CN109257217B - Privacy protection-based Web service QoS prediction method under mobile edge environment - Google Patents

Privacy protection-based Web service QoS prediction method under mobile edge environment Download PDF

Info

Publication number
CN109257217B
CN109257217B CN201811091441.1A CN201811091441A CN109257217B CN 109257217 B CN109257217 B CN 109257217B CN 201811091441 A CN201811091441 A CN 201811091441A CN 109257217 B CN109257217 B CN 109257217B
Authority
CN
China
Prior art keywords
edge
user
qos
users
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
CN201811091441.1A
Other languages
Chinese (zh)
Other versions
CN109257217A (en
Inventor
张鹏程
金惠颖
王继民
贾旸旸
赵齐
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.)
Hohai University HHU
Original Assignee
Hohai University HHU
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 Hohai University HHU filed Critical Hohai University HHU
Priority to CN201811091441.1A priority Critical patent/CN109257217B/en
Publication of CN109257217A publication Critical patent/CN109257217A/en
Application granted granted Critical
Publication of CN109257217B publication Critical patent/CN109257217B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/147—Network analysis or design for predicting network behaviour
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/50—Network service management, e.g. ensuring proper service fulfilment according to agreements
    • H04L41/5003—Managing SLA; Interaction between SLA and QoS
    • H04L41/5009—Determining service level performance parameters or violations of service level contracts, e.g. violations of agreed response time or mean time between failures [MTBF]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/02—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/50—Network services
    • H04L67/52—Network services specially adapted for the location of the user terminal

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Environmental & Geological Engineering (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

The invention discloses a privacy protection-based QoS (quality of service) prediction method for Web services under a mobile edge environment, which collects user information and QoS attribute data of the Web services, performs data preprocessing on the data to obtain a data set conforming to the characteristics of the edge environment, and combines differential privacy to realize the QoS prediction method for protecting privacy under the edge environment. When data preprocessing is carried out, the data are divided into areas which are positioned in the same edge server, the same edge area and different edge areas according to the latitude and longitude of a user; processing the services in the divided edge server, deleting the services which have no attribute value and are greatly different from the attribute value at the same edge end, and obtaining a service set in the similar edge environment; in the prediction process, in order to protect the privacy of the user, Laplace noise is added into the original data set to obtain a disguised data set, and the disguised data set is used as historical data to be predicted.

Description

Privacy protection-based Web service QoS prediction method under mobile edge environment
Technical Field
The invention relates to a Web service QoS (quality of service) prediction method, in particular to a Web service QoS prediction method for protecting user privacy in a mobile edge environment, and belongs to the technical field of information.
Background
In recent years, with the rapid development of internet technology, many Web services having the same or similar functions have appeared. The QoS attributes of common Web services include: response time, throughput, cost, etc. In a real environment, QoS data often has a partial missing situation, which makes predicting the missing QoS according to the known QoS data an important issue in the service recommendation process, and the quality of service (QoS) attribute of the Web service becomes an important issue in the service selection process. In addition, mobile edge computing is an emerging technology nowadays, and due to the characteristics of short response time and high processing speed, in a service recommendation system, providing reliable service for mobile devices or mobile users by deploying an edge server becomes a new problem of predicting service quality in a new environment at present, and quality of service (QoS) becomes a key factor.
The existing prediction methods mainly aim at Web services in general environments, and mainly comprise a QoS prediction method based on trust perception, a Web service QoS prediction method based on temporal perception, a matrix factorization QoS prediction method based on context perception and a QoS prediction method based on privacy protection. The privacy protection-based prediction method mainly considers the confidentiality of original data, selectively adds a Laplace mechanism or predicts based on a position sensitive Hash function, for example, Liu and the like consider the privacy of users, adds Laplace noise to an original data set to carry out data camouflage, and predicts by utilizing a collaborative filtering method; qi and other privacy protection service recommendation methods using locality sensitive hashing obtain efficient trade-offs among service accuracy, privacy protection and high efficiency; zhang et al adopts an extensible big data multidimensional anonymization method based on a distributed computing system. However, the existing method cannot predict data with short timeliness and fast change under the mobile edge environment. Wang et al propose a service recommendation QoS prediction method based on collaborative filtering in a mobile edge environment, but this method directly predicts a user's data set as historical data, and does not consider privacy of the user.
With the development of technology, a great deal of service information of users is acquired in more and more prediction methods, and privacy protection becomes a great demand for users in the prediction process.
Disclosure of Invention
The purpose of the invention is as follows: the characteristics of short data timeliness and fast change in the mobile edge environment and the user requirements of privacy protection are considered. The invention provides a Web service QoS prediction method based on privacy protection in a mobile edge environment, wherein the chaos and instability of data can influence the QoS value prediction, so that the service attribute value is preprocessed according to the user geographical position information in a common data set to obtain the data meeting the edge end characteristics. Under the edge environment that the user has the characteristics of mobility and activity, the original data of the user is protected, and the privacy of the user is effectively protected while the rapid prediction is achieved.
The technical scheme is as follows: a Web service QoS prediction method based on privacy protection under a mobile edge environment comprises the following steps:
step 1: collecting user information and QoS attribute data of a Web service called by a user;
step 2: preprocessing the geographic position in the user information to construct an edge end environment;
and step 3: screening an edge service data set with the characteristics of a mobile edge environment from the preprocessed data;
and 4, step 4: setting privacy parameters, and adding Laplace noise to an original QoS data set;
and 5: searching for similar users by continuously increasing the distance by taking an edge server where the predicted user is located as a center;
step 6: and 5, obtaining the optimal distance by the minimum error of the MAE and the RMSE predicted values along with the distance change condition in the step 5, and recommending the Top-k similar users in the distance to realize the QoS attribute value prediction for protecting the privacy of the users in the marginal environment.
The data collection in the step 1 mainly comprises two aspects: (1) and collecting user information including the geographical position of the country, longitude, latitude value and the like from the wsdream. The wsdream is used as a channel for collecting user information and can provide reliability and quality evaluation of Web service. According to the obtained latitude and longitude information of the user, the user can be divided into several areas such as North America, Europe, Asia and the like. The user information is used to divide the location and area of the edge server in the mobile edge environment. The north american regions with the largest amount of data were selected to improve the sufficiency of the experiment. (2) QoS attribute data such as response time, throughput and the like of Web service called by a user are used as an original data set, the user after being divided into areas and the service called by the user at each place are in one-to-one correspondence by means of geographical position information of the user, attribute values of the service called by the user with the latitude and longitude difference within 10 degrees are obtained, the error ranges of the response time and the throughput attribute values are compared, and the processing of the following steps is carried out on the original data set. The QoS attribute data is the source of the service set that satisfies the edge environment characteristics.
The geographical position preprocessing mainly constructs an edge environment, and the step 2 further comprises:
step 21: the users are divided into continents according to the latitude and longitude values of the users. Latitude values (-86.9162,40.4249), (-122.2536,37.8668) in north america, (9.1833,48.7667), (-1.6743,48.112) in europe, and (114.1667,22.25), (139.69,35.69) in asia;
step 22: and selecting North America of China with the largest user information data quantity as an edge position point so as to reduce the prediction error. There are 87 location points with different latitude and longitude in north america, which are approximately regarded as 87 edge servers located at the edge end, wherein there are 60 location points with multiple users and 27 location points with only one user;
step 23: users with the same latitude and longitude are considered to be located in one edge server. Let North American user set U ═ U1,u2,...,unU, user u28,u29,u30Has latitude longitude value of (38, -97), user u130,u131,u132,u133,u134,u135The latitude longitude values of (37.8668, -122.2536) are respectively regarded as the same server in the same area of the edge end, and data in the same server can be used as a historical data set during prediction;
step 24: users whose latitude and longitude values are different within a certain range are approximately regarded as being located in the same area of the edge end. User U in north american user set U10Has a latitude and longitude of (39.1023, -96.5246), and the user u11Has a latitude longitude of (39.1668, -84.5382), and the latitude longitude difference is small, approximately regarded as a user in two edge servers located in the same area of the edge end;
step 25: users with large differences in latitude and longitude values are approximately regarded as being located in different areas of the edge end. User U in set U2Has a latitude and longitude of (44.5642, -123.2790), and the user u89Has a latitude and longitude of (33.7866, -118.2987), and the user u101Has a latitude longitude of (18.2500, -66.5000), the inter-user latitudes and longitudes are very different, approximately regarded as different users located in different areas of the marginal end.
Compared with the general environment, the edge environment has the characteristics of short time effect and quick response, so that the data set in the edge environment is more accurate. Screening out an edge service data set with a mobile edge environment characteristic, wherein the step 3 is further as follows:
step 31: services with response time and throughput attribute values of-1 in the original dataset are removed. The attribute value is-1, which indicates that the user does not call the service, and in order to ensure high call rate of the service and achieve the characteristics of high prediction speed and high accuracy at the edge end, a high-density user-service matrix set is obtained after the service set which is not called by the user is removed, so that the prediction efficiency is improved, and the error is reduced;
step 32: and determining an error range of the service called by the user in the same edge area, and removing the services with the response time error value of more than 2ms and the throughput error value of more than 20Mbps so as to meet the characteristics of the edge end service. The same area of the edge end has a more similar edge environment, therefore, users in different servers in the same edge area call the QoS value of the same service within a certain error range, wherein the response time error value is within 2ms and the throughput error value is within 20 Mbps. Let North American service set S ═ S1,s2,...,smE.g. service s48At user u11The QoS attribute value response time of the edge area is about 0.1, the throughput is about 2.0, the edge service data set is finally obtained after screening processing, and the response time-user matrix is as follows: 2354. the throughput-user matrix is: 2258*174.
In order to protect the privacy of the original data and the privacy of the user, the step 4 further includes:
step 41: and performing z-score processing on the screened edge service data set. The core idea of z-score is: z ═ x-mean (x) ·/std (x), where x is the column vector of the two-dimensional matrix response time and throughput, mean (x) is the mean of vector x, std (x) is the standard deviation of vector x. Performing z-score transformation on the response time attribute value rt under the edge environment to obtain a new matrix rt 1;
step 42: adding satisfies the Laplace function
Figure GDA0003151524230000041
Distributed noise. Wherein mu and b are respectively a position parameter and a scale parameter;
step 43: let μ be 0 or more,
Figure GDA0003151524230000042
wherein epsilon is a privacy parameter, and deltaf is the range (maximum value minus minimum value) of the column vector, obtaining a formula
Figure GDA0003151524230000043
Satisfying a symmetric exponential distribution;
step 44: and generating a random number x by using a rand function, and obtaining a random noise value by using a Laplace distribution function. Let the random number generated by the edge end response time matrix set column vector be x _ rt, let epsilon belong to the interval 0.5 to 5, step length be 0.5, continuously update parameter epsilon to obtain scale parameter b, substitute formula
Figure GDA0003151524230000044
Obtaining a Laplace random noise value L _ rt;
step 45: and adding the new data set rt1 after z-score transformation in the step 41 and the laplacian random noise value L _ rt obtained in the step 44 to obtain a new data set X _ rt after data disguise by using a difference privacy formula X- (g), (X) + Laplace (delta f/epsilon), wherein g (X) refers to an original data set, and Laplace (delta f/epsilon) refers to an obtained Laplacian random noise value.
In order to find suitable similar users, the step 5 further includes:
step 51: and continuously searching other edge servers outwards in a diffused mode by taking the edge server where the predicted user is located as the center of a circle and taking the distance as the radius. Let a certain predicted user UnLatitude longitude value of (alpha)1,β1) The edge server is b, and the latitude longitude value of other users to be searched is (alpha)2,β2) The edge server is biUsing a formula
Figure GDA0003151524230000051
In which theta is two-edge garmentCenter angle between servers, by distance formula
Figure GDA0003151524230000052
Get edge servers b and biThe distance of (2) is to make the point where the edge server b is located as the center of a circle and the distance as the radius r, and the distance continuously spreads outwards to cover other edge servers bi；
Step 52: and calculating the similarity between the covered users in the edge server and the predicted users, and performing QoS prediction based on the similarity. (1) The similarity calculation formula of two users u and v located in the edge server is as follows:
Figure GDA0003151524230000053
wherein, the service set S is a service commonly called by the users u and v, Si represents the service set to be predicted, ru,iIs the QoS value at which user u invokes service i,
Figure GDA0003151524230000054
is the average of all services invoked by user u, rv,iIs the QoS value at which user v invokes service i,
Figure GDA0003151524230000055
is the average of all services invoked by user v; (2) from the calculation formula of standard deviation
Figure GDA0003151524230000056
(cuThe number of the service i called for the user u) is obtained by substituting the number of the service i into the similarity calculation formula
Figure GDA0003151524230000057
(cvNumber of calls to service i for user v); (3) from the z-score normalized calculation formula, it is known
Figure GDA0003151524230000058
Substituting into simplified similarity calculation formula to obtain
Figure GDA0003151524230000059
(4) Let response time two-dimensional matrixes a and b, a ═ a respectively1,a2,…,an)，b＝(b1,b2,…,bn) Wherein a isn、bnThe column vectors are two-dimensional matrixes respectively, and after data camouflage, the two matrixes are changed into A ═ A (A)1,A2,…,An)，B＝(B1,B2,…,Bn)。
Figure GDA00031515242300000510
aiAnd Laplace (Δ f)b/εb) Are two independent vectors, and Laplace (Δ f)b/εb) If the symmetric exponential distribution is satisfied when the Laplace probability density function mu is 0, then Sigma aiLaplace(Δfb/εb)≈0，∑biLaplace(Δfa/εa)≈0,∑Laplace(Δfa/εa)Laplace(Δfb/εb) 0. available, AB ≈ Σ aibiAb; (5) substituting the result in (4) into the similarity formula in (3) to obtain the similarity
Figure GDA0003151524230000061
Qu,i、Qv,iThe disguised values respectively represent attribute values of the user u and the user v for calling the service i; (6) based on the similarity, the QoS value of service i invoked by user u can be predicted directly,
Figure GDA0003151524230000062
q`u,ithe predicted value is represented by a value of the prediction,
Figure GDA0003151524230000063
an average value of attribute value masquerading values representing the user u invoking the service; (7) using a formula
Figure GDA0003151524230000064
The root mean square error is calculated and,
Figure GDA0003151524230000065
calculating the average absolute error, wherein N is the number of the service sets i;
step 53: the number of Top-k similar users within the distance is counted according to the optimal distance worth the minimum error in step 52. And predicting the optimal distance of the edge server where the user is located to search for similar users according to the minimum error value of the RMSE and the MAE, and counting the number of the edge servers in the distance and the number of the users in each edge server to obtain the total number of Top-k similar users.
Aiming at the characteristics of strong user mobility and short QoS attribute value timeliness under the mobile edge environment, in the step 6, in the prediction process, the coverage range of the minimum error value is continuously searched according to the position of the edge server where the predicted user is located, and Top-k similar user recommendation is carried out within the optimal distance, so that the QoS attribute value prediction for protecting the user privacy under the edge environment is realized.
Has the advantages that: compared with the prior art, the method for predicting the Web service QoS based on privacy protection in the mobile edge environment improves the efficiency and accuracy of Web service QoS prediction in the mobile edge environment, considers rapid prediction and simultaneously protects the privacy of a user.
Drawings
FIG. 1 is an overall step diagram of an embodiment of the present invention;
FIG. 2 is a flow chart of a method according to an embodiment of the present invention.
Detailed Description
The present invention is further illustrated by the following examples, which are intended to be purely exemplary and are not intended to limit the scope of the invention, as various equivalent modifications of the invention will occur to those skilled in the art upon reading the present disclosure and fall within the scope of the appended claims.
As shown in fig. 1, the method for predicting the QoS of the Web service based on privacy protection in the mobile edge environment mainly includes 6 steps:
step 1: collecting user information and QoS attribute data of a user calling Web service in a general environment;
step 2: preprocessing the geographic position in the user information to construct an edge end environment;
and step 3: screening an edge service data set with the characteristics of a mobile edge environment from the preprocessed data;
and 4, step 4: setting privacy parameters, and adding Laplace noise to an original QoS data set;
and 5: searching for similar users by continuously increasing the distance by taking an edge server where the predicted user is located as a center;
step 6: and 5, obtaining the optimal distance from the minimum error in the step 5, and recommending the Top-k similar users in the distance to realize the QoS attribute value prediction for protecting the privacy of the users in the marginal environment.
As shown in fig. 2, the method for predicting the QoS of the Web service based on privacy protection in the mobile edge environment includes the following specific steps:
step 101: collecting user information and QoS attribute values under a general environment, wherein the QoS attribute values mainly comprise representative attributes such as response time, throughput and the like;
the specific steps of collecting information including the user geographical position and the like from the wsdream comprise:
step 102: collecting user information including geographical positions such as the country, longitude and latitude values according to the description of the dataset file in the wsdream;
step 103: dividing the obtained latitude and longitude information of the user into several areas such as North America, Europe, Asia and the like;
step 104: selecting the North American region with the largest data volume as a preprocessed experimental data set;
the method comprises the following steps of utilizing QoS attribute data such as response time, throughput and the like of Web service called by a user in a general environment as an original data set, and specifically comprising the following steps:
step 105: the geographic position information of the user is used for corresponding the geographic position information of the user to the service called by each place one by one;
step 106: and acquiring the attribute value of the service called by the similar user, comparing the error range and processing the service set.
The specific processing steps of the user geographical position preprocessing are as follows:
step 107: the users are divided into continents according to the latitude and longitude values of the users. Latitude values (-86.9162,40.4249), (-122.2536,37.8668) in north america, (9.1833,48.7667), (-1.6743,48.112) in europe, and (114.1667,22.25), (139.69,35.69) in asia;
step 108: and selecting North America of China with the largest user information data quantity as an edge position point. There are 87 location points with different latitude and longitude in north america, which are approximately regarded as 87 edge servers located at the edge end, wherein there are 60 location points with multiple users and 27 location points with only one user;
step 109: users with the same latitude and longitude are considered to be located in one edge server. Let North American user set U ═ U1,u2,...,uiWhere n is 1,228,u29,u30Has latitude longitude value of (38, -97), user u130,u131,u132,u133,u134,u135The latitude longitude values of (37.8668, -122.2536) are respectively regarded as the same server in the same area of the edge end, and data in the same server can be used as a historical data set during prediction;
step 110: users whose latitude and longitude values are different within a certain range are approximately regarded as being located in the same area of the edge end. User U in north american user set U10Has a latitude and longitude of (39.1023, -96.5246), and the user u11Has a latitude longitude of (39.1668, -84.5382), and the latitude longitude difference is small, approximately regarded as a user in two edge servers located in the same area of the edge end;
step 111: users with large differences in latitude and longitude values are approximately regarded as being located in different areas of the edge end. User U in set U2Has a latitude and longitude of (44.5642, -123.2790), and the user u89Has a latitude and longitude of (33.7866, -118.2987), and the user u101The latitude and longitude of (18.2500, -66.5000), the latitude and longitude between users are greatly different, and the user can be approximately regarded asDifferent users located in different areas of the edge terminal.
The specific screening steps of the edge service data set with the mobile edge environment characteristic are as follows:
step 112: and removing the service with the attribute value of-1 in the original data set. The attribute value is-1, which indicates that the user does not call the service, and in order to ensure high call rate of the service and achieve the characteristics of high prediction speed and high accuracy at the edge end, a high-density user-service matrix set is obtained after the service set which is not called by the user is removed, so that the prediction efficiency is improved, and the error is reduced;
step 113: and determining the error range of the service called by the user in the same edge area, and removing the service with a larger error value so as to meet the characteristics of the edge end service. The same area of the edge end has a more similar edge environment, so the QoS values of the same service invoked by users in different servers in the same edge area are also within a certain error range. Let North American service set S ═ S1,s2,...,smWhere m is 1,2, k (k is a positive integer), such as the service s48At user u11The QoS attribute value response time of the edge region is [0.1,0.3 ]]Throughput is [1.0,5.0 ]]After screening, the response time-user matrix size of the final edge service data set is: 2354. the throughput-user matrix is: 2258*174.
The specific steps of adding Laplace noise to the original data are as follows:
step 114: and performing z-score processing on the screened edge service data set. The core idea of z-score is: z ═ x-mean (x) ·/std (x), where x is a vector, mean (x) is the mean of vector x, std (x) is the standard deviation of vector x. Performing z-score transformation on the response time attribute value rt under the edge environment to obtain a new matrix rt 1;
step 115: adding satisfies the Laplace function
Figure GDA0003151524230000091
Distributed noise. Where μ and b are a position parameter and a scale parameter, respectively, and x refers to a random number that generates Laplace noise；
Step 116: let μ be 0 or more,
Figure GDA0003151524230000092
wherein epsilon is a privacy parameter, and deltaf is the range (maximum value minus minimum value) of the column vector, obtaining a formula
Figure GDA0003151524230000093
Satisfying a symmetric exponential distribution;
step 117: and generating a random number x by using a rand function, and obtaining a random noise value by using a Laplace distribution function. Let the random number generated by the edge response time matrix set column vector be x _ rt, and let epsilon belong to the interval [0.5,5 ]]Step length is 0.5, parameter epsilon is continuously updated to obtain scale parameter b, and the scale parameter b is substituted into the formula
Figure GDA0003151524230000094
Obtaining a Laplace random noise value L _ rt;
step 118: and adding the new data set rt1 after z-score transformation in the step 41 and the laplacian random noise value L _ rt obtained in the step 44 to obtain a new data set X _ rt after data disguise by using a difference privacy formula X- (g), (X) + Laplace (delta f/epsilon), wherein g (X) refers to an original data set, and Laplace (delta f/epsilon) refers to an obtained Laplacian random noise value.
The specific steps for searching similar users are as follows:
step 119: and continuously searching other edge servers outwards by taking the edge server where the predicted user is located as the circle center and the distance between the edge servers as the radius. Let a certain predicted user UnLatitude longitude value of (alpha)1,β1) The edge server is b, and the latitude longitude value of other users to be searched is (alpha)2,β2) The edge server is biUsing a formula
Figure GDA0003151524230000095
Where θ is the center angle between two edge servers, expressed by the distance
Figure GDA0003151524230000101
Get edge servers b and biThe distance of (2) is to make the point where the edge server b is located as the center of a circle and the distance as the radius r, and the distance continuously spreads outwards to cover other edge servers bi；
Step 120: and calculating the similarity between the covered users in the edge server and the predicted users, and performing QoS prediction based on the similarity. (1) The similarity calculation formula of two users u and v located in the edge server is as follows:
Figure GDA0003151524230000102
wherein, the service set S is a service commonly called by the users u and v, ru,iIs the QoS value at which user u invokes service i,
Figure GDA0003151524230000103
is the average of all services invoked by user u, rv,iIs the QoS value at which user v invokes service i,
Figure GDA0003151524230000104
is the average of all services invoked by user v; (2) from the calculation formula of standard deviation
Figure GDA0003151524230000105
(cuThe number of the service i called for the user u) is obtained by substituting the number of the service i into the similarity calculation formula
Figure GDA0003151524230000106
(cvNumber of calls to service i for user v); (3) from the z-score normalized calculation formula, it is known
Figure GDA0003151524230000107
Substituting into simplified similarity calculation formula to obtain
Figure GDA0003151524230000108
(4) Let two vectors a ═ a1，a2，…，an)，b＝(b1，b2，…，bn) After data masquerading, the two vectors become A ═ A1,A2,…,An)，B＝(B1,B2,…,Bn)。
Figure GDA0003151524230000109
aiAnd Laplace (Δ f)b/εb) Are two independent vectors, and Laplace (Δ f)b/εb) If the symmetric exponential distribution is satisfied when the Laplace probability density function mu is 0, then Sigma aiLaplace(Δfb/εb)≈0，∑biLaplace(Δfa/εa)≈0,∑Laplace(Δfa/εa)Laplace(Δfb/εb) 0. available, AB ≈ Σ aibiAb; (5) substituting the result in the step (4) into the similarity formula in the step (3) to obtain the similarity
Figure GDA0003151524230000111
(6) Based on the similarity, the QoS value of service i invoked by user u can be predicted directly,
Figure GDA0003151524230000112
q 'will be based on the mean and standard deviation of the predicted users in z-score normalization'u,iBefore the value returns to normalization; (7) using a formula
Figure GDA0003151524230000113
The root mean square error is calculated and,
Figure GDA0003151524230000114
calculating the average absolute error, wherein N is the number of the service sets i;
step 121: and counting the number of Top-k similar users within the optimal distance according to the minimum error value in the step 120. The minimum error value of RMSE and MAE is used for predicting the optimal distance of the edge server where the user is located to search for similar users, and the number of the edge servers in the distance and the number of the users in each edge server are counted to obtain the total number of Top-k similar users;
step 122: in the prediction process, the coverage range of the minimum error value is continuously searched according to the position of the edge server where the predicted user is located, and the Top-k similar user recommendation is carried out within the optimal distance.

Claims (4)

1. A method for predicting the QoS of Web service based on privacy protection in a mobile edge environment is characterized by comprising the following steps:
step 1: collecting user information and QoS attribute data of a user calling Web service;
step 2: preprocessing the geographic position in the user information to construct an edge terminal environment, wherein the step 2 comprises the following steps:
step 21: dividing the users into continents according to the latitude and longitude values of the users;
step 22: selecting a continent with the maximum user information data quantity as an edge position point;
step 23: regarding users with the same latitude and longitude as being located in an edge server, and taking data in the same server as a historical data set during prediction;
step 24: regarding users with the latitude and longitude value difference within a preset range as being located in the same area of the edge end;
step 25: regarding users with the difference between the latitude value and the longitude value larger than a preset range as different areas positioned at the edge end;
and step 3: screening an edge service data set with the characteristics of a mobile edge environment from the preprocessed data;
and 4, step 4: setting a privacy parameter epsilon, and adding Laplace noise to an original QoS data set, wherein the step 4 comprises the following steps:
step 41: performing z-score processing on the screened edge service data set;
step 42: adding satisfies the Laplace function
Figure FDA0003070542480000011
Distributed noise of which muAnd b are a position parameter and a scale parameter, respectively;
step 43: order to
Figure FDA0003070542480000012
Wherein epsilon is a privacy parameter and deltaf is the range of the column vector to obtain a formula
Figure FDA0003070542480000013
Satisfying a symmetric exponential distribution;
step 44: generating a random number x by using a rand function, and obtaining a random noise value by using a Laplace distribution function;
step 45: adding the new data set after z-score transformation in step 41 and the Laplace random noise value obtained in step 44 to obtain a new data set after data camouflage by using a difference privacy formula X ═ g (X) + Laplace (delta f/epsilon), wherein g (X) refers to an original data set, and Laplace (delta f/epsilon) refers to an obtained Laplace random noise value;
and 5: searching for similar users by continuously increasing the distance by taking an edge server where the predicted user is located as a center;
step 6: and 5, obtaining the optimal distance from the minimum error in the step 5, and recommending the Top-k similar users in the distance to realize the QoS attribute value prediction for protecting the privacy of the users in the marginal environment.
2. The method for predicting QoS of Web service based on privacy protection in mobile edge environment according to claim 1, wherein the data collection in step 1 mainly includes two aspects: (1) collecting user information of the geographical position containing the country, longitude and latitude values from the wsdream; (2) the QoS attribute data of the Web service invoked by the user is taken as the raw data set.
3. The method for predicting QoS of Web service based on privacy protection in mobile edge environment according to claim 1, wherein the step 3 comprises the following steps:
step 31: removing services with the response time and the throughput attribute value of-1 in the original data set;
step 32: and determining an error range of the service called by the user in the same edge area, and removing the services with the response time error value of more than 2ms and the throughput error value of more than 20Mbps so as to meet the characteristics of the edge end service.
4. The method for predicting QoS of Web service based on privacy protection in mobile edge environment according to claim 1, wherein the step 5 comprises the following steps:
step 51: continuously searching other edge servers by outward diffusion by taking the edge server where the user is predicted as the center of a circle and taking the distance as the radius;
step 52: calculating the similarity between the covered users in the edge server and the predicted users, and performing QoS prediction based on the similarity;
step 53: the number of Top-k similar users within the distance is counted according to the optimal distance worth the minimum error in step 52.
CN201811091441.1A 2018-09-19 2018-09-19 Privacy protection-based Web service QoS prediction method under mobile edge environment Active CN109257217B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811091441.1A CN109257217B (en) 2018-09-19 2018-09-19 Privacy protection-based Web service QoS prediction method under mobile edge environment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811091441.1A CN109257217B (en) 2018-09-19 2018-09-19 Privacy protection-based Web service QoS prediction method under mobile edge environment

Publications (2)

Publication Number Publication Date
CN109257217A CN109257217A (en) 2019-01-22
CN109257217B true CN109257217B (en) 2021-08-10

Family

ID=65047505

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811091441.1A Active CN109257217B (en) 2018-09-19 2018-09-19 Privacy protection-based Web service QoS prediction method under mobile edge environment

Country Status (1)

Country Link
CN (1) CN109257217B (en)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110417867B (en) * 2019-07-03 2020-06-16 河海大学 Web service QoS monitoring method under mobile edge environment
CN111131424B (en) * 2019-12-18 2020-12-18 武汉大学 Service quality prediction method based on combination of EMD and multivariate LSTM
CN111327480B (en) * 2020-03-02 2021-04-27 河海大学 Method for monitoring multiple QoS of Web service under mobile edge environment
CN112700067B (en) * 2021-01-14 2024-05-24 安徽师范大学 Method and system for predicting service quality in unreliable mobile edge environment
CN113099387B (en) * 2021-03-22 2022-02-22 河海大学 Method for constructing service quality data set in real edge environment
CN113225209B (en) * 2021-04-20 2022-03-18 河海大学 Network monitoring real-time early warning method based on time series similarity retrieval
CN114466326B (en) * 2021-12-31 2024-05-28 河海大学 Mobile terminal position privacy protection method for marine observation monitoring sensor network
CN116595254B (en) * 2023-05-18 2023-12-12 杭州绿城信息技术有限公司 Data privacy and service recommendation method in smart city

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2012050913A1 (en) * 2010-09-28 2012-04-19 The Ohio State University Predictive network system and method
CN105791392A (en) * 2016-02-16 2016-07-20 中国科学院信息工程研究所 Mobile network based edge service communication method and system
CN107609421A (en) * 2017-09-25 2018-01-19 深圳大学 Secret protection cooperates with the collaborative filtering method based on neighborhood of Web service prediction of quality
CN107679415A (en) * 2017-09-25 2018-02-09 深圳大学 Secret protection cooperates with the collaborative filtering method based on model of Web service prediction of quality
CN107659444A (en) * 2017-09-25 2018-02-02 深圳大学 Secret protection cooperates with the difference privacy forecasting system and method for Web service quality

Also Published As

Publication number Publication date
CN109257217A (en) 2019-01-22

Similar Documents

Publication Publication Date Title
CN109257217B (en) Privacy protection-based Web service QoS prediction method under mobile edge environment
Xiao et al. Loclok: Location cloaking with differential privacy via hidden markov model
Wang et al. Protecting personal trajectories of social media users through differential privacy
Al-Dhubhani et al. An adaptive geo-indistinguishability mechanism for continuous LBS queries
Gao et al. DPLCF: differentially private local collaborative filtering
Gu et al. Supporting both range queries and frequency estimation with local differential privacy
CN105992338B (en) Positioning method and device
US9159030B1 (en) Refining location detection from a query stream
Yang et al. Location privacy preservation mechanism for location-based service with incomplete location data
CN110062324B (en) Personalized position privacy protection method based on k-anonymity
CN109977324B (en) Interest point mining method and system
CN111093191B (en) Crowd sensing position data issuing method based on differential privacy
Ghinita Private queries and trajectory anonymization: a dual perspective on location privacy
US10219115B2 (en) Facilitation of mobile device geolocation
CN112818216B (en) Client recommendation method and device, electronic equipment and storage medium
US20210400615A1 (en) Wi-fi z-axis positioning
CN106465175A (en) Method for collecting and aggregating network quality data
Budgaga et al. A framework for scalable real‐time anomaly detection over voluminous, geospatial data streams
Cunningham et al. Privacy-preserving synthetic location data in the real world
Li et al. DALP: A demand‐aware location privacy protection scheme in continuous location‐based services
CN110298687B (en) Regional attraction assessment method and device
Ma et al. Data level privacy preserving: A stochastic perturbation approach based on differential privacy
WO2020223908A1 (en) Privacy management
Huang et al. A differential private mechanism to protect trajectory privacy in mobile crowd-sensing
CN111861526A (en) Method and device for analyzing object source

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