CN113645564B - Hybrid excitation method for crowd funding of indoor position information - Google Patents

Hybrid excitation method for crowd funding of indoor position information Download PDF

Info

Publication number
CN113645564B
CN113645564B CN202110758771.7A CN202110758771A CN113645564B CN 113645564 B CN113645564 B CN 113645564B CN 202110758771 A CN202110758771 A CN 202110758771A CN 113645564 B CN113645564 B CN 113645564B
Authority
CN
China
Prior art keywords
crowd
funder
target position
funding
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110758771.7A
Other languages
Chinese (zh)
Other versions
CN113645564A (en
Inventor
张晖
遆宁
赵海涛
孙雁飞
朱洪波
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Posts and Telecommunications
Original Assignee
Nanjing University of Posts and Telecommunications
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing University of Posts and Telecommunications filed Critical Nanjing University of Posts and Telecommunications
Priority to CN202110758771.7A priority Critical patent/CN113645564B/en
Publication of CN113645564A publication Critical patent/CN113645564A/en
Application granted granted Critical
Publication of CN113645564B publication Critical patent/CN113645564B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/023—Services making use of location information using mutual or relative location information between multiple location based services [LBS] targets or of distance thresholds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W12/00—Security arrangements; Authentication; Protecting privacy or anonymity
    • H04W12/12—Detection or prevention of fraud
    • H04W12/121—Wireless intrusion detection systems [WIDS]; Wireless intrusion prevention systems [WIPS]
    • H04W12/122—Counter-measures against attacks; Protection against rogue devices
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/021—Services related to particular areas, e.g. point of interest [POI] services, venue services or geofences
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/33—Services specially adapted for particular environments, situations or purposes for indoor environments, e.g. buildings

Abstract

The invention relates to a mixed incentive method facing indoor position information crowd funding, which applies crowd funding participation thought, considers the quality difference of indoor position information provided by crowd funding persons, designs and filters potential malicious crowd funding persons by adopting the correlation analysis of the maximum information coefficient, and the quality of the data is quantified by adopting an unsupervised learning algorithm fusing weighted clustering and Relieff, therefore, based on the acquisition of the sensing data of the beacon of the target position, the acquisition of the marking information corresponding to the target position is designed to be realized, the logic positioning of each indoor target position is realized, in application, a hybrid excitation framework is introduced, a perception crowd funder is stimulated to provide high-quality perception data for high-quality perception service, the individual rationality and approximately maximized perception platform efficiency of the perception crowd funder are met, and the performance is better than that of the existing method in the aspects of quality assurance and efficiency management.

Description

Hybrid excitation method for crowd funding of indoor position information
Technical Field
The invention relates to a hybrid excitation method for crowd funding of indoor position information, and belongs to the technical field of wireless communication.
Background
Indoor positioning services have attracted attention in recent years due to their social and commercial value, and WiFi fingerprint-based indoor positioning technology has become popular due to its minimal hardware requirements and relatively good positioning accuracy. In indoor positioning based on WiFi fingerprint, positioning is divided into an offline training phase and an online positioning phase, wherein the offline training phase mainly comprises the steps of enabling some professionals to collect WiFi fingerprint data at each position through site survey in a positioning area, and then building a fingerprint database. In the on-line positioning stage, the system matches the WiFi fingerprint of the position of the user with the fingerprint of the fingerprint database to obtain the position of the user with the positioning function, and the matching process is completed by some excellent positioning algorithms. However, one of the technical difficulties of indoor positioning based on WiFi fingerprints is that the construction of a fingerprint database requires a professional to perform a tedious field survey in a positioning area, which consumes a lot of time, manpower and financial resources, and thus the application range of the indoor positioning based on WiFi fingerprints is greatly limited.
Different from the excitation mechanism with a great deal of research work in the directions of a P2P network, an opportunistic network, a delay tolerant network, wireless spectrum allocation and the like, a new architecture and network relationship are developed by the masses, so that the excitation mechanism mode is more diversified. In crowd funding, the general participation of mobile users enables objects of incentives to have mobile attributes, such as geographic positions and the influence of task coverage, which makes the incentive problem more complex, and new technologies and theories also enable crowd funding incentive mechanisms to be different from incentive mechanisms in traditional networks, such as the influence of social network relations and the like. The main objective of the group perception incentive mechanism research is to incentivize holders (crowd funders) with perception devices to join in perception projects, actively participate in perception tasks and submit high-quality and reliable perception data under the management of a server platform.
Research on crowd funding incentive mechanisms mainly focuses on external incentives, which are not much involved and are single in form. The auction mechanism becomes the most common incentive mode in group collaborative computing in the external incentive. The method is basically suitable for all perception workers, but often brings high cost problem. The design of many incentive mechanisms has the problems of integrity, privacy and safety of workers and high cost of workers, so that the incentive effect is not ideal. In addition, existing incentive mechanisms are designed for guaranteeing the number of perception crowd funders mostly, and the problem of perception data quality is rarely considered. Blind incentives often result in inefficient sensing, poor quality of service, unnecessary waste of human and material resources and platform capital for the sensing crowd funders.
Although the related art has been proposed to solve the above problems, such as a method of mixing an external incentive scheme with an internal incentive scheme, which gives rewards of status, honor, etc. to incentivize workers to participate in tasks, from the motivation of the workers, there are problems in that it is difficult to detect data quality without knowing the perception of crowd funder behavior information and data truth values. Data validation may require deployment of conventional sensors to collect noise truth values in real-time in the field. The deployment needs a large amount of capital investment, has poor flexibility and expandability and negates the significance of crowd funding. Second, it is difficult to design incentive schemes that simultaneously satisfy both psychology and maximization of platform performance. The individual rationality means that the crowd funding person needs to obtain a reward which is not less than the crowd funding expense to complete the crowd funding task; platform performance refers to the value gained by the platform providing services minus the payment paid by the hiring crowd funder. A well-designed incentive mechanism not only ensures that the crowd-funding crowd-sourcing crowd-funding crowd-sourcing crowd-funding crowd-sourcing crowd-funding crowd-supporting crowd-funding. Furthermore, it is difficult to efficiently combine data quality with incentive schemes. The crowd funding person has different physical resources and individual abilities, submits crowd funding data with different qualities, and naturally requires rewards with different values. However, the existing uniform price reward mechanism lacks fairness and incentive, and the price reward mechanism is despite the careless behavior of some crowd-funding crowd-funders, and both of the two ways cannot maintain long-term efficient data collection and high-quality crowd-funding service.
Disclosure of Invention
The invention aims to solve the technical problem of providing a crowd funding hybrid incentive method for indoor position information, which is applied to a crowd funding participant mode, realizes the acquisition of target position effective data through set data screening and quality mechanisms based on the acquisition of target position beacon sensing data, and improves the efficiency of indoor positioning application.
The invention adopts the following technical scheme for solving the technical problems: the invention designs a mixed incentive method for crowd funding of indoor position information, which is based on the fact that each crowd funder uploads beacon perception data about each indoor target position respectively to realize the screening of each crowd funder corresponding to each target position respectively, and comprises the following steps:
step A. based on each crowd funding xiThe uploaded time sequence perception data sequences respectively corresponding to preset beacon perception data types of indoor target positions form crowd funders xiSeparately uploaded beacon-aware datasets for all indoor target locations
Figure GDA0003512165700000021
Obtaining the maximum mutual information value between beacon sensing data sets corresponding to two different crowd funders
Figure GDA0003512165700000022
Then entering the step B; wherein I is more than or equal to 1 and less than or equal to I, j is more than or equal to 1 and less than or equal to I, I is not equal to j, I represents the total number of crowd funding people, xiDenotes the ith crowd funding, xjRepresents the j-th crowd funder,
Figure GDA0003512165700000023
indicating that the ith crowd funder uploaded a beacon-aware data set for all target locations indoors,
Figure GDA0003512165700000024
representing a beacon perception data set uploaded by a jth crowd funder on all target locations indoors;
step B, aiming at each crowd funding x respectivelyiObtaining the sum of the maximum mutual information value between each crowd funder and the rest crowd funders as the corresponding mutual information value of the crowd funder, namely obtaining x of each crowd funderiRespectively corresponding mutual information values, deleting all crowd-funders of which the mutual information values are smaller than a preset mutual information threshold value, updating the total number I of the crowd-funders, and then entering the step C;
step C, aiming at each indoor target position, executing the following steps C1 to C4 to obtain each target crowd funder corresponding to each target position;
step C1. respectively aiming at each crowd funding xiBased on crowd funding person xiThe uploaded time sequence sensing data sequence corresponding to the preset beacon sensing data type relative to the target position obtains the latest uploaded time sequence
Figure GDA0003512165700000031
Data of a person
Figure GDA0003512165700000032
And wherein historical upload
Figure GDA0003512165700000033
Data of a person
Figure GDA0003512165700000034
Then proceed to step C2; wherein the content of the first and second substances,
Figure GDA0003512165700000035
represents the crowd funding person xiThe length of the uploaded time sequence sensing data sequence corresponding to the preset beacon sensing data type relative to the target position is half, I is more than or equal to 1 and less than or equal to I, I represents the total number of crowd funding people, xiRepresenting the ith crowd funder;
step C2. targets each crowd funder xiBased on crowd funding person xiThe preset initial weight value corresponding to the target position is as follows:
Figure GDA0003512165700000036
iterating according to preset iteration times to obtain crowd funding person xiWeight value Tod [ x ] corresponding to the target positioni]Then proceed to step C3; wherein j is more than or equal to 1 and less than or equal to I, I is not equal to j, xjRepresents the j-th crowd funder,
Figure GDA0003512165700000037
representing crowd funding x of comparesiOther crowd funding people xjUploaded in time sequence sensing data sequence corresponding to preset beacon sensing data type relative to target position according to latest time sequence
Figure GDA0003512165700000038
The number of the data is one,
Figure GDA0003512165700000039
to represent
Figure GDA00035121657000000310
And
Figure GDA00035121657000000311
the difference between the two properties is that,
Figure GDA00035121657000000312
to represent
Figure GDA00035121657000000313
And
Figure GDA00035121657000000314
the difference between them, prop (x)j) Represents the crowd funding person xjThe ratio of the number of data in the time sequence sensing data sequence of the target position corresponding to the preset beacon sensing data type to the total number of data of the time sequence sensing data sequence of the target position corresponding to the preset beacon sensing data type uploaded by all crowd funders is calculated;
step C3. is based on the crowd-funder x uploading a chronological awareness data sequence corresponding to a predetermined beacon awareness data type for the target locationiAccording to crowd funding person xiWeight value Tod [ x ] corresponding to the target positioni]Combining all crowd funders xiRespectively corresponding to the initial data quality
Figure GDA00035121657000000315
And the initial clustering mass center w corresponding to each crowd funder as followsFormula (II):
Figure GDA00035121657000000316
Figure GDA00035121657000000317
Figure GDA00035121657000000318
Figure GDA0003512165700000041
performing iteration to obtain data quality corresponding to each crowd funder
Figure GDA0003512165700000042
Obtaining various crowd funders x by taking the condition that the change in two adjacent iterations is smaller than a preset change threshold as an end conditioniData quality relating to the target position respectively
Figure GDA0003512165700000043
Then proceed to step C4; wherein ε represents a predetermined positive real number, dist (w, x)i) Representing clustering centroid w and crowd funder xiBased on the distance of intercepting the time sequence sensing data sequences with the same length, W represents a set of time sequence sensing data sequences which are uploaded by all people raising persons and respectively related to all indoor target positions;
step C4, according to each crowd funding person xiData quality relating to the target position respectively
Figure GDA0003512165700000044
In descending order of the average number of people xiSorting is carried out, and from the first, the order is selected before
Figure GDA0003512165700000045
The individual crowd funder forms each target crowd funder corresponding to the target position; wherein, a represents a preset screening proportion,
Figure GDA0003512165700000046
indicating rounding up.
As a preferred technical scheme of the invention: said representation
Figure GDA0003512165700000047
And
Figure GDA0003512165700000048
the difference between them
Figure GDA0003512165700000049
To represent
Figure GDA00035121657000000410
And
Figure GDA00035121657000000411
the difference between them
Figure GDA00035121657000000412
And clustering centroid w and crowd funder xiDistance dist (w, x) between the two time series sensing data sequences based on intercepting preset same length time series sensing data sequencesi) The Euclidean distances are adopted.
As a preferred technical scheme of the invention: and C, based on the step C, after obtaining each target crowd funder corresponding to each target position, respectively aiming at each target position, according to the time sequence perception data sequence of each target crowd funder corresponding to the target position, respectively relating to the target position, constructing the marking information of the unique identifier corresponding to the target position, namely obtaining the marking information corresponding to each target position, and realizing the logic positioning of each indoor target position.
As a preferred technical scheme of the invention: the preset beacon sensing data type is wireless signal strength.
Compared with the prior art, the hybrid incentive method for crowd funding of indoor position information has the following technical effects:
the invention designs a mixed incentive method facing indoor position information crowd funding, applies crowd funding participation thought, considers the quality difference of indoor position information provided by crowd funding persons, designs and filters potential malicious crowd funding persons by adopting the correlation analysis of the maximum information coefficient, and the quality of the data is quantified by adopting an unsupervised learning algorithm fusing weighted clustering and Relieff, therefore, based on the acquisition of the sensing data of the beacon of the target position, the acquisition of the marking information corresponding to the target position is designed to be realized, the logic positioning of each indoor target position is realized, in application, a hybrid excitation framework is introduced, a perception crowd funder is stimulated to provide high-quality perception data for high-quality perception service, the individual rationality and approximately maximized perception platform efficiency of the perception crowd funder are met, and the performance is better than that of the existing method in the aspects of quality assurance and efficiency management.
Drawings
FIG. 1 is a schematic diagram of an embodiment of an incentive framework design in a hybrid incentive method for crowd funding of indoor location information according to the invention;
FIG. 2 is a schematic flow chart of a reverse auction-based system provided in the present invention;
FIG. 3 is a schematic flow diagram of a reverse auction-based, game-based hybrid incentive framework provided in the present invention;
fig. 4 is a flow diagram of the voting mechanism provided in the present invention.
Detailed Description
The following description will explain embodiments of the present invention in further detail with reference to the accompanying drawings.
The invention designs a mixed incentive method for crowd funding of indoor position information, which is based on the fact that each crowd funder uploads beacon perception data about each indoor target position respectively, screening of the crowd funders corresponding to each target position respectively is achieved, and in practical application, the following steps A to C are specifically designed and executed.
Step A. based on each crowd funding xiThe uploaded time sequence sensing data sequence respectively corresponding to preset beacon sensing data types such as wireless signal intensity of indoor target positions forms various crowd funders xiSeparately uploaded beacon-aware datasets for all indoor target locations
Figure GDA0003512165700000051
Obtaining the maximum mutual information value between beacon sensing data sets corresponding to two different crowd funders
Figure GDA0003512165700000052
Then entering the step B; wherein I is more than or equal to 1 and less than or equal to I, j is more than or equal to 1 and less than or equal to I, I is not equal to j, I represents the total number of crowd funding people, xiDenotes the ith crowd funding, xjRepresents the j-th crowd funder,
Figure GDA0003512165700000053
indicating that the ith crowd funder uploaded a beacon-aware data set for all target locations indoors,
Figure GDA0003512165700000054
indicating that the j-th crowd funder uploaded a beacon-aware data set for all target locations indoors.
The relationship between location information provided by crowd funders is evaluated by means of MICs with more extensive correlation identification capabilities, where mutual information is defined as the amount of information contained in one random variable about another random variable, or the uncertainty that one random variable reduces because another random variable is known, and is used in feature selection studies to measure the correlation of two features. For two features X ═ Xi,x2,...,xKY ═ Yi,y2,...,yKAnd expanding according to the definition of entropy:
Figure GDA0003512165700000061
the maximum mutual information method approximates probability distribution using an idea similar to a histogram, thereby calculating mutual information. On the basis of meshing a random variable scatter plot, the probability is the number of data points falling into the mesh that is proportional. The calculation process is divided into three steps, firstly, under the condition of giving two integer values r and c, r rows and c columns of grid division are carried out on a scatter diagram of two random variables, and the maximum mutual information value is solved; secondly, normalizing the maximum mutual information value; thirdly, the maximum value of mutual information under different division scales is selected as the Maximum Information Coefficient (MIC) value.
Step A above relates to the maximum mutual information value
Figure GDA0003512165700000062
In practical application, the method is specifically aimed at various crowd funders xiSeparately uploaded beacon-aware datasets for all indoor target locations
Figure GDA0003512165700000063
Is provided with a division G
Figure GDA0003512165700000064
Dividing the corresponding scatter diagram into r × c grids, calculating the mutual information of G in each division mode,
Figure GDA0003512165700000065
the maximum mutual information under partition G is written as:
Figure GDA0003512165700000066
will I*Divided by logmin { r, c } for normalization. Combining the maximum normalized mutual information obtained under different scale division into a characteristic matrix, and recording the matrix as
Figure GDA0003512165700000067
Each element m in the matrixr,cFor the maximum normalized mutual information obtained by arbitrary r × c trellis division:
Figure GDA0003512165700000068
finally, the Maximum Information Coefficient (MIC) is the maximum mutual information value of the squares in the matrix, defined as:
Figure GDA0003512165700000069
maximum information factor between different crowd funders
Figure GDA00035121657000000610
The larger the value, the stronger the correlation between them, the higher the possibility of mutual substitution when used for fingerprint library construction, and the higher the redundancy.
Crowd funder x for each individualiFor the uploaded time-series sensing data sequence of the target position corresponding to the preset beacon sensing data type, the true value is unknown, and the data quality of each user needs to be estimated by taking the weighted aggregation of the user data and the crowd funder feasibility as a reference, so the following steps are continuously executed.
Step B, aiming at each crowd funding x respectivelyiObtaining the sum of the maximum mutual information value between each crowd funder and the rest crowd funders as the corresponding mutual information value of the crowd funder, namely obtaining x of each crowd funderiAnd D, respectively corresponding mutual information values, deleting all the crowd funders of which the mutual information values are smaller than a preset mutual information threshold value, updating the total number I of the crowd funders, and then entering the step C.
And C, aiming at each indoor target position, executing the following steps C1 to C4 to obtain each target crowd funder corresponding to each target position.
Step C1. respectively aiming at each crowd funding xiBased on crowd funding person xiThe uploaded time sequence sensing data sequence corresponding to the preset beacon sensing data type relative to the target position obtains the latest uploaded time sequence
Figure GDA0003512165700000071
Data of a person
Figure GDA0003512165700000072
And wherein historical upload
Figure GDA0003512165700000073
Data of a person
Figure GDA0003512165700000074
Then proceed to step C2; wherein the content of the first and second substances,
Figure GDA0003512165700000075
represents the crowd funding person xiThe length of the uploaded time sequence sensing data sequence corresponding to the preset beacon sensing data type relative to the target position is half, I is more than or equal to 1 and less than or equal to I, I represents the total number of crowd funding people, xiIndicating the ith crowd funder.
The Relief algorithm is a classical feature selection method for classification problems, and a family of Relief algorithms comprises a series of algorithms including an earliest proposed Relief algorithm, a later expanded reliefF algorithm and an RReliefF algorithm. The Relief algorithm is a classical feature selection method for the class II data classification problem, has the advantages of easiness in implementation, high execution efficiency and the like, and has the core idea that a feature subset with the best class discrimination capability is selected, the discrimination capability of each feature is represented by a weight value, and the weight value is obtained by calculating the condition that similar samples are distinguished based on the value of the feature, namely the following step C2 is continuously executed.
Step C2. targets each crowd funder xiBased on crowd funding person xiThe preset initial weight value corresponding to the target position is as follows:
Figure GDA0003512165700000076
iterating according to preset iteration times to obtain crowd funding person xiWeight value Tod [ x ] corresponding to the target positioni]Then proceed to step C3; wherein j is more than or equal to 1 and less than or equal to I, I is not equal to j, xjRepresents the j-th crowd funder,
Figure GDA0003512165700000077
representing crowd funding x of comparesiOther crowd funding people xjUploaded in time sequence sensing data sequence corresponding to preset beacon sensing data type relative to target position according to latest time sequence
Figure GDA0003512165700000078
The number of the data is one,
Figure GDA0003512165700000079
to represent
Figure GDA00035121657000000710
And
Figure GDA00035121657000000711
the difference between the two properties is that,
Figure GDA00035121657000000712
to represent
Figure GDA00035121657000000713
And
Figure GDA00035121657000000714
the difference between them, prop (x)j) Represents the crowd funding person xjThe ratio of the number of data in the time sequence sensing data sequence corresponding to the preset beacon sensing data type about the target position to the total number of data of the time sequence sensing data sequence corresponding to the preset beacon sensing data type about the target position uploaded by all crowd funders is obtained.
In practice, it is shown here
Figure GDA00035121657000000715
And
Figure GDA00035121657000000716
the difference between them
Figure GDA00035121657000000717
To represent
Figure GDA00035121657000000718
And
Figure GDA00035121657000000719
the difference between them
Figure GDA0003512165700000081
Are realized by adopting Euclidean distance.
The Relief algorithm is based on the idea that: "good" features, i.e., features that are discriminative, should be those that bring samples of the same class closer together, while keeping samples of different classes farther apart.
Step C3. is based on the crowd-funder x uploading a chronological awareness data sequence corresponding to a predetermined beacon awareness data type for the target locationiAccording to crowd funding person xiWeight value Tod [ x ] corresponding to the target positioni]Combining all crowd funders xiRespectively corresponding to the initial data quality
Figure GDA0003512165700000082
And the initial clustering mass center w corresponding to each crowd funder is as follows:
Figure GDA0003512165700000083
Figure GDA0003512165700000084
Figure GDA0003512165700000085
Figure GDA0003512165700000086
performing iteration to obtain data quality corresponding to each crowd funder
Figure GDA0003512165700000087
Obtaining various crowd funders x by taking the condition that the change in two adjacent iterations is smaller than a preset change threshold as an end conditioniData quality relating to the target position respectively
Figure GDA0003512165700000088
Then proceed to step C4; wherein ε represents a predetermined positive real number, dist (w, x)i) Representing clustering centroid w and crowd funder xiBased on the distance of intercepting the time sequence sensing data sequences with the preset same length, W represents the set of the time sequence sensing data sequences which are uploaded by each crowd funder and respectively related to each indoor target position.
In practical application, the centroid w of the cluster and the crowd funder x are clusterediDistance dist (w, x) between the two time series sensing data sequences based on intercepting preset same length time series sensing data sequencesi) The method is realized by adopting the Euclidean distance.
Step C4, according to each crowd funding person xiData quality relating to the target position respectively
Figure GDA0003512165700000089
In descending order of the average number of people xiSorting is carried out, and from the first, the order is selected before
Figure GDA00035121657000000810
The individual crowd funder forms each target crowd funder corresponding to the target position; wherein, a represents a preset screening proportion,
Figure GDA00035121657000000811
indicating rounding up.
In fact, under the condition of the same data quantity, the sensing data with high quality can give more effective information to the sensing platform than the sensing data with low quality. High quality data is inherently more useful, while low quality data often requires additional processing to be put to use, such as channel check codes (ECC) to detect and correct transmission errors. Group awareness typically corrects errors using multiple people repeating tasks, additional hirers checking data quality, and awareness of other ancillary data, etc.
After the steps a to C are executed to obtain each target crowd funder corresponding to each target position, in practical application, it is further designed to construct, for each target position, label information of a unique identifier corresponding to the target position according to a time sequence perception data sequence of each target crowd funder corresponding to the target position, that is, to obtain label information corresponding to each target position, so as to implement logical positioning of each indoor target position.
The design introduces the idea of crowd funding, and can well solve the problems in the prior art. Based on the indoor fingerprint positioning technology of mobile crowdsourcing, WiFi fingerprint data can be collected by using a mobile phone of a common user, the process of field survey of a specially-assigned person is avoided, the frequent updating of a fingerprint database can be kept, and the environment change is adapted. Through the research on the indoor fingerprint positioning technology based on mobile crowdsourcing, the sensor nodes which need the most service can be conveniently found for the energy collection system in the indoor environment to carry out wireless energy service. In addition, in many daily life scenes, including shopping malls, hospitals, museums, parking lots and the like, there is an urgent need for positioning services.
With the popularization and widespread use of various mobile portable devices, such as smart phones, tablet computers, wearable devices, and the like, the crowd funds a new mode of sensing the environment, collecting data, and providing information services. Crowd funding refers to forming a perception network through mobile equipment existing in people and releasing tasks to individuals or groups in the network to complete the crowd funding, so that professionals or the public are helped to collect data, analyze information and share knowledge. Traditional data collection methods, such as wireless sensor networks and distributed sensor networks, generally adopt a static method, are easily limited by, for example, insufficient node coverage, costly installation and management costs, and lack of expandability, and encounter many difficulties in actual deployment and application. However, in crowd funding, combining human perception judgment capability and perception capability of the mobile device itself, such as geographical location, climate environment, traffic condition, etc., it is a critical issue how to attract more users to participate in tasks through reasonable incentive mechanisms and to promote the users to provide high-quality data.
In practical application, the execution of the steps A to C in the designed hybrid incentive method for the crowd funding of the indoor position information is regarded as one round of execution based on the acquisition of each target crowd funder corresponding to each target position, namely, a task requester issues a position information task for obtaining each target position in the indoor space, each crowd funder executes the steps according to the steps, after the execution of the round, each target position corresponds to each target crowd funder respectively, namely, each target position can be regarded as a trust crowd funder of the target position, namely, an incentive mechanism can be introduced to the trust crowd funder, and the trust crowd funder is encouraged to continue to provide a time sequence perception data sequence for the corresponding target position.
Purely from external incentives, it is easy for workers to work for the purpose of obtaining rewards, resulting in poor crowd funding task results. In addition, compared with the intrinsic incentive mechanism, the extrinsic incentive mechanism based on money in long-term application can lead workers to reduce participation in a single intrinsic incentive, and because the characters of human beings have difference and the demands are inconsistent, a universal mechanism is difficult to form from the intrinsic incentive of the workers. As one of the solutions, a hybrid incentive framework is specifically adopted, which integrates widely popular incentive methods: the reverse auction and the gamification ensure the long-term participation of the users, effectively reduce the time loss of task release, further stimulate the crowd funders to provide high-quality data and stimulate the crowd funders to provide real feedback. On one hand, the method combines an incentive mechanism and data quality to encourage crowd funding people to carry out effective crowd funding activities and submit high-quality crowd funding data, and has good practicability; on the other hand, the method improves the participation degree of crowd funding personnel by combining the internal and external incentive mechanisms.
The incentive framework has three stakeholders participating in the mobile crowdsourcing system: crowd funders, task publishers (i.e., crowdsourcing) and crowdsourcing platforms. The proposed incentive scheme framework, which is relevant to all three stakeholders, consists of two parts: the two-round crowd funder selection process based on reverse auction is based on verification of a gamification voting mechanism.
First, a two-round crowd funder selection mechanism based on reverse auctions to encourage the crowd to actively participate and provide high quality sensory data. The publisher provides a short description of the task, describes its nature, time constraints, etc. The platform filters potential malicious crowd-funding persons through two rounds of selection, and according to each crowd-funding person xiRank price siSelected by ascending sort, wherein the rank price is based primarily on crowd funder's offer biAnd the data quality q provided by itiWherein s isi＝bi×(1-qi). On the basis, the platform ranks s according to the quotationiSelecting m crowd funding persons while satisfying
Figure GDA0003512165700000101
Second, to avoid the publisher's unreal feedback on perceived data quality, a gamification-based verification mechanism aims to evaluate the authenticity of the publisher's feedback. There are generally two cases if the publisher's feedback on the quality of the data is bad. One situation is that the publisher's evaluation is real, that is, the crowd funder i's data quality is really bad. Alternatively, the publisher may want to use crowd funding data from crowd funders, but want to pay as little consideration as possible, perhaps by requiring a false report. To address this problem, the platform will employ n players who will provide an assessment of the trustworthiness of the data e by votinggAnd each player is given a small prize c:
Figure GDA0003512165700000102
wherein q isjIs the quality of the location information provided by user j, which is used to measure the voting ability of the user,
Figure GDA0003512165700000103
the actual ticketing for user i.
According to the integrated incentive framework, the problem that free-riding easily occurs in a reverse auction mechanism is solved to a certain extent, namely, a task requester pays consideration to cooperative workers before a task is executed, and the workers tend to pay less effort, so that the data quality is reduced. The integrated incentive frame adopts an unsupervised learning algorithm of the fusion of weighted clustering and Relieff to quantify the data quality, and adopts the correlation analysis of the maximum information coefficient to filter potential malicious crowd funders, so that the incentive perception crowd funders provide high-quality perception data for high-quality perception service, the personal rationality and the approximately maximized perception platform efficiency of the perception crowd funders are met, and the method has better performance in the aspects of quality assurance and efficiency management compared with the conventional method.
The hybrid incentive method for crowd funding of indoor location information and the introduced hybrid incentive framework are applied to practice, as shown in fig. 1, a reverse auction-based and game-based hybrid incentive framework flow proposed by the invention is shown, and the hybrid incentive method comprises the following two crowd funder selection processes based on the reverse auction, as shown in fig. 2.
The reverse auction component completes task allocation and payment, crowd funding selection and recruitment, data credibility calculation and publisher feedback on the quality of the sensed data, and a flow diagram based on the reverse auction is shown in fig. 3.
Figure 3 shows a reverse auction based, game-based hybrid incentive framework flow diagram that generally has two scenarios if the publisher's feedback on data quality is bad. One situation is that the publisher's evaluation is real, that is, the crowd funder i's data quality is really bad. Alternatively, the publisher wants to use the perception data from crowd funders, but wants to payPayment is made as little as possible and may be accomplished by the need for false reports. To address this issue, the incentive framework explicitly uses a gamification verification mechanism to infer whether the publisher's feedback is authentic. Figure 4 shows a specific voting process, in particular the platform will hire n players and give each player a small prize c. These players will then provide an assessment of the quality of the data by voting eg：
Figure GDA0003512165700000111
Wherein q isjIs the quality of the location information provided by user j during the reverse auction phase, which is used to measure the voting ability of the user,
Figure GDA0003512165700000112
the actual ticketing for user i.
It should be noted that the prize, c, paid to each player is optional for the platform. The reason for setting a comparatively small monetary reward is mainly based on the consideration that the gaming mechanism can be understood as an interesting crowd-sourced task in the proposed incentive framework, wherein the platform plays the role of the publisher and the game player is a crowd-funder. Furthermore, it is observed that users, while primarily seeking fun for gambling, also readily accept economic incentives as an additional incentive. Thus, initially, such bonus settings may entice players to complete verification tasks at a relatively low rate of labor, and then once users enjoy the gaming process, they will continue to participate without expecting economic compensation.
The basis of the comparison algorithm is explained as follows.
If eg∈[0,0.6]Meaning that both ratings (publisher's feedback and gamer validation) are bad ratings; in this case, the feedback of the publisher is considered to be real. In this case, the final rating is further calculated as follows: when delta is more than or equal to 0.3, adopting e ═ e (e)g+epu) 2 to smooth out relatively large numbers between two ratingsQuantity difference; otherwise, let e equal to epu(meaning that the gap between the two ratings is small enough to accommodate this assumption).
If eg∈[0.6,1]And Δ ≧ 0.4, this means that the gamer scored very high, but the publisher gave poor ratings, the gap between these ratings was sufficient for the gamer's ratings to be considered correct (i.e., the publisher provided false feedback). Then let e equal to eg(ii) a Otherwise, if Δ ∈ [0.2,0.4 ]]This means that the gap between the two ratings is moderate; then using e ═ e (e)g+epu) 2 and a threshold of 0 between it and the good-bad rating. 6 to infer whether the feedback of the publisher is real. When Delta epsilon [0,0.2 ∈]It is only necessary to make e equal to epu. This means that the gap is too small to trust the feedback of the publisher.
Through this comparison algorithm, the platform can infer whether the publisher's feedback is authentic. If the publisher's rating is authentic, the publisher will be refunded bi+ n · c, which means that the publisher does not have to pay additional game fees; otherwise, no refund is given and the actual payment by the publisher will be bi+biP + 2. n.c, the first term biIs a reward to crowd funder i, and the second item is biP platform's yield, where one n-c is the cost of gameplay and the other n-c is the penalty of providing false feedback to the publisher. In summary, after gamification, it is possible to obtain the value of the final rating e (correctly evaluated data quality) and know whether the publisher feedback is authentic.
The application is executed as follows:
a. task allocation and deposit payment
Publishers purchase awareness services from the community awareness platform by paying a deposit to the platform. The amount paid by the publisher depends on the crowd funder's bid, platform profit, and additional gamification fees. Note that the publisher's deposit is refunded if and only if the publisher reports a service failure (poorly rated sensory data) and the platform successfully verifies the failure report by gamification. Thus, the deposit payment itself is only for a guarantee. The publisher needs to provide a short description of the task, describe its nature, time constraints, etc. Note that the tasks should be divided into small pieces for the mobile device to complete. The platform then distributes the micro-tasks to the crowd, recruiting crowd funders by open calls.
b. First round crowd funder screening
In the first round, the platform selects the candidate to participate according to the data similarity provided by the crowd funder, measures the redundancy (correlation) among the crowd funder by adopting the maximum information coefficient, calculates the maximum information coefficient MIC among the crowd funder, and the larger the value is, the stronger the correlation is, the stronger the possibility of mutual substitution is when the platform is used for position estimation, and the higher the redundancy is.
c. Second-round crowd funder selection and employment
According to ranking price s of each crowd funder iiSelected by ascending sort, wherein the rank price is based primarily on crowd funder's offer biAnd its data quality qiWherein s isi＝bi×(1-qi). On the basis, the platform ranks s according to the quotationiSelecting M crowd funding persons while satisfying
Figure GDA0003512165700000131
The reward that the crowd funder i can ultimately obtain is still its offer.
d. Publisher feedback on location information
The publisher will evaluate the data to obtain epuFor epuThe criteria for evaluation are as follows:
Figure GDA0003512165700000132
if the publisher's c rating is good, the transaction is successfully completed and the crowd funder's offer is paid to the crowd funder as an award. Otherwise, the platform executes a gamification mechanism to verify that the publisher's evaluation is authentic.
The mixed incentive method for crowd funding of indoor position information is designed by the technical scheme, the crowd funding participation thought is applied, the quality difference of indoor position information provided by crowd funders is considered, the potential malicious crowd funders are analyzed and filtered by adopting the correlation of the maximum information coefficient, and the quality of the data is quantified by adopting an unsupervised learning algorithm fusing weighted clustering and Relieff, therefore, based on the acquisition of the sensing data of the beacon of the target position, the acquisition of the marking information corresponding to the target position is designed to be realized, the logic positioning of each indoor target position is realized, in application, a hybrid excitation framework is introduced, a perception crowd funder is stimulated to provide high-quality perception data for high-quality perception service, the individual rationality and approximately maximized perception platform efficiency of the perception crowd funder are met, and the performance is better than that of the existing method in the aspects of quality assurance and efficiency management.
The embodiments of the present invention have been described in detail with reference to the drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the gist of the present invention.

Claims (4)

1. A hybrid incentive method for crowd funding of indoor position information is characterized by comprising the following steps: the method is characterized in that the method comprises the following steps of based on the fact that each crowd-funder uploads beacon perception data about each indoor target position respectively, screening of the crowd-funders corresponding to each target position respectively is achieved, and the method comprises the following steps:
step A. based on each crowd funding xiThe uploaded time sequence perception data sequences respectively corresponding to preset beacon perception data types of indoor target positions form crowd funders xiSeparately uploaded beacon-aware datasets for all indoor target locations
Figure FDA0003512165690000011
Obtaining the maximum mutual information value between beacon sensing data sets corresponding to two different crowd funders
Figure FDA0003512165690000012
Then entering the step B; wherein I is more than or equal to 1 and less than or equal to I, j is more than or equal to 1 and less than or equal to I, I is not equal to j, and I represents the total number of crowd funding people，xiDenotes the ith crowd funding, xjRepresents the j-th crowd funder,
Figure FDA0003512165690000013
indicating that the ith crowd funder uploaded a beacon-aware data set for all target locations indoors,
Figure FDA0003512165690000014
representing a beacon perception data set uploaded by a jth crowd funder on all target locations indoors;
step B, aiming at each crowd funding x respectivelyiObtaining the sum of the maximum mutual information value between each crowd funder and the rest crowd funders as the corresponding mutual information value of the crowd funder, namely obtaining x of each crowd funderiRespectively corresponding mutual information values, deleting all crowd-funders of which the mutual information values are smaller than a preset mutual information threshold value, updating the total number I of the crowd-funders, and then entering the step C;
step C, aiming at each indoor target position, executing the following steps C1 to C4 to obtain each target crowd funder corresponding to each target position;
step C1. respectively aiming at each crowd funding xiBased on crowd funding person xiThe uploaded time sequence sensing data sequence corresponding to the preset beacon sensing data type relative to the target position obtains the latest uploaded time sequence
Figure FDA0003512165690000015
Data of a person
Figure FDA0003512165690000016
And wherein historical upload
Figure FDA0003512165690000017
Data of a person
Figure FDA0003512165690000018
Then proceed to step C2; wherein the content of the first and second substances,
Figure FDA0003512165690000019
represents the crowd funding person xiThe length of the uploaded time sequence sensing data sequence corresponding to the preset beacon sensing data type relative to the target position is half, I is more than or equal to 1 and less than or equal to I, I represents the total number of crowd funding people, xiRepresenting the ith crowd funder;
step C2. targets each crowd funder xiBased on crowd funding person xiThe preset initial weight value corresponding to the target position is as follows:
Figure FDA00035121656900000110
iterating according to preset iteration times to obtain crowd funding person xiWeight value Tod [ x ] corresponding to the target positioni]Then proceed to step C3; wherein j is more than or equal to 1 and less than or equal to I, I is not equal to j, xjRepresents the j-th crowd funder,
Figure FDA0003512165690000021
representing crowd funding x of comparesiOther crowd funding people xjUploaded in time sequence sensing data sequence corresponding to preset beacon sensing data type relative to target position according to latest time sequence
Figure FDA00035121656900000218
The number of the data is one,
Figure FDA0003512165690000022
to represent
Figure FDA0003512165690000023
And
Figure FDA0003512165690000024
the difference between the two properties is that,
Figure FDA0003512165690000025
to represent
Figure FDA0003512165690000026
And
Figure FDA0003512165690000027
the difference between them, prop (x)j) Represents the crowd funding person xjThe ratio of the number of data in the time sequence sensing data sequence of the target position corresponding to the preset beacon sensing data type to the total number of data of the time sequence sensing data sequence of the target position corresponding to the preset beacon sensing data type uploaded by all crowd funders is calculated;
step C3. is based on the crowd-funder x uploading a chronological awareness data sequence corresponding to a predetermined beacon awareness data type for the target locationiAccording to crowd funding person xiWeight value Tod [ x ] corresponding to the target positioni]Combining all crowd funders xiRespectively corresponding to the initial data quality
Figure FDA0003512165690000028
And the initial clustering mass center w corresponding to each crowd funder is as follows:
Figure FDA0003512165690000029
Figure FDA00035121656900000210
Figure FDA00035121656900000211
Figure FDA00035121656900000212
performing an iteration toData quality corresponding to each crowd funder
Figure FDA00035121656900000213
Obtaining various crowd funders x by taking the condition that the change in two adjacent iterations is smaller than a preset change threshold as an end conditioniData quality relating to the target position respectively
Figure FDA00035121656900000214
Then proceed to step C4; wherein ε represents a predetermined positive real number, dist (w, x)i) Representing clustering centroid w and crowd funder xiBased on the distance of intercepting the time sequence sensing data sequences with the same length, W represents a set of time sequence sensing data sequences which are uploaded by all people raising persons and respectively related to all indoor target positions;
step C4, according to each crowd funding person xiData quality relating to the target position respectively
Figure FDA00035121656900000215
In descending order of the average number of people xiSorting is carried out, and from the first, the order is selected before
Figure FDA00035121656900000216
The individual crowd funder forms each target crowd funder corresponding to the target position; wherein, a represents a preset screening proportion,
Figure FDA00035121656900000217
indicating rounding up.
2. The hybrid incentive method for crowd funding of indoor location information according to claim 1, wherein: said representation
Figure FDA0003512165690000031
And
Figure FDA0003512165690000032
the difference between them
Figure FDA0003512165690000033
To represent
Figure FDA0003512165690000034
And
Figure FDA0003512165690000035
the difference between them
Figure FDA0003512165690000036
And clustering centroid w and crowd funder xiDistance dist (w, x) between the two time series sensing data sequences based on intercepting preset same length time series sensing data sequencesi) The Euclidean distances are adopted.
3. The hybrid incentive method for crowd funding indoor location information according to claim 1 or 2, wherein: and C, based on the step C, after obtaining each target crowd funder corresponding to each target position, respectively aiming at each target position, according to the time sequence perception data sequence of each target crowd funder corresponding to the target position, respectively relating to the target position, constructing the marking information of the unique identifier corresponding to the target position, namely obtaining the marking information corresponding to each target position, and realizing the logic positioning of each indoor target position.
4. The hybrid incentive method for crowd funding indoor location information according to claim 1 or 2, wherein: the preset beacon sensing data type is wireless signal strength.
CN202110758771.7A 2021-07-05 2021-07-05 Hybrid excitation method for crowd funding of indoor position information Active CN113645564B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110758771.7A CN113645564B (en) 2021-07-05 2021-07-05 Hybrid excitation method for crowd funding of indoor position information

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110758771.7A CN113645564B (en) 2021-07-05 2021-07-05 Hybrid excitation method for crowd funding of indoor position information

Publications (2)

Publication Number Publication Date
CN113645564A CN113645564A (en) 2021-11-12
CN113645564B true CN113645564B (en) 2022-04-08

Family

ID=78416707

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110758771.7A Active CN113645564B (en) 2021-07-05 2021-07-05 Hybrid excitation method for crowd funding of indoor position information

Country Status (1)

Country Link
CN (1) CN113645564B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102946635A (en) * 2012-10-17 2013-02-27 上海京颐信息科技有限公司 Method and system of room-level indoor location based on exciter technology
CN108337656A (en) * 2018-01-16 2018-07-27 武汉工程大学 A kind of mobile intelligent perception motivational techniques

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8874477B2 (en) * 2005-10-04 2014-10-28 Steven Mark Hoffberg Multifactorial optimization system and method
US20150019310A1 (en) * 2013-07-15 2015-01-15 Jeffrey Young Electronic system and method for group purchasing promotions
CN112950251B (en) * 2021-05-11 2024-01-26 北京航空航天大学 Reputation-based vehicle crowd sensing node reverse combination auction excitation optimization method

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102946635A (en) * 2012-10-17 2013-02-27 上海京颐信息科技有限公司 Method and system of room-level indoor location based on exciter technology
CN108337656A (en) * 2018-01-16 2018-07-27 武汉工程大学 A kind of mobile intelligent perception motivational techniques

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
"Wireless Virtual Embedding Algorithm Considering Inter-cell Interference in 5G Ultra-dense Network";赵海涛;《IEEE》;20200813;全文 *
"一种基于信誉综合测评的移动群智感知激励机制";朱会;《科学技术创新》;20180305;全文 *
"业务迁移场景下利用信誉值的拓扑构造激励算法";张晖;《西安交通大学学报》;20170313;全文 *

Also Published As

Publication number Publication date
CN113645564A (en) 2021-11-12

Similar Documents

Publication Publication Date Title
CN108055119B (en) Safety excitation method and system based on block chain in crowd sensing application
CN110189174A (en) A kind of mobile intelligent perception motivational techniques based on quality of data perception
Xu et al. Incentive mechanism for multiple cooperative tasks with compatible users in mobile crowd sensing via online communities
Wachs et al. Social capital predicts corruption risk in towns
CN108364190B (en) Mobile crowd sensing online excitation method combined with reputation updating
KR101300517B1 (en) Optimum Tender Price prediction method and system
CN110009174A (en) Risk identification model training method, device and server
Pouryazdan et al. Game-theoretic recruitment of sensing service providers for trustworthy cloud-centric Internet-of-Things (IoT) applications
Nan et al. A cross-space, multi-interaction-based dynamic incentive mechanism for mobile crowd sensing
CN108648000B (en) Method and device for evaluating user retention life cycle and electronic equipment
CN112950251A (en) Reputation-based vehicle crowd sensing node reverse combination auction excitation optimization method
CN108764943B (en) Suspicious user monitoring and analyzing method based on fund transaction network
Wang et al. Promoting users’ participation in mobile crowdsourcing: A distributed truthful incentive mechanism (DTIM) approach
CN111626563B (en) Dual-target robust mobile crowd sensing system and excitation method thereof
Jia et al. An incentive mechanism in expert-decision-based crowdsensing networks
Hu et al. Incentive mechanism for mobile crowdsensing with two-stage stackelberg game
CN109325657A (en) A kind of measures and procedures for the examination and approval of credit operation, storage medium and server
Xiong et al. MAIM: A novel incentive mechanism based on multi-attribute user selection in mobile crowdsensing
Wang et al. TVD-RA: A truthful data value discovery based reverse auction incentive system for mobile crowd sensing
Schreiner Simple Poverty Scorecard Poverty-Assessment Tool: Cambodia
CN113645564B (en) Hybrid excitation method for crowd funding of indoor position information
Tang et al. A Semi-supervised Sensing Rate Learning based CMAB scheme to combat COVID-19 by trustful data collection in the crowd
CN117114178A (en) Electricity selling company optimization decision method and system
CN112907340B (en) Excitation method and system based on two-way auction model in crowd sensing
CN113139792B (en) Specific material collection method based on crowd sensing technology

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant