer to step 510 and its related description, which will not be repeated here.

The second obtaining module 420 may obtain first information associated with the target positioning request, where the first information includes at least: signal characteristics associated with the target positioning request. In some embodiments, the first information may further include one or more of the following associated with the target positioning request: popularity information, minimum signal strength, maximum signal strength, crowd density, center grid pixel value size, characteristics The total size of the values of all grids in the graph. In some embodiments, the first information may also include the communication relationship between the target positioning request and the primary base station, the sum of the signal strength matching probabilities of multiple base stations adjacent to the primary base station, and the previous base station of the primary base station. At least one of the center distance, the communication relationship with the former base station, and the sum of heat information with the former base station. Wherein, the primary base station is the base station to which the terminal corresponding to the target positioning request is currently connected, and the previous base station is the base station to which the terminal is connected before connecting to the primary base station. For more details about the first information, refer to step 520 and related descriptions.

The first determining module 430 may be configured to determine a preset point associated with the target positioning request based on the first information. In some embodiments, the first determination module 430 may also be used to obtain multiple signal strengths of the target positioning request, and based on the multiple signal strengths, determine a preset point associated with the target positioning request. In some embodiments, the first determining module 430 may be used to obtain second information including a plurality of preset grids, calculate the similarity between the first information and the second information, determine at least one similar grid based on the similarity, and , Determine the preset point based on at least one similar grid. In some embodiments, the first determining module 430 may be further configured to sort a plurality of preset grids based on similarity to obtain a similarity ranking result, and determine at least one similar grid based on the similarity ranking result. In some embodiments, the first determining module 430 may also be used to determine whether the similarity satisfies a preset condition, and if so, the preset grid whose similarity satisfies the condition is determined as a similar grid. In some embodiments, the first determining module 430 may be further configured to determine the preset point based on the median, average, and geometric center of the position coordinates of the at least one similar grid. In some embodiments, the first determining module 430 may be further configured to determine the weight of the plurality of preset grids based on the similarity, and determine the preset point based on the weight and the similarity. For more details about the preset points and similar grids, refer to step 530, step 610 and related descriptions, which will not be repeated here.

The generating module 440 is configured to generate at least one first feature map based on preset points. The first feature map may include feature information related to the target positioning request. For more details about the first feature map, please refer to step 540 and related descriptions, which will not be repeated here.

The second determining module 450 is configured to determine at least one second characteristic map based on the at least one first characteristic map. In some embodiments, the second determining module 450 may also be used to obtain multiple first feature maps, and perform fusion processing on the multiple first feature maps to obtain at least one second feature map. For more details about the second feature map, please refer to step 550 and related descriptions, which will not be repeated here.

The third acquiring module 460 is configured to process at least one second feature map to acquire target positioning, and the processing at least includes: processing at least one second feature image based on a convolution kernel. The third acquisition module 460 may also be used to determine the position correction information through a convolutional neural network model based on the at least one second feature map. The convolutional neural network model includes the convolution kernel, and acquires the position correction information based on the position correction information and preset points. The target positioning. For more details about processing the second feature map, refer to step 560 and related descriptions, which will not be repeated here.

The third determining module 470 is configured to determine the confidence of the target positioning based on at least the confidence network model and at least one second feature map. In some embodiments, the confidence network model is used to output the confidence of the target location according to the proportion of non-empty feature pixels of the at least one second feature map. For more details about the confidence level and the confidence level network model, please refer to Figure 5 and its related descriptions, which will not be repeated here.

It should be understood that the system and its modules shown in FIG. 4 can be implemented in various ways. For example, in some embodiments, the system and its modules may be implemented by hardware, software, or a combination of software and hardware. Among them, the hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those skilled in the art can understand that the above-mentioned methods and systems can be implemented using computer-executable instructions and/or included in processor control code, for example on a carrier medium such as a disk, CD or DVD-ROM, such as a read-only memory (firmware Such codes are provided on a programmable memory or a data carrier such as an optical or electronic signal carrier. The system and its modules in this specification can not only be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc. It may also be implemented by software executed by various types of processors, or may be implemented by a combination of the above hardware circuit and software (for example, firmware).

It should be noted that the above description of the processing device 112 and its modules of the network positioning system 100 is only for convenience of description, and does not limit this specification within the scope of the examples mentioned. It can be understood that for those skilled in the art, after understanding the principle of the system, it is possible to arbitrarily combine various modules, or form a subsystem to connect with other modules without departing from this principle. For example, the first acquisition module 410 and the second acquisition module 420 may be two different modules, or may be combined into the same module. Such deformations are all within the protection scope of this specification.

Fig. 5 is a flowchart of an exemplary process of a network positioning method according to some embodiments of the present specification. In some embodiments, process 500 may be performed by a processing device (eg, processing device 112 or other processing device). For example, the process 500 may be stored in a storage device (for example, the storage device 140 or a storage unit of a processing device) in the form of a program or instruction. When the processor 220 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 500. In some embodiments, the process 500 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 5 is not restrictive.

Step 510: Obtain a target positioning request. In some embodiments, step 510 may be performed by the first obtaining module 410.

The location request can be any request based on location services. For example, request to locate the origin coordinates, request to locate the destination coordinates, and so on. In some embodiments, the target location request may be a location request issued by a target user in a transportation service (e.g., taxi service, courier service, taxi-hailing service), or it may be related to a target location (e.g., place of departure, destination). Location) related positioning request.

The positioning request can be a real-time request, a reservation request, etc., or any combination thereof. As used herein, a real-time request may include a service that the requester expects to receive at the current moment or at a designated time close to the current moment. For example, if the specified time is within a certain period of time from the current time, such as 5 minutes from the current time, 10 minutes from the current time, or 20 minutes from the current time, the positioning request may be a real-time request. The reservation request may include the service that the requester expects to receive at a future time at the current moment. For example, if the service is booked in the future time, for example, within a specified time after the current time, the positioning request may be a reservation request. The specified time period can be 20 minutes after the current time, 2 hours after the current time, or 1 day after the current time. In some embodiments, the server 110 may specify a real-time request or a reservation request based on a time threshold. The time threshold can be a default value of the system 100, or can be adjusted according to different situations. For example, during peak traffic hours, the time threshold can be set smaller (for example, 10 minutes), during off-peak hours (for example, 10:00-12:00 in the morning), the time threshold can be set larger (for example, 1 hour) ).

There are many ways to obtain the target location request. In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first obtaining module 410)) may obtain the target location request initiated by the user from the user terminal 130. Active initiation methods may include, but are not limited to, clicking a positioning button, touching a positioning button, checking a positioning option, voice inputting a positioning request, and so on. For example, the user may open the software (such as navigation software, travel software, etc.) installed on the user terminal 130 that has the permission to obtain the location, and in response to the user's operation of opening the software, the positioning request is initiated. In some embodiments, the positioning request may be automatically initiated by the user terminal 130. For example, when the user uses navigation software, the user terminal 130 may automatically initiate a positioning request at a preset time interval to determine the user's real-time position and realize navigation.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first obtaining module 410)) may obtain the target positioning request through the server 110. In some embodiments, the server 110 may determine which technology to use for positioning according to the information. Network positioning may include base station-based positioning, WIFI-based positioning, and so on.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first obtaining module 410)) may obtain the target positioning request in the form of network positioning through the network 120. Taking network positioning as an example, the acquired target positioning request may include the user's current network data. The user's current network data can be collected by the user terminal 130. The user's current network data may include one or more of any combination of the user's current network address, network signal strength, and network cache time.

The processing device (for example, the processing device 112 or other processing devices (for example, the first acquisition module 410)) can also filter the historical data of the positioning request in the user terminal 130, for example, it can filter out the positioning request acquisition with the closest time Request for targeting. This manual does not limit the acquisition method of the target positioning request.

Step 520: Acquire first information associated with the target positioning request. In some embodiments, step 520 may be performed by the second acquisition module 420.

The first information may be any information related to the target positioning request.

In some embodiments, the first information may include at least signal characteristics associated with the target positioning request.

In some embodiments, the signal characteristics may include, but are not limited to, WIFI signals, base station signals, GPS signals, Bluetooth signals, (NFC) radio frequency signals, and the like. The first information may also include first fingerprint information, and the first fingerprint information may include the identification and signal strength of the wireless access point AP scanned at the current location. Wherein, the identifier of the wireless access point AP may be the name or MAC address of the wireless network. The MAC address is written inside the hardware of the network device when it is produced by the network device manufacturer, and serves as the only network identifier of the network device.

In some embodiments, the first information may also include non-signal characteristics associated with the target positioning request. In some embodiments, non-signal characteristics may include, but are not limited to, weather, wind speed, temperature, humidity, indoor and outdoor, signal mode (for example, 4G or 5G), user's mobile phone model, etc. at the current location or target location.

In some scenarios, for the same target positioning request, its non-signal characteristics may affect (for example, enhance, weaken) the strength of its signal characteristics. For example, an environment with bad weather (e.g., heavy rain, thunder and lightning) will weaken the intensity of the corresponding signal feature (e.g., base station signal). For another example, in the process of responding to the same target positioning request, the user switches the signal mode of the user terminal 130 from 4G to 5G, which will enhance the strength of the corresponding signal feature (for example, GPS signal).

In some embodiments, the first information may further include one or more of the following associated with the target positioning request: popularity information, minimum signal strength, maximum signal strength, crowd density, center grid pixel value size, characteristics The total size of the values of all grids in the graph.

The popularity information may refer to the frequency of the user reaching the destination of the target positioning request within a predetermined area (for example, a city) and within a predetermined time period (for example, the last three months, the last six months, and the last year). The popularity information can be used to characterize the reachable state of the corresponding input grid and the number of arrivals within a predetermined time period. For more details about the heat information, please refer to the following embodiments and related descriptions, which will not be repeated here.

The maximum signal strength and the minimum signal strength are respectively used to indicate the upper limit and lower limit of the signal strength corresponding to the above-mentioned signal characteristics. People flow density is used to indicate the flow of people at the destination of the target positioning request.

The pixel value size of the center grid and the total size of the values of all grids in the feature map are used to represent the image information corresponding to the target positioning request. For more details about the grid, the pixel value size of the center grid, and the feature map, please refer to the following embodiments and related descriptions, which will not be repeated here.

In some embodiments, the first information may also include the communication relationship between the target positioning request and the main base station, the sum of the signal strength matching probabilities of multiple base stations adjacent to the main base station, the center distance from the former base station of the main base station, and At least one of the communication relationship of the previous base station and the sum of the heat information of the previous base station; wherein the main base station is the base station to which the terminal corresponding to the target positioning request is currently connected, and the previous base station is the base station that the terminal is connected to before connecting to the main base station Base station.

In some embodiments, the base station may be any device installed with lidar equipment, radar, and cameras. The base station may be a movable platform, for example, a vehicle (for example, a car, an airplane, a boat, etc.). The base station can also be a fixed platform, such as a detection station or an airport control tower.

It should be noted that the signal strength can be divided into S levels, for example, 7 levels of 0-6. In some embodiments, the matching probability between different signal intensities can be expressed as:

Where f _p,i is the matching probability, S is the highest level of signal strength, hi _,s is the number of times the i-th input grid receives the signal strength of s, and t is the signal strength in the positioning request, h _i,t is the number of times the i-th input grid received signal strength is t, hi _,t-1 is the number of times the i-th input grid received signal strength is t-1, hi _,t-1 is the number The number of times the received signal strength of i input grids is t+1, and w1, w2, and w3 are the corresponding weights.

Thus, by adding base station information, the accuracy of positioning can be further improved.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the second acquisition module 420)) may acquire different types of first information in a variety of ways.

Take acquiring the signal characteristic of the first information (for example, the MAC address of the wireless network) as an example. In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the second obtaining module 420)) may obtain the MAC address from the user terminal 130. In some embodiments, when the user terminal 130 searches for one or more wireless network hotspots around, the user terminal 130 may record and save the MAC address corresponding to the above-mentioned wireless network hotspot for the second obtaining module 420 to obtain.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the second acquisition module 420)) may also acquire the first information from a storage device (for example, the storage device 140) that stores the first information. Signal characteristics.

Take the non-characteristic signal (for example, the weather of the destination corresponding to the target positioning request) of acquiring the first information as an example. In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the second acquisition module 420)) may acquire the non-signal characteristics input by the user from the user terminal 130. Specifically, the user may receive the information presented by the processing device in the form of a pop-up window at the user terminal 130 (for example, asking about the weather conditions of the user’s current location), and the user may input the answer corresponding to the information in the user terminal 130 for the processing Equipment acquisition.

The first information can also be obtained in other ways. For example, a processing device (for example, the processing device 112 or another processing device (for example, the second acquisition module 420)) may obtain the historical data of the first information from the user terminal 130 and process the first information to obtain the first information. In some embodiments, the processing device may periodically update the historical data of the above-mentioned first information.

It should be noted that obtaining different first information will have varying degrees of impact on the confidence of target positioning. Therefore, comprehensively and accurately acquiring the first information through the foregoing acquisition method can improve the accuracy of target positioning. For more details about the confidence level, please refer to the following embodiments and related descriptions, which will not be repeated here.

Step 530: Determine a preset point associated with the target positioning request based on the first information. In some embodiments, step 530 may be performed by the first determining module 430.

The preset point may be any geographic coordinate point associated with the target positioning request.

The preset point may be a coordinate point within a preset range of the destination of the target positioning request. In some embodiments, the preset point may include: at least one of the center point of the grid where the target positioning request is located, the maximum signal strength point, and the position coordinate point.

In some embodiments, the preset point may include the center point of the grid where the target positioning request is located. In some embodiments, the center point may be the center coordinate point of the grid corresponding to the destination of the target positioning request. The aforementioned corresponding grid may be a grid obtained by locating the destination through a network. For more details about the grid, please refer to Figure 6 and its related descriptions, which will not be repeated here.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determining module 430)) may determine the preset associated with the target positioning request in a variety of ways based on the first information. point.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determining module 430)) may determine the preset point corresponding to the target positioning request according to the signal strength of the first information. Taking the acquired preset point as the maximum signal strength point as an example, in some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determining module 430)) may acquire the target from the user terminal 130 Multiple signal strengths for positioning requests. It can be understood that since the signal strength depends on the distance between the user terminal 130 and the WIFI device and/or the base station, the greater the distance, the lower the corresponding signal strength. Therefore, the determining module 430 can determine the distance between the user terminal 130 and the WIFI device and/or the base station according to the obtained multiple signal strengths, and based on this, determine the preset point associated with the target positioning request, for example, according to the aforementioned WIFI The position coordinates of the device and/or base station and the corresponding distance are calculated, and the position coordinates of the point of maximum signal strength within the range are calculated.

The preset point associated with the target positioning request may also be determined in other ways, which is not limited in this embodiment. For example, the preset point can be determined by the location coordinate point of the corresponding grid by the target positioning request. For more details about determining the preset point by the position coordinate point of the grid corresponding to the target positioning request, please refer to FIG. 6 and its related description, which will not be repeated here.

Step 540: Generate at least one first feature map based on the preset points. In some embodiments, step 540 may be performed by the generation module 440.

The feature map may refer to an image containing feature information related to a target positioning request. In some embodiments, the characteristic information may include at least one of traffic speed, traffic flow and traffic density, heat information, geographic information, minimum signal strength, and maximum signal strength of the destination of the target positioning request. Taking feature information as heat information as an example, heat information can be used to characterize whether the grid corresponding to the target positioning request is in a place reachable by vehicles. For example, if there is an intersection in the grid, the grid is reachable. If the grid is a green area, such as a lawn, the grid is in an unreachable state. The predetermined time period may be one day, one week, one month, etc., which is not limited in this embodiment. The geographic information of the grid may represent the density of buildings and/or greening density input into the grid. It should be understood that other characteristic information may also be included, such as crowd density, etc., which is not limited in this embodiment. Optionally, the number of feature information corresponds to the number of feature maps.

In some embodiments, the characteristic information may also be signal characteristics or non-signal characteristics included in the first information, for example, signal strength, weather conditions, and so on.

In some embodiments, the first feature map may be generated in a variety of ways based on preset points. In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the generating module 440)) may acquire a preset range of images centered on a preset point (for example, a center point, a maximum signal strength point) , Process the acquired image to generate the corresponding first feature map. Specifically, the above-mentioned image may be captured by an image acquisition device (for example, a camera, a vehicle-mounted device), and the above-mentioned image is combined with feature information for manual annotation to generate a first feature map. For example, the corresponding heat information, crowd density, etc. can be marked in the image.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the generating module 440)) may determine multiple input grids corresponding to a preset point (for example, a center point, a maximum signal strength point), And multiple feature maps are determined by the feature information corresponding to the multiple input grids. For more details about inputting the grid and determining the feature map through the inputting grid, please refer to FIG. 11, FIG. 12, FIG. 15 and related descriptions, which will not be repeated here.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the generation module 440)) may also generate the first feature map through a machine learning model or algorithm. Specifically, the above-mentioned acquired image may be input to a machine learning model, or the above-mentioned acquired image may be processed through a machine learning algorithm to generate a first feature map. Machine learning models or algorithms can include, but are not limited to, gradient boosting decision tree (GBDT) model, decision tree algorithm, random forest algorithm, logistic regression algorithm, support vector machine (SVM) algorithm, naive Bayes algorithm, adaptive enhancement algorithm, K nearest neighbor (KNN) algorithm, Markov chain algorithm, etc., or any combination thereof.

The first feature map can also be generated in other ways. For example, it can be obtained from a storage device (for example, the storage device 140), a map service provider (for example, Google Maps ^™ ), and/or any other device and/or service provider that can provide a feature map related to the target positioning request The first feature map.

Step 550: Determine at least one second characteristic map based on the at least one first characteristic map. In some embodiments, step 550 may be performed by the second determining module 450.

The second feature map may be an image obtained after processing the first feature map. In some embodiments, the second feature map can be used to determine the target location. For more details about determining the target location through the second feature map, please refer to step 560, FIG. 16 and related descriptions, which will not be repeated here.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the second determining module 450)) may determine the second characteristic map in a variety of ways based on the first characteristic map.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the second determining module 450)) may obtain multiple first feature maps, and perform fusion processing on the multiple first feature maps to obtain at least A second feature map. Just as an example, the processing device can use the acquisition method in step 540 to acquire two first feature maps with different feature information (for example, signal strength and heat information) respectively, and the processing device can analyze the above two different features. The information is fused to generate at least one second feature map. It can be understood that if multiple second feature maps are generated, each second feature map has the above two feature information.

In some embodiments, the above-mentioned method for fusing different feature information may include: directly splicing different feature information, or generating a combined feature from multiple feature information through set algorithm processing. In some embodiments, the method of generating combined features may include, but is not limited to: feature combination, feature dimensionality reduction, feature intersection, and the like. Feature combination means to obtain a new feature through some linear or non-linear superposition of features. Common feature combination methods can be Cartesian product methods. For example, feature A={0,1}, feature B={0,1}, then feature A×B={(0,0),(0,1),(1,0),(1,1) }. The feature combination method can also adopt the decision tree + LR method. In some embodiments, a gradient boosting tree + logistic regression (GBDT+LR) approach can be used. Gradient boosting tree + logistic regression (GBDT+LR) is an automatic feature extraction method. GBDT is a gradient boosting decision tree. First, a decision tree will be constructed. First, a decision tree will be constructed on the residuals of the existing model and the actual sample output, and iteratively, each iteration will produce a larger gain classification Therefore, as many leaf nodes as there are in the decision tree constructed by GBDT, the resulting feature space will be as large, and this feature will be used as the input of the LR model. The decision tree is represented in the form of a tree of nodes. Each node makes a binary decision based on the characteristics of the data, and each leaf node of the tree contains a prediction result.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the second determining module 450)) may group the obtained multiple first characteristic maps according to the category of the characteristic information. Specifically, the first feature maps with feature information A in the above-mentioned multiple first feature maps can be grouped into one group, and the first feature maps with feature information B can be divided into another group. The foregoing groups may be determined as different second feature map sets, and the processing device may merge the foregoing second feature map sets to determine at least one second feature map.

The second feature map can also be determined in other ways, for example, the first feature map can be directly determined as the second feature map, etc., which is not limited in this specification.

Step 560: Process the at least one second feature map to obtain target positioning. In some embodiments, step 560 may be performed by the third acquisition module 460.

The target positioning may be the location information of the destination corresponding to the target positioning request, the environmental information of the geographic area to which the destination belongs, and the like. Wherein, the location information may be coordinate points, latitude and longitude, and so on. The environmental information can be weather, wind speed, etc.

In some embodiments, the target location may be obtained by means of network positioning technology or positioning model processing. In some embodiments, the positioning model may be a convolutional neural network model.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the third acquisition module 460)) may process at least one second feature map based on the convolution kernel to acquire the target location.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the third acquisition module 460)) may process the at least one second feature image in a variety of ways based on the convolution kernel.

In some embodiments, the processing method for the at least one second feature map may be: based on the at least one second feature map, the position correction information is determined through a convolutional neural network model. In some embodiments, the position correction information may include a position offset. The position offset may include the longitude offset and the latitude offset of the actual position relative to the preset point position. Specifically, at least one second feature map may be input to the convolutional neural network model for feature processing, and the position offset is output.

In some embodiments, the convolutional neural network model may include a convolution kernel for processing the second feature map. The convolutional neural network may be a feedforward neural network, and the convolution kernel may be composed of several convolutional layers and pooling layers. The structure of the convolutional neural network model may also be U-Net, ResNet, DenseNet, etc., and this specification does not limit the structure of the convolutional neural network model.

For more details about the structure of the convolutional neural network model, please refer to FIG. 17 and related descriptions, which will not be repeated here.

In some embodiments, the input of the convolutional neural network model may be one or more second feature maps, and the output may be position correction information. In some embodiments, the input of the convolutional neural network model may also be the actual location of the target location, the grid where the actual location is located, or the location of the actual location in the second feature map.

For more details about the input and output of the convolutional neural network model, please refer to FIG. 15, FIG. 16, FIG. 17, and related descriptions, which will not be repeated here.

Because the convolutional neural network model has the characteristics of local area connection, weight sharing, downsampling, etc., the weight sharing reduces the number of weights that need to be trained, reduces the computational complexity of the network, and makes the network locally transform the input It has certain invariance, such as translation invariance, zoom invariance, etc., which improves the generalization ability of the network. At the same time, the convolutional neural network model has the characteristics of feature cross extraction and fusion of peripheral information. Therefore, the present embodiment adopts the convolutional neural network model, which does not need to do a lot of complicated feature work, and can expand the feature information more conveniently and reduce the input The model's feature information is lost, and it is more flexible to add features.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the third acquisition module 460)) may acquire the target location based on the above-mentioned position correction information and preset points. Taking the position correction information as the position offset as an example, the processing device can adjust the coordinates of the preset point to obtain the target positioning according to the longitude and latitude coordinates corresponding to the acquired position offset. For example, the processing device can perform a summation calculation on the coordinates of the preset point and the latitude and longitude coordinates corresponding to the position offset to obtain the destination coordinates of the target positioning.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the third determining module 470)) may determine the confidence of the target positioning in a variety of ways.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the third determining module 470)) may determine the confidence of the target location based at least on the confidence network model and the at least one second feature map.

In some embodiments, the confidence network model may be a machine learning model, which may include, but is not limited to, a linear classifier (LC), a K-Nearest Neighbor (kNN) model, and a naive Baye Naive Bayes (NB) model, Support Vector Machine (SVM), Decision Tree (DT) model, Random Forests (RF) model, Classification and Regression Trees, CART) model, Gradient Boosting Decision Tree (GBDT) model, xgboost (eXtreme Gradient Boosting), Gradient Boosting Machines (GBM), Light Gradient Boosting Machine (LightGBM), LASSO (Least Absolute Shrinkage and Selection Operator, LASSO), Artificial Neural Networks (Artificial Neural Networks, ANN) models, etc., or any combination thereof.

In some embodiments, the input of the confidence model may be one or more second feature maps, and the output may be the confidence of the target position. The confidence level may indicate the accuracy of the corresponding positioning information. The confidence level can be any number between [0, 1]. 0 means completely inaccurate, 1 means completely accurate, the larger the value, the higher the accuracy.

Taking the confidence network model as the GBDT model as an example, in some embodiments, the proportion of non-empty feature pixels in the feature map can be calculated, and the proportion of non-empty feature pixels can be input into the confidence network model for feature processing, and output Error distance, and then obtain the confidence of positioning information according to the mapping formula. Optionally, the confidence network model of this embodiment is a gradient boosting decision tree network model GBDT.

In some embodiments, the mapping formula (2) of the GBDT model is:

Among them, confidence is the degree of confidence, which is used to characterize the credibility of the positioning information obtained based on the output of the convolutional neural network model. Optionally, this embodiment obtains multiple feature maps based on the training data, trains the GBDT model according to the proportion of non-empty feature pixels in the feature maps, regresses the error distance between the predicted position and the satellite positioning position of the sample, and according to the error distance And the above mapping formula to obtain confidence. Therefore, in this embodiment, while predicting the position, the confidence level of the obtained positioning information can be evaluated as a basis for other services.

Fig. 6 is a flowchart of an exemplary process of determining a preset point according to some embodiments of the present specification. In some embodiments, process 600 may be performed by a processing device (eg, processing device 112 or other processing device). For example, the process 600 may be stored in a storage device (for example, the storage device 140 or a storage unit of a processing device) in the form of a program or instruction. When the processor 220 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 600. In some embodiments, the process 600 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 6 is not restrictive.

Step 610: Obtain second information including multiple preset grids.

In some embodiments, the preset grid may be a geographical area pre-divided by network positioning. In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determining module 430)) may pre-divide a large area (for example, the area of a city) into multiple parts by means of network positioning or the like. Grids, optionally, the size of each grid is N*N, and N is greater than or equal to 1 meter.

In some embodiments, the second information may be information associated with a preset grid. In some embodiments, the second information may include signal characteristics and non-signal characteristics of the preset network. For example, multiple signal strengths of the preset grid, weather conditions at the corresponding location of the preset grid, and so on.

In some embodiments, the second information may also include position information of a preset grid.

In some embodiments, the preset grid may have corresponding signal characteristics, non-signal characteristics, and location information.

In some embodiments, the second information may include second fingerprint information. For example, the second fingerprint information may include the identification and signal strength of the wireless access point AP scanned in the preset grid. Optionally, the identifier of the wireless access point AP may be the name or MAC address of the wireless network.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determination module 430)) may obtain the second information in a variety of ways. In some embodiments, the processing device may obtain one or more second information corresponding to each preset grid from the storage device.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determining module 430)) may obtain all historical positioning requests from the storage device, and establish and preset grids based on the above positioning requests. The corresponding second information database. In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determination module 430)) may establish the second information database based on the pre-scan. The processing device (for example, the processing device 112 or other processing devices (for example, the first determining module 430)) may obtain the second information corresponding to each preset grid from the second information database.

In some embodiments, the second information database may be updated in real time. For example, the processing device can update AP-related information in real time, such as changes in the name of the AP, new APs, or failures of the original APs.

In some embodiments, the processing device may obtain the second fingerprint information of different grids from the fingerprint database. Optionally, when each target user requests positioning in the grid through the user terminal 130, the second fingerprint information corresponding to the grid is determined and uploaded to the fingerprint database, or the terminal is used to scan in multiple grids in advance. Obtain the correspondence between the location information of each grid and the fingerprint information scanned in the grid, and upload the correspondence between the location information of each grid and the fingerprint information scanned in the grid to the fingerprint database in. Optionally, the second fingerprint information of each grid is updated in real time. For example, changes in the name of the AP, new APs, or failure of the original AP can cause the second fingerprint information of the grid to be changed. Therefore, The accuracy of positioning can be improved by updating the second fingerprint information of each grid in real time.

In addition to the above methods, the second information can also be obtained in other ways. For example, the second information may be obtained in the manner of obtaining the first information in the foregoing step 520. For example, the second information may be obtained from a storage device (for example, the storage device 140) that stores the second information.

Step 620: Calculate the similarity between the first information and the second information.

In some embodiments, the processing device may calculate the similarity between the first information and the second information to indicate the probability that the location of the target positioning request is located in the preset grid. For example, the higher the similarity, the higher the probability that the current location request is issued in the preset grid corresponding to the second information.

In some embodiments, the processing device may determine the similarity between the first information and the second information through the similarity model. The similarity model can determine the similarity between the vector corresponding to the first information and the vector corresponding to the second information through the similarity algorithm. Exemplary similarity algorithms may include Cosine Similarity, Euclidean Distance Algorithm, Pearson Correlation Coefficient Algorithm, Tanimoto Coefficient Algorithm, Manhattan Distance Algorithm, Ma Mahalanobis distance algorithm, Lance Williams distance algorithm, Chebyshev distance algorithm, Hausdorff distance algorithm, etc.

In some embodiments, the AP identification and signal strength in the first information may be directly compared with the AP identification and signal strength in the second information respectively for similarity and difference to determine the similarity.

In some embodiments, the first information and the second information may be subjected to discrete processing and/or normalization processing to form the corresponding first vector and the second vector. For example, the first information and the second information may be processed by a softmax function. The information is normalized to obtain a first vector corresponding to the first information and a second vector corresponding to the second information. Then calculate the cosine similarity (or Euclidean distance, etc.) between the first vector and the second vector to determine the similarity.

It is easy to understand that the higher the similarity between the first information and the second information, the higher the probability that the current position of the user terminal 130 is in the grid corresponding to the second information.

In addition to the above methods, the second information can also be obtained in other ways. For example, the second information can be obtained in the manner of obtaining the first information in step 520 described above. For example, the second information may be obtained from a storage device (for example, the storage device 140) that stores the second information.

Step 630: Determine at least one similar grid based on the similarity.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determining module 430)) may be based on the similarity by judging whether the above-mentioned similarity satisfies a preset condition, or by performing an evaluation on the above-mentioned similarity. At least one similar grid is determined by means such as sorting.

In some embodiments, the processing device may set a certain preset condition for the similarity between the first information and the second information, and determine the preset grid that satisfies the condition as a similar grid. Among them, in some embodiments, the preset condition may be whether the similarity between the first information and the second information is greater than a preset threshold. For example, the preset threshold may be 0.8. If the similarity is greater than 0.8, it means that the preset grid meets the preset condition. The preset threshold may be set according to requirements, for example, different thresholds may be set for preset grids corresponding to different cities and service areas.

In some embodiments, the processing device may also sort a plurality of preset grids based on the similarity, and determine at least one similar grid based on the similarity ranking result. For more details about similarity ranking and determining similar grids, please refer to FIG. 7 and related descriptions, which will not be repeated here.

In addition to the above methods, similar grids can also be determined by other methods such as comparing the weights of multiple preset grids.

Step 640: Determine the preset point based on the at least one similar grid.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determination module 430)) may be based on the median, average, or geometric center of the position coordinates of at least one similar grid. One or more of them determine the preset point.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determination module 430)) determines the preset point based on the average value of the position coordinates of the at least one similar grid. Specifically, the processing device may use the average of the longitude and latitude of at least one similar grid to characterize the position coordinates of the preset point, and use the position coordinate of the center of the similar grid to characterize the position coordinates of the similar grid. In some embodiments, the position coordinates of the preset point satisfy the following formula:

Among them, center _{_lon} is the longitude of the preset point, center _{_lat} is the latitude of the preset point, K is the number of similar grids, g _{k_lon} is the longitude of the center of the Kth similar grid, and g _{k_lat} is the Kth similar The latitude of the center of the grid. It can be understood that by summing the position coordinates of the K similar grids and then performing an average calculation, the average value of the position coordinates of the K similar grids can be obtained.

This embodiment uses the coordinates of the center of the similar grid to characterize the position coordinates of the similar grid. It should be understood that the coordinates of other points in the grid can also be used as the position coordinates of the grid, and this embodiment does not limit this. . It should be understood that other methods for determining the center point according to the positions of similar grids can be applied in this embodiment. For example, weights are assigned to each grid based on the corresponding similarity, and the weighted average of the position coordinates of each similar grid is calculated. To obtain the position coordinates of the center point, etc., this embodiment does not limit this.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determination module 430)) may determine the preset point based on the geometric center of the at least one similar grid. Taking a similar grid as an example, the geometric center can be the center point of the similar grid. Taking multiple similar grids as an example, the geometric center may be the coordinate point with the smallest sum of distances from the similar grids.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determination module 430)) may determine the weights of a plurality of preset grids based on the similarity, and determine the preset grids based on the weights and the similarity. Set point. For example, the processing device may assign higher weights to preset grids with higher similarity, and sort the preset grids according to the weights corresponding to the multiple preset grids (for example, in ascending order), and sort the preset grids with the highest weight. The center point of is determined as the preset point.

In some embodiments, the preset point may also be determined based on signal characteristics. Specifically, the processing device (for example, the processing device 112 or other processing devices (for example, the first determining module 430)) may obtain the signal strengths corresponding to the multiple signal characteristics in the target positioning request, and locate the respective grids based on the above signal strengths. Or range, and the intersection position of the grid or range or the midpoint of the intersection range is determined as the preset point. This manual does not limit the way to determine the preset point.

Fig. 7 is a flowchart of an exemplary process of determining a preset point according to some embodiments of the present specification. In some embodiments, process 700 may be performed by a processing device (e.g., processing device 112 or other processing device). For example, the process 700 may be stored in a storage device (for example, the storage device 140 or a storage unit of a processing device) in the form of a program or instruction. When the processor 220 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 700. In some embodiments, the process 700 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 7 is not restrictive.

Step 710: Sort the multiple preset grids based on the similarity to obtain a similarity sorting result.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determining module 430)) can perform various methods such as ascending and descending order on the similarity of multiple preset grids. Sort multiple preset networks to obtain similarity ranking results.

In some embodiments, the processing device may sort a plurality of preset networks based on the magnitude of the similarity value corresponding to each similarity grid to obtain the similarity ranking result. Among them, the similarity ranking results can be sorted according to the similarity value from large to small, and can also be sorted according to the similarity value from small to large.

In addition to the above methods, the similarity ranking results can also be obtained by other methods, which are not limited in this specification.

In step 720, the processing device (for example, the processing device 112 or another processing device (for example, the first determining module 430)) determines at least one similar grid based on the similarity ranking result.

In some embodiments, the processing device (for example, the processing device 112 or other processing devices (for example, the first determination module 430)) may determine at least one preset grid that meets the preset conditions as similar based on the similarity ranking result. grid. The preset condition may be that the similarity is greater than the threshold, the top ones in the order of similarity, and so on. For example, according to the similarity ranking result, the three preset grids with the highest similarity ranking are determined as similar grids. For another example, the preset grid with the highest degree of similarity may be determined as a similar grid.

Similar grids can also be determined by the weight ranking results of the preset grids, which is not limited in this specification.

Fig. 8 is a schematic diagram of a machine learning model of the related technology. As shown in Figure 8, in the machine learning model based on the NLP service of the related technology, the current fingerprint information is obtained while the terminal scans the GPS location. The fingerprint information may include the unique identification of the wireless access point AP scanned by the terminal (SSID and/or MAC address), the scanned signal strength, the unique identification of the base station, etc., and upload the GPS location and the acquired fingerprint information through the network 81 to form a fingerprint database 82. In addition, the correspondence between the GPS location and fingerprint information can be constructed based on the fingerprint database 82. Specifically, the fingerprint information is used as the input, and the GPS location is the output to construct a sample to train a machine learning model, such as the machine learning model 83 shown in FIG. 8. In subsequent use, the fingerprint information uploaded on the terminal side is input into the machine learning model 83 to predict the current position.

As shown in FIG. 8, in the related art, the machine learning model 83 includes a recall module 831, a ranking module 832, and a smoothing module 833. Wherein, a region (for example, an urban area) is pre-divided into multiple geographic regions to form multiple grids, each grid size is N*N meters, and N is greater than or equal to 1. Among them, each grid can record rich fingerprint information. When using the machine learning model 83 to predict the current position, input the fingerprint information of all grids and the fingerprint information uploaded by the current terminal into the machine learning model 83, and select a predetermined number from all grids through some manual rules in the recall module 831 The candidate grids of, are input to the sorting module 832, and the sorting module 832 sorts the candidate grids according to predetermined rules, so that the grids corresponding to the true value (accurate position) are sorted first, thus, the positioning problem can be transformed into a sorting problem . In the processing of the sorting module 832, the deviation of fingerprint information and other data may cause the sorting result to deviate far from the true value. Therefore, the smoothing module 833 is designed to correct the sorting result.

Among them, the machine learning model used in related technologies (for example, machine learning model 83) is mainly a tree-based model. Feature engineering is required when constructing the model, that is, the scanned fingerprint information is transformed into a suitable machine by artificial means. Learning the advanced features of the model, this process will lead to the loss of features, which will lead to lower accuracy of model prediction. Specifically, in the related art, in the feature processing process of the sorting module 832, there is a lack of characterization of the spatial relationship between grids, for example, the lack of local correlation and overall correlation of features between grids, and, By adding a smoothing module to correct the predicted position, it does not optimize the positioning accuracy from the feature level, but smooths the sorting result of the sorting module. Therefore, the deviation between the predicted result and the true value may still exist. Therefore, In related technologies, the accuracy and precision of using a machine learning model to predict the current position is low.

Therefore, the embodiment of this specification provides a network positioning method, according to a plurality of preset points corresponding to the first information in a predetermined target positioning request, a plurality of preset grids are determined according to the foregoing multiple preset points, and a plurality of preset grids are determined according to the preset points. It is assumed that one or more feature maps are determined corresponding to the grid, and the one or more feature maps are input to a pre-trained convolutional neural network model to obtain target positioning. Therefore, the embodiment of the present specification can reduce the feature loss corresponding to the feature map, and improve the accuracy of obtaining target positioning.

Fig. 9 is a flowchart of an exemplary process of a network positioning method according to some embodiments of the present specification. In some embodiments, the process 900 may be executed by the processing device 112 or other processing devices. For example, the process 900 may be stored in a storage device (for example, the storage device 140 or a storage unit of a processing device) in the form of a program or instruction. When the processor 220 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 900. In some embodiments, the process 900 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 9 is not restrictive.

In some embodiments, the network positioning method may include the following steps:

Step S910: Receive a positioning request. The target positioning request may be referred to as a positioning request, and the positioning request includes the first fingerprint information corresponding to the current position. As mentioned above, the first information may include first fingerprint information. The first fingerprint information may include the identification and signal strength of the wireless access point AP scanned at the current location. Optionally, the identifier of the wireless access point AP may be the name or MAC address of the wireless network.

Step S920: Determine the center point corresponding to the first fingerprint information according to the predetermined second fingerprint information of different grids. As mentioned above, the second information may include the second fingerprint information, the preset point may include a center point, and the grid is a pre-divided geographic area. In this embodiment, a large area (for example, an area of a city) is divided into multiple grids in advance. Optionally, the size of each grid is N*N, and N is greater than or equal to 1 meter.

In some embodiments, the second fingerprint information of different grids can be obtained from the fingerprint database. Optionally, when each user terminal 130 requests positioning in the grid, the second fingerprint information corresponding to the grid is determined and uploaded to the fingerprint database, or the terminal is used to scan multiple grids in advance to obtain each grid. The corresponding relationship between the position information of the grid and the fingerprint information scanned in the grid, and the corresponding relationship between the position information of each grid and the fingerprint information scanned in the grid is uploaded to the fingerprint database. Optionally, the second fingerprint information of each grid is updated in real time. For example, changes in the name of the AP, new APs, or failure of the original AP can cause the second fingerprint information of the grid to be changed. Therefore, The accuracy of positioning can be improved by updating the second fingerprint information of each grid in real time.

Step S930: Determine multiple input grids according to the center point. In some embodiments, with the center point as the center, a grid corresponding to the center point can be expanded to M*M grids as input grids, and M is greater than 1. Among them, similar grids can be called input grids. For more details about determining the input grid, please refer to FIG. 7, FIG. 11, FIG. 12 and related descriptions, which will not be repeated here.

Step S940: Determine multiple feature maps according to feature information corresponding to each input grid. In this embodiment, the value of each pixel of the feature map corresponds to the feature value of the feature information of each input grid, and at least one type of feature information is related to the first fingerprint information. Therefore, even if the calculated center points are the same, because the first fingerprint information in the positioning request is different, the characteristic information related to the first fingerprint information is also different, and the finally obtained positioning information is also different. Optionally, the characteristic value of the characteristic information may be a numerical value, a vector or other expression forms, which is not limited in this embodiment.

In some embodiments, the characteristic information may include the matching probability between the signal strength received by each input grid and the signal strength in the positioning request, and the characteristic information is related to the first fingerprint information in the positioning request. As a result, by introducing certain features in the positioning request into the convolutional neural network for processing, the loss of feature information is further reduced, and the accuracy of positioning is improved.

For more details about the matching rate between the signal strengths, please refer to FIG. 5 and related descriptions, which will not be repeated here.

Step S950, input multiple feature maps to a pre-trained convolutional neural network model to obtain positioning information. Among them, target positioning can be called positioning information.

In an optional implementation manner, a plurality of feature maps are input to a convolutional neural network model for feature processing, a position offset is output, and the positioning information is obtained according to the position offset and the center point position. Wherein, the position offset is the offset of the positioning information relative to the position of the center point.

Convolutional Neural Network CNN is a feedforward neural network composed of several convolutional layers and pooling layers. The CNN model has the characteristics of local area connection, weight sharing, downsampling, etc. Through weight sharing, the number of weights that need to be trained is reduced, the computational complexity of the network is reduced, and the network has a certain degree of difference in the local transformation of the input. Degeneration, such as translation invariance, scaling invariance, etc., improves the generalization ability of the network. At the same time, the CNN model has the features of feature cross extraction and fusion of surrounding information. Therefore, the CNN model is adopted in this embodiment without a lot of complicated feature work, and feature information can be expanded more conveniently and the loss of feature information of the input model is reduced. , And it is more flexible to add features.

In this embodiment, the CNN model calculates the position offset of the current position corresponding to the positioning request. Optionally, the position offset may include the longitude offset and the latitude offset relative to the center point position, based on the position offset and the center point. The position information of the current position can be calculated. As a result, the positioning problem can be transformed into a regression problem, which improves the accuracy of positioning.

The CNN model can complete classification tasks and regression tasks. If the localization problem in this embodiment is transformed into a classification problem, the CNN model can only determine whether each grid is or is not the grid where the truth point is located. In this case, if the obtained M*M grids do not cover the truth value points, a situation where the positioning cannot be performed will occur. Therefore, in this embodiment, the CNN model is used to convert the positioning problem into a regression problem, and the position offset can always be output, which improves the accuracy of positioning.

In an optional implementation manner, the convolutional neural network model of this embodiment is trained according to a predetermined loss function. The loss function can directly affect the performance of the convolutional neural network model. Optionally, the loss function used in this embodiment is:

Wherein, Δlon_pred and Δlat_pred are the offsets in longitude and latitude between the predicted position of the convolutional neural network model and the center point, and Δlon_label and Δlat_label are the offsets in longitude and latitude between the satellite positioning position of the sample and the center point .

Therefore, the convolutional neural network model is trained based on the above loss function until the loss function is minimized, that is, the error distance is minimized, and a trained convolutional neural network model is obtained.

Fig. 10 is a flowchart of an exemplary process of determining a center point according to some embodiments of the present specification. In some embodiments, the process 1000 may be executed by the processing device 112 or other processing devices. For example, the process 1000 may be stored in a storage device (for example, the storage device 140 or a storage unit of a processing device) in the form of a program or instruction. When the processor 220 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 1000. In some embodiments, the process 1000 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 10 is not restrictive.

As shown in FIG. 10, step S920 may include the following steps.

Step S921: Calculate the similarity between the first fingerprint information and each second fingerprint information. In some embodiments, the AP identification and signal strength in the first fingerprint information may be directly compared with the AP identification and signal strength in the second fingerprint information, respectively, for similarity and difference to determine the similarity. In other embodiments, the first fingerprint information and the second fingerprint information may be subjected to discrete processing and/or normalization processing to form the corresponding first vector and the second vector, and the cosine of the first vector and the second vector can be calculated. Similarity (or Euclidean distance, etc.) to determine the similarity. It is easy to understand that the higher the similarity between the first fingerprint information and the second fingerprint information, the higher the probability that the current position of the user terminal 130 is in the grid corresponding to the second fingerprint information.

In step S922, the grids are sorted according to the degree of similarity. Optionally, the grids can be sorted according to the corresponding similarity degrees from large to small, and the grids can also be sorted according to the corresponding similarities from small to large, which is not limited in this embodiment.

Step S923: Obtain at least one similar grid according to the similarity ranking result. Optionally, the first predetermined grid with the highest similarity is obtained as the similar grid.

Step S924: Determine the center point according to the position coordinates of the at least one similar grid. In an optional implementation manner, the median or average value of the position coordinates of each similar grid is calculated, and the median or average value is determined as the position coordinates of the center point. For more details about determining the preset point through similar grids, please refer to step 640 and its related description, which will not be repeated here.

FIG. 11 is a flowchart of an exemplary process of determining an input grid according to some embodiments of the present specification. As shown in Fig. 11, according to the aforementioned step S930, when M is an odd number, taking the grid where the center point C is located as the central grid Cg, the central grid Cg can be expanded to M*M grids as the input grid Pin . Among them, M=5 is taken as an example in FIG. 11. For more details on determining the input grid, please refer to FIG. 12 and its related descriptions, which will not be repeated here.

Fig. 12 is a flowchart of another exemplary process of determining an input grid according to some embodiments of the present specification. As shown in Figure 12, according to the aforementioned step S930, when M is an even number, the center point is taken as the relative center, so that there are more input grids on one side of the boundary close to the center point than on the other side. The number of grids expanded in the opposite direction of the x-axis and the opposite direction of the y-axis is 3, and the number of grids expanded in the positive direction of the x-axis and the positive direction of the y-axis is 2, to form 6*6 input grid Pin' . It should be understood that other expansion methods can also be applied in this embodiment, for example, random expansion makes one side of the grid where the center point is located more rows or one column of grids than the other side to form M*M grids.

Fig. 13 is a schematic diagram of a characteristic diagram according to some embodiments of the present specification. As shown in FIG. 13, the feature information in the feature map 13 is used to characterize the popularity information of each input grid. That is, in this embodiment, each input grid corresponds to a pixel in the characteristic map 13, and the heat information corresponding to the input grid is the pixel value of the pixel. Among them, a pixel value of 0 indicates that the number of times the vehicle reaches the grid within a predetermined period of time is 0, and a pixel value of 24 indicates that the number of times the vehicle reaches the grid within a predetermined period of time is 24. Optionally, the input information of each input grid can be obtained from a fingerprint database. Thus, by taking the feature value representing a type of feature information as the value of the pixel corresponding to each input grid, a feature map corresponding to the type of feature information of each input grid is obtained. It is easy to understand that the characteristic value of the characteristic information may be a numerical value, a vector, a character, or other manifestations, which is not limited in this embodiment.

FIG. 14 is a flowchart of an exemplary process of training a convolutional neural network model according to some embodiments of this specification. In some embodiments, the process 1400 may be executed by the processing device 112 or other processing devices. For example, the process 1400 may be stored in a storage device (for example, the storage device 140 or a storage unit of a processing device) in the form of a program or instruction. When the processor 220 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 1400. In some embodiments, the process 1400 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 14 is not restrictive.

As shown in FIG. 14, in an optional implementation manner, the training method of the convolutional neural network model of this embodiment includes the following steps:

Step S1: Obtain training data. The training data may include multiple sample data, and the sample data may include the satellite positioning position of the sample, the first fingerprint information corresponding to the grid where the satellite positioning position of the sample is located, and the predetermined second fingerprint information of each grid.

Step S2: Determine the center point corresponding to each first fingerprint information according to the second fingerprint information of each grid.

Step S3: Determine multiple input grids according to the center point corresponding to each first fingerprint information.

Step S4: Determine multiple feature maps according to feature information corresponding to each input grid.

Step S5: Input multiple feature maps into the convolutional neural network model to predict the location information corresponding to each first fingerprint information.

It should be understood that steps S2-S5 are similar to steps S920-S950 in the foregoing embodiment, and will not be repeated here.

Step S6: Based on the above loss function, the loss is calculated according to the predicted positioning information and the satellite positioning position of the corresponding sample.

Step S7: Adjust the parameters of the convolutional neural network model according to the loss until the loss function is minimized.

The embodiment of this specification determines the center point by calculating the similarity between the first fingerprint information and each second fingerprint information of the grid where the satellite positioning position of the sample is located, and determines multiple features based on multiple input grids determined by the center point Figure, the feature map corresponding to each sample data is used as input, and the convolutional neural network model is trained based on the above loss function, so that the trained convolutional neural network model can more accurately predict the position based on the input feature map information.

FIG. 15 is a flowchart of an exemplary process of training a convolutional neural network model according to some embodiments of this specification. In some embodiments, the process 1500 may be executed by the processing device 112 or other processing devices. For example, the process 1500 may be stored in a storage device (for example, the storage device 140 or a storage unit of a processing device) in the form of a program or instruction. When the processor 220 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 1500. In some embodiments, the process 1500 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 15 is not restrictive.

As shown in FIG. 15, it is assumed that the user uses the user terminal 130 at the location D to send a target positioning request. As mentioned above, the target location request may include first information, the first information may include first fingerprint information, and the first fingerprint information may include at least the identification of the wireless access point AP (such as name, MAC address) scanned by the user terminal. Etc.), calculate the similarity between the second fingerprint information of each grid and the first fingerprint information, and sort the grids based on the similarity, determine multiple similar grids according to the similarity ranking results, and according to the multiple similar grids The average or median of the position coordinates of the grid determines the center point C1, and takes the grid where the center point C1 is located as the center grid, and expands outward to obtain M*M input grid Pin1. According to the feature information corresponding to each input grid, such as the matching probability of each input grid and the first fingerprint information, the maximum signal strength and minimum signal strength in each input grid, the popularity information of each grid, geographic information, etc., determine Multiple feature maps. Wherein, the feature value of the feature information corresponding to each input grid is filled into each pixel of the corresponding feature map to obtain the feature map corresponding to the feature information. Thus, assuming that each input grid has c feature information, then c feature maps with a size of M*M can be obtained.

In this embodiment, at least one of the feature information of the input grid is related to the first fingerprint information in the target location request. Therefore, even if the calculated center points are the same, because the first fingerprint information in the target positioning request is different, the characteristic information related to the first fingerprint information is also different, and the finally obtained positioning information is also different. At the same time, some features in the target positioning request are introduced into the convolutional neural network for processing, which further reduces the loss of feature information and improves the accuracy of positioning.

FIG. 16 is a schematic diagram of an exemplary process of determining target location and confidence based on a convolutional neural network model according to some embodiments of the present specification. In some embodiments, the process 1600 may be executed by the processing device 112 or other processing devices. For example, the process 1600 may be stored in a storage device (for example, the storage device 140 or a storage unit of a processing device) in the form of a program or instruction. When the processor 220 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 1600. In some embodiments, the process 1600 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 16 is not restrictive.

As shown in Fig. 16, in this embodiment, the positioning model may include a pre-trained CNN model and a GBDT model. Among them, the CNN model is used to predict the current location information corresponding to the positioning request, and the GBDT model is used to obtain the confidence of the current location information obtained according to the CNN model.

In some embodiments, the obtained c feature maps of size M*M are input into the pre-trained CNN model, and the longitude offset Δlon and latitude offset Δlat relative to the center point are output, according to the position coordinates of the center point C1 Calculate the location information of the current position L with the longitude offset Δlon and latitude offset Δlat output by the CNN model.

By calculating the proportion of non-empty feature pixels corresponding to each feature map, the proportion of non-empty feature pixels of each feature map is input as a feature (feature 161) into the GBDT model for feature processing to obtain the error distance, and the corresponding mapping formula is obtained according to the above mapping formula The confidence of confidence. Optionally, the proportion of non-empty feature pixels may be the ratio of the number of pixels with a pixel value other than 0 in the feature map to the total number of pixels. For example, in the feature map corresponding to the feature information shown in FIG. 13, the feature map The size of is 12*12, and the number of pixels whose pixel value is not zero is 62, and the proportion of non-empty feature pixels corresponding to feature map 13 is 62/144.

In this embodiment, the center point is determined by the first fingerprint information in the positioning request and the second fingerprint information of each grid, and the input grid is determined according to the center point, and the feature map corresponding to the positioning request is determined according to the feature information of the input grid. Input multiple feature maps into the pre-trained CNN model, output the position offset relative to the center point position coordinates, determine the current positioning information according to the position offset, and use the proportion of non-empty feature pixels in each feature map as the feature Input into the confidence network model to obtain the confidence of the predicted positioning information. Therefore, this embodiment can reduce the loss of characteristic information and improve the accuracy of positioning. At the same time, the obtained positioning information can be evaluated for confidence as a basis for other services.

FIG. 17 is a schematic diagram of an exemplary process of determining target positioning based on a convolutional neural network model according to some embodiments of the present specification. In some embodiments, the process 1700 may be executed by the processing device 112 or other processing devices. For example, the process 1700 may be stored in a storage device (for example, the storage device 140 or a storage unit of a processing device) in the form of a program or instruction. When the processor 220 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 1700. In some embodiments, the process 1700 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 17 is not restrictive.

As shown in FIG. 17, in this embodiment, the convolutional neural network model may include three convolutional layers conv1-conv3 and three fully connected layers connect1-connect3. Among them, in the first conv1 conv1, 3 *3, 5*5 and 7*7 three size convolution kernels are used for feature processing, and the second convolution layer and the third convolution layer are both 5*5 size convolution kernels for feature processing. After each convolution, the pooling layer Max_pooling is used for processing. In this embodiment, multiple convolution kernels of different sizes can be used to effectively extract features on different receptive fields, thus, the loss of feature information can be further reduced. After the feature map is processed by the three convolutional layers conv1-conv3, the corresponding feature vector 103 is obtained, and the feature vector 103 is input to the first fully connected layer connect1. The neural network model can also include a global feature extraction layer 101. The global feature extraction layer 101 introduces the corresponding global features in the positioning request, such as the signal strength of the AP scanned by the terminal that issued the positioning request, and discretizes the corresponding global features in the positioning request. After conversion, the corresponding feature vector 102 is obtained, and the feature vector 102 is input to the first fully connected layer connect1. The feature vector 102 and feature vector 103 are processed by the fully connected layer connect1-connect3 and output position offset, that is, relative to the center The longitude offset Δlon and the latitude offset Δlat of the point can be calculated according to the position coordinates of the center point, the longitude offset Δlon and the latitude offset Δlat relative to the center point, and the positioning information corresponding to this positioning request can be calculated. It should be understood that this embodiment does not limit the number of convolutional layers and the size of the convolution kernel in the convolutional neural network, and may be appropriately adjusted according to actual application scenarios.

In this embodiment, convolutional neural networks are used to perform convolution processing, pooling processing, and full connection processing on the acquired feature maps to obtain positioning information. Among them, the corresponding global features in the positioning request are introduced together with the convolution processed feature maps. Input to the fully connected layer for processing, so that the fingerprint information in the positioning request further participates in the position prediction, thereby further reducing the loss of characteristic information and improving the accuracy of positioning.

Fig. 18 is a block diagram of a positioning device according to some embodiments of this specification. As shown in FIG. 18, in some embodiments, the positioning device 1800 includes a positioning request receiving unit 1810, a center point determining unit 1820, an input grid determining unit 1830, a feature map determining unit 1840, and a positioning information acquiring unit 1850.

As mentioned above, the first obtaining module 410 may be used to obtain a target positioning request. In some embodiments, the first acquisition module 410 may include one or more sub-modules. For example, the first obtaining module 410 may include a request receiving unit 1810. In some embodiments, the request receiving unit 1810 is configured to receive a target positioning request. As mentioned above, the target location request may include the first information, and the first information may include the first fingerprint information. The first fingerprint information may include first fingerprint information corresponding to the current location, and the first fingerprint information includes the identification and signal strength of the wireless access point AP scanned at the current location. For more details about obtaining the target location request, please refer to step 510 and its related description, which will not be repeated here.

As mentioned above, the first determining module 430 may be used to determine the preset point associated with the target positioning request based on the first information. In some embodiments, the first determining module 430 may include one or more sub-modules. For example, the first acquisition module 410 may include a center point determination unit 1820 and an input grid determination unit 1830. In some embodiments, the center point determining unit 1820 is configured to determine a center point corresponding to the first fingerprint information according to predetermined second fingerprint information of different grids, where the grid is a pre-divided geographic area . For more details about determining the preset point, refer to step 530 and its related description, which will not be repeated here.

In some embodiments, the input grid determining unit 1830 is configured to determine a plurality of input grids according to the center point. For more details about determining the input grid, please refer to FIG. 11 and FIG. 12 and related descriptions, which will not be repeated here.

As mentioned above, the generating module 440 may be used to generate at least one first feature map based on preset points. In some embodiments, the generation module 440 may include one or more sub-modules. For example, the generating module 440 may include a feature map determining unit 1840. In some embodiments, the feature map determining unit 1840 is configured to determine one or more feature maps according to the feature information corresponding to each input grid, and the value of each pixel of the feature map corresponds to the feature of each input grid. Information, wherein at least one type of characteristic information is related to the first fingerprint information. For more details about determining the feature map, please refer to step 540 and step 550 and related descriptions, which will not be repeated here.

As mentioned above, the third acquiring module 460 may be used to process at least one second feature map to acquire the target location. In some embodiments, the third acquisition module 460 may include one or more sub-modules. For example, the third obtaining module 460 may include a positioning information obtaining unit 1850. In some embodiments, the positioning information obtaining unit 1850 is configured to input the one or more feature maps into a pre-trained convolutional neural network model to obtain target positioning. For more details about obtaining target positioning, please refer to step 560 and its related description, which will not be repeated here.

The embodiment of this specification determines the center point corresponding to the first fingerprint information in the positioning request according to the predetermined second fingerprint information of different grids, determines a plurality of input grids according to the center point, and determines the characteristics corresponding to each input grid. The information determines multiple feature maps, and the multiple feature maps are input to the pre-trained convolutional neural network model to obtain positioning information, where the feature information corresponding to each input grid may include at least one feature that is the same as the first fingerprint information Therefore, the embodiments of this specification can expand feature information more conveniently, reduce the loss of feature information, and improve the accuracy of positioning.

The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only an example, and does not constitute a limitation to this specification. Although it is not explicitly stated here, those skilled in the art may make various modifications, improvements and amendments to this specification. Such modifications, improvements, and corrections are suggested in this specification, so such modifications, improvements, and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

Meanwhile, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and/or "some embodiments" mean a certain feature, structure, or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that “one embodiment” or “one embodiment” or “an alternative embodiment” mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. . In addition, some features, structures, or characteristics in one or more embodiments of this specification can be appropriately combined.

In addition, those skilled in the art can understand that various aspects of this specification can be explained and described through a number of patentable categories or situations, including any new and useful process, machine, product, or combination of substances, or a combination of them. Any new and useful improvements. Correspondingly, various aspects of this specification can be completely executed by hardware, can be completely executed by software (including firmware, resident software, microcode, etc.), or can be executed by a combination of hardware and software. The above hardware or software can be called "data block", "module", "engine", "unit", "component" or "system". In addition, various aspects of this specification may be embodied as a computer product located in one or more computer-readable media, and the product includes computer-readable program codes.

The computer storage medium may contain a propagated data signal containing a computer program code, for example on a baseband or as part of a carrier wave. The propagated signal may have multiple manifestations, including electromagnetic forms, optical forms, etc., or a suitable combination. The computer storage medium may be any computer readable medium other than the computer readable storage medium, and the medium may be connected to an instruction execution system, device, or device to realize communication, propagation, or transmission of the program for use. The program code located on the computer storage medium can be transmitted through any suitable medium, including radio, cable, fiber optic cable, RF, or similar medium, or any combination of the above medium.

The computer program codes required for the operation of each part of this manual can be written in any one or more programming languages, including object-oriented programming languages such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python Etc., conventional programming languages such as C language, Visual Basic, Fortran2003, Perl, COBOL2002, PHP, ABAP, dynamic programming languages such as Python, Ruby and Groovy, or other programming languages. The program code can run entirely on the user's computer, or as an independent software package on the user's computer, or partly on the user's computer and partly on a remote computer, or entirely on the remote computer or processing equipment. In the latter case, the remote computer can be connected to the user's computer through any network form, such as a local area network (LAN) or a wide area network (WAN), or connected to an external computer (for example, via the Internet), or in a cloud computing environment, or as a service Use software as a service (SaaS).

In addition, unless explicitly stated in the claims, the order of processing elements and sequences, the use of numbers and letters, or the use of other names described in this specification are not used to limit the order of processes and methods in this specification. Although the foregoing disclosure uses various examples to discuss some embodiments of the invention that are currently considered useful, it should be understood that such details are only for illustrative purposes, and the appended claims are not limited to the disclosed embodiments. On the contrary, the rights are The requirements are intended to cover all modifications and equivalent combinations that conform to the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on existing processing devices or mobile devices.

For the same reason, it should be noted that, in order to simplify the expressions disclosed in this specification and help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of this specification, multiple features are sometimes combined into one embodiment. In the drawings or its description. However, this method of disclosure does not mean that the subject of the specification requires more features than those mentioned in the claims. In fact, the features of the embodiment are less than all the features of the single embodiment disclosed above.

In some embodiments, numbers describing the number of ingredients and attributes are used. It should be understood that such numbers used in the description of the embodiments use the modifier "about", "approximately" or "substantially" in some examples. Retouch. Unless otherwise stated, "approximately", "approximately" or "substantially" indicates that the number is allowed to vary by ±20%. Correspondingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, and the approximate values can be changed according to the required characteristics of individual embodiments. In some embodiments, the numerical parameter should consider the prescribed effective digits and adopt the method of general digit retention. Although the numerical ranges and parameters used to confirm the breadth of the ranges in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values is as accurate as possible within the feasible range.

For each patent, patent application, patent application publication and other materials cited in this specification, such as articles, books, specifications, publications, documents, etc., the entire contents are hereby incorporated into this specification as a reference. The application history documents that are inconsistent or conflict with the content of this specification are excluded, and the documents that restrict the broadest scope of the claims of this specification (currently or later appended to this specification) are also excluded. It should be noted that if there is any inconsistency or conflict between the description, definition, and/or use of terms in the accompanying materials of this manual and the content of this manual, the description, definition and/or use of terms in this manual shall prevail. .

Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, the alternative configuration of the embodiment of the present specification can be regarded as consistent with the teaching of the present specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

A network positioning method, characterized in that it comprises:

Obtain the target positioning request;

Acquiring first information associated with the target positioning request, where the first information includes at least: signal characteristics associated with the target positioning request;

Determine a preset point associated with the target positioning request based on the first information;

Generating at least one first feature map based on the preset point;

Determine at least one second feature map based on the at least one first feature map; and

The at least one second feature map is processed to obtain a target location, and the processing at least includes: processing the at least one second feature image based on a convolution kernel.
The method according to claim 1, wherein the determining a preset point associated with the target positioning request based on the first information comprises:

Acquiring multiple signal strengths of the target positioning request; and

Based on the multiple signal strengths, a preset point associated with the target positioning request is determined.
The method according to claim 1, wherein the determining a preset point associated with the target positioning request based on the first information comprises:

Acquiring second information including multiple preset grids;

Calculating the similarity between the first information and the second information;

Determining at least one similar grid based on the similarity; and

The predetermined point is determined based on the at least one similar grid.
The method according to claim 3, wherein the calculating the similarity between the first information and the second information comprises:

Sorting the plurality of preset grids based on the similarity to obtain a similarity sorting result; and,

At least one similar grid is determined based on the similarity ranking result.
The method according to claim 3, wherein the calculating the similarity between the first information and the second information comprises:

Judging whether the similarity satisfies a preset condition; and

If yes, the preset grid whose similarity meets the condition is determined as a similar grid.
The method according to claim 3, wherein the determining the preset point based on the at least one similar grid comprises:

The predetermined point is determined based on one or more of the median, average, and geometric center of the position coordinates of the at least one similar grid.
The method according to claim 3, wherein the determining the preset point based on the at least one similar grid comprises:

Determining the weights of the plurality of preset grids based on the similarity; and

The preset point is determined based on the weight and the similarity.
The method according to claim 1, wherein the processing the at least one second feature map based on the convolution kernel, and obtaining the target location comprises:

Based on the at least one characteristic image, determining position correction information through a convolutional neural network model, the convolutional neural network model including the convolution kernel; and

Acquiring the target location based on the position correction information and the preset point.
The method according to claim 1, wherein the at least one second characteristic map is determined based on the at least one first characteristic map:

Acquire multiple first feature maps, and perform fusion processing on the multiple first feature maps to obtain the at least one second feature map.
The method according to claim 1, wherein the first information further includes one or more of the following associated with the target positioning request:

Heat information, minimum signal strength, maximum signal strength, crowd density, the size of the pixel value of the center grid, and the total size of the values of all grids in the feature map.
The method according to claim 1, wherein the method further comprises:

Determine the confidence of the target location based at least on the confidence network model and the at least one second feature map.
The method according to claim 11, wherein the confidence network model is used to output the confidence of the target positioning according to the proportion of non-empty feature pixels of the at least one second feature map.
The method according to claim 1, wherein the first information further includes the communication relationship between the target positioning request and the primary base station, and the sum of the signal strength matching probabilities of multiple base stations adjacent to the primary base station , At least one of the center distance from the former base station of the main base station, the communication relationship with the former base station, and the sum of heat information with the former base station;

Wherein, the primary base station is the base station to which the terminal corresponding to the target positioning request is currently connected, and the previous base station is the base station to which the terminal is connected before connecting to the primary base station.
The method according to claim 1, wherein the preset point comprises: at least one of a center point of a grid where the target positioning request is located, a maximum signal strength point, and a position coordinate point.
A network positioning system, characterized in that it comprises:

At least one storage medium, the storage medium including an instruction set for network positioning;

At least one processor, the at least one processor is in communication with the at least one storage medium, wherein, when the instruction set is executed, the at least one processor is configured to:

Obtain the target positioning request;

Acquiring first information associated with the target positioning request, where the first information includes at least: signal characteristics associated with the target positioning request;

Determine a preset point associated with the target positioning request based on the first information;

Generating at least one first feature map based on the preset point;

Determine at least one second feature map based on the at least one first feature map; and

The at least one second feature map is processed to obtain a target location, and the processing at least includes: processing the at least one second feature image based on a convolution kernel.
The system according to claim 15, wherein in order to determine a preset point associated with the target positioning request based on the first information, the at least one processor is configured to:

Acquiring multiple signal strengths of the target positioning request; and

Based on the multiple signal strengths, a preset point associated with the target positioning request is determined.
The system according to claim 16, wherein in order to determine a preset point associated with the target positioning request based on the first information, the at least one processor is configured to:

Acquiring second information including multiple preset grids;

Calculating the similarity between the first information and the second information;

Determining at least one similar grid based on the similarity; and

The predetermined point is determined based on the at least one similar grid.
The system according to claim 17, wherein the calculating the similarity between the first information and the second information comprises:

Sorting the plurality of preset grids based on the similarity to obtain a similarity sorting result; and,

At least one similar grid is determined based on the similarity ranking result.
The system according to claim 17, wherein in order to calculate the similarity between the first information and the second information, the at least one processor is configured to:

Judging whether the similarity satisfies a preset condition; and

If yes, the preset grid whose similarity meets the condition is determined as a similar grid.
The system according to claim 17, wherein in order to determine the preset point based on the at least one similar grid, the at least one processor is configured to:

The predetermined point is determined based on one or more of the median, average, and geometric center of the position coordinates of the at least one similar grid.
The system according to claim 17, wherein in order to determine the preset point based on the at least one similar grid, the at least one processor is configured to:

Determining the weights of the plurality of preset grids based on the similarity; and

The preset point is determined based on the weight and the similarity.
The system according to claim 15, wherein, in order to process the at least one second characteristic image based on a convolution kernel to obtain target positioning, the at least one processor is configured to:

Based on the at least one characteristic image, determining position correction information through a convolutional neural network model, the convolutional neural network model including the convolution kernel; and

Acquire the target location based on the position correction information and the preset point.
The system according to claim 15, wherein for the at least one first characteristic map, at least one second characteristic map is determined, and the at least one processor is configured to:

Acquire multiple first feature maps, and perform fusion processing on the multiple first feature maps to obtain the at least one second feature map.
The system according to claim 15, wherein the first information further includes one or more of the following associated with the target positioning request:

Heat information, minimum signal strength, maximum signal strength, crowd density, the size of the pixel value of the center grid, and the total size of the values of all grids in the feature map.
The system of claim 15, wherein the at least one processor is further configured to:

Determine the confidence of the target location based at least on the confidence network model and the at least one second feature map.
The system according to claim 25, wherein the confidence network model is used to output the confidence of the target location according to the proportion of non-empty feature pixels of the at least one second feature map.
The system according to claim 15, wherein the first information further includes the communication relationship between the target positioning request and the primary base station, and the sum of the signal strength matching probabilities of multiple base stations adjacent to the primary base station , At least one of the center distance from the former base station of the main base station, the communication relationship with the former base station, and the sum of heat information with the former base station;

Wherein, the primary base station is the base station to which the terminal corresponding to the target positioning request is currently connected, and the previous base station is the base station to which the terminal is connected before connecting to the primary base station.
The system according to claim 15, wherein the preset point comprises: at least one of a center point of a grid where the target positioning request is located, a maximum signal strength point, and a position coordinate point.
A network positioning system, characterized in that it comprises:

The first obtaining module is used to obtain a target positioning request;

The second acquisition module is configured to acquire first information associated with the target positioning request, where the first information includes at least: signal characteristics associated with the target positioning request;

A first determining module, configured to determine a preset point associated with the target positioning request based on the first information;

A generating module, configured to generate at least one first feature map based on the preset point;

The second determining module is configured to determine at least one second characteristic map based on the at least one first characteristic map; and,

The third acquisition module is configured to process the at least one second feature map to acquire target positioning, and the processing at least includes: processing the at least one second feature image based on a convolution kernel.
A computer-readable storage medium that stores computer instructions for network positioning. After the computer reads the computer instructions for network positioning in the storage medium, the computer executes the following network positioning methods:

Obtain the target positioning request;

Acquiring first information associated with the target positioning request, where the first information includes at least: signal characteristics associated with the target positioning request;

Determine a preset point associated with the target positioning request based on the first information;

Generating at least one first feature map based on the preset point;

Determine at least one second feature map based on the at least one first feature map; and

The at least one second feature map is processed to obtain a target location, and the processing at least includes: processing the at least one second feature image based on a convolution kernel.