CN110266539B - Internet of things service QoS prediction method based on collaborative filtering - Google Patents

Internet of things service QoS prediction method based on collaborative filtering Download PDF

Info

Publication number
CN110266539B
CN110266539B CN201910550459.1A CN201910550459A CN110266539B CN 110266539 B CN110266539 B CN 110266539B CN 201910550459 A CN201910550459 A CN 201910550459A CN 110266539 B CN110266539 B CN 110266539B
Authority
CN
China
Prior art keywords
service
user
qos
similarity
time
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
CN201910550459.1A
Other languages
Chinese (zh)
Other versions
CN110266539A (en
Inventor
黄学臻
张琦
陈彦如
童恩栋
李秋香
吴薇
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.)
First Research Institute of Ministry of Public Security
Original Assignee
First Research Institute of Ministry of Public Security
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 First Research Institute of Ministry of Public Security filed Critical First Research Institute of Ministry of Public Security
Priority to CN201910550459.1A priority Critical patent/CN110266539B/en
Publication of CN110266539A publication Critical patent/CN110266539A/en
Application granted granted Critical
Publication of CN110266539B publication Critical patent/CN110266539B/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
    • 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
    • H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention discloses an internet of things service QoS prediction method based on collaborative filtering1，T2，…，Tk) And the QoS data respectively calculates the similarity of the user and the service, and further obtains the final user and service similarity by combining the k groups of similarities. And selecting the most similar TopN users and services, fusing the user-based and service-based collaborative prediction methods, and finally predicting to obtain the missing QoS. The method of the invention is based on the CF algorithm, and introduces a time perception mechanism, thereby realizing QoS prediction.

Description

Internet of things service QoS prediction method based on collaborative filtering
Technical Field
The invention relates to the technical field of Internet of things, in particular to a collaborative filtering-based Internet of things service QoS prediction method.
Background
The Internet of Things (IoT) is subject to scale development, and the network complexity is inevitably increased dramatically. Service-oriented Architecture (SOA) provides an ideal solution for the construction and application integration of such complex systems. The SOA is a method for modularizing business logic of applications into different functional units (called services) and connecting the services through well-defined interfaces and contracts. The service-oriented architecture changes the cooperation mode among the applications of the traditional Internet of things, so that different applications can share resources through service discovery and combination.
In an actual service-oriented architecture environment of the internet of things, a plurality of service providers exist to provide services to the outside. Thus, there may be multiple services implemented for the same function. This raises the problem of service selection, i.e. how to select the best service out of a set of alternative services. By introducing a non-functional description, Quality of Service (QoS), a set of functionally identical services can be distinguished, i.e. a set of functionally identical services have different QoS attributes, such as Service price, Service response time, Service reliability, etc. The service requester has not only its functional requirements but also its QoS requirements when requesting services, and the purpose of service selection is to select the optimal service that satisfies both the requester's functional requirements and QoS requirements.
However, it is possible that the same user invokes the same service at different times, and the QoS achieved by different users invoking the same service is almost always different. Therefore, the QoS of each service needs to be calculated separately for any user. In practice, the user invokes only a small fraction of the services, and many services have missing QoS data that needs to be predicted. The missing QoS prediction generally has three methods of matrix decomposition, time series prediction and collaborative filtering. Among them, the collaborative filtering is widely used due to its good performance. Jiang et al indicate that QoS attributes that do not vary much among different users should be given less weight in calculating user similarity. Liu et al found that similar users obtained with the conventional collaborative filtering algorithm were not similar, but only because they coincidentally obtained similar QoS during some service invocation. Therefore, they consider the location information of users and services, and improve the collaborative filtering algorithm. Ma et al consider two users similar, whose similarity remains unchanged before and after a service invocation. Thereby obtaining the missing QoS by solving a quadratic equation. However, network dynamics is high in the scene of the internet of things, QoS information has high timeliness, historical QoS values are simply averaged in the prior art, and time attributes of QoS fine granularity are not considered, so that QoS prediction accuracy is low.
Reference to the literature
[1]Y.Hu,Q.Peng,X.Hu,and R.Yang,“Web service recommendation based on time series forecasting and collaborative filtering,”in Proceedings of IEEE International Conference on Web Services.IEEE,2015,pp.233–240.
[2]Y.Koren,“Factor in the neighbors:Scalable and accurate collaborative filtering,”ACM Transactions on Knowledge Discovery from Data,vol.4,no.1,pp.1–24,2010.
[3]G.Adomavicius and A.Tuzhilin,“Toward the next generation of recommender systems:A survey of the state-of-the-art and possible extensions,”IEEE Transactions on Knowledge and Data Engineering,vol.17,no.6,pp.734–749,2005.
[4]J.S.Breese,D.Heckerman,and C.Kadie,“Empirical analysis of predictive algorithms for collaborative filtering,”in Proceedings of International Conference on Uncertainty in Artificial Intelligence.ACM,1998,pp.43–52.
[5]W.Lo,J.Yin,Y.Li,and Z.Wu,“Efficient web service qos prediction using local neighborhood matrix factorization,”Engineering Applications of Artificial Intelligence,vol.38,pp.14–23,2015.
[6]G.Xue,C.Lin,Q.Yang,W.Xi,H.Zeng,Y.Yu,and Z.Chen,“Scalable collaborative filtering using cluster-based smoothing,”in Proceedings of ACM Sigir Conference on Information Retrieval.ACM,2005,pp.114–121.
[7]B.Sarwar,G.Karypis,J.Konstan,and J.Riedl,“Item-based collaborative filtering recommendation algorithms,”in Proceedings of IEEE International Conference on World Wide Web.IEEE,2001,pp.285–295.
[8]J.Wang,A.P.Vries,and M.J.Reinders,“Unifying user-based and item-based collaborative filtering approaches by similarity fusion,”in Proceedings of ACM Sigir Conference on Information Retrieval.ACM,2006,pp.501–508.
[9]“Pearson correlation coefficient,”https://en.wikipedia.org/wiki/Pearson correlation coefficient,2018.
[10]Y.Jiang,J.Liu,M.Tang,and X.Liu,“An effective web service recommendation method based on personalized collaborative filtering,”in Proceedings of IEEE International Conference on Web Services,2011,pp.211–218.
[11]J.Wu,L.Chen,Y.Feng,Z.Zheng,M.C.Zhou,and Z.Wu,“Predicting quality of service for selection by neighborhood-based collaborative filtering,”IEEE Transactions on Systems,Man,and Cybernetics:Systems,vol.43,no.2,pp.428–439,2013.
[12]J.Liu,M.Tang,Z.Zheng,X.F.Liu,and S.Lyu,“Locationaware and personalized collaborative filtering for web service recommendation,”IEEE Transactions on Services Computing,vol.9,no.5,pp.686–699,2016.
[13]Y.Ma,S.Wang,P.C.K.Hung,C.Hsu,Q.Sun,and F.Yang,“A highly accurate prediction algorithm for unknown web service qos values,”IEEE Transactions on Services Computing,vol.9,no.4,pp.511–523,2016.
[14]B.Cavallo,M.D.Penta,and G.Canfora,“An empirical comparison of methods to support qos-aware service selection,”in Proceedings of International Workshop on Principles of Engineering Service-Oriented Systems,2010,pp.64–70.
[15]C.Wu,W.Qiu,X.Wang,Z.Zheng,and X.Yang,“Time-aware and sparsity-tolerant qos prediction based on collaborative filtering,”in Proceedings of IEEE International Conference on Web Services.IEEE,2016,pp.637–640.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide the method for predicting the QoS of the service of the Internet of things based on collaborative filtering, which can improve the QoS prediction accuracy.
In order to achieve the purpose, the invention adopts the following technical scheme:
an internet of things service QoS prediction method based on collaborative filtering comprises the following steps:
s1, filtering to obtain QoS data of each service called by each user in the latest T time;
s2, dividing time T into k groups T1，T2，…，T k1 st group time T1Closest to the current time; the time span of each group is T/k, each element in the QoS matrix is the average value of all QoS data of a certain service called by a certain user in each group of time;
s3, predicting T based on QoS matrix in latest T time1Missing QoS value of time：
S3.1, QoS prediction based on users:
3.1.1) respectively calculating the similarity between the target user and each other user based on the QoS data in each group of time;
3.1.2) for the similarity between the target user and each other user, respectively calculating as follows:
calculating QoS data based on k groups of time to obtain similarity distribution weights among k users, and calculating comprehensive inter-user similarity between a target user and a corresponding user based on the distributed weights;
3.1.3) based on the calculated comprehensive inter-user similarity between the target user and other users, selecting the first N users with the maximum similarity with the target user, and further obtaining a QoS predicted value based on the users;
s3.2, QoS prediction based on service:
3.2.1) respectively calculating the service similarity of the target service and other services based on the QoS data of each group of time;
3.2.2) for the similarity between the target service and each service, respectively calculating as follows:
distributing weight values for the similarity among k services obtained by calculating QoS data based on k groups of time, and calculating to obtain the comprehensive similarity among the services between a target service and the corresponding service based on the distributed weight values;
3.2.3) calculate service-based QoS prediction:
selecting the target service s based on the similarity between the target service and other serviceseThe first M services with the maximum similarity further obtain a QoS predicted value based on the services;
and S3.3, integrating the QoS predicted value based on the user calculated in the step S3.1 and the QoS predicted value based on the service calculated in the step S3.2, and calculating the QoS predicted value based on the collaborative filtering.
Further, in step S1, the filtering calculation formula is as follows:
r(ui,sj)＝{r(ui,sj,t)|tcurrent-T≤t≤tcurrent}，i＝1,2,...,m，j＝1,2,...,n；
wherein r (u)i,sjT) represents user uiSelecting service s at a certain time T within T timejQoS data of r (u)i,sj) Representing user uiSelecting service s during time TjAll QoS data of (a); t is tcurrentRepresenting the current time, m is the number of users and n is the number of services.
Further, in step S2, the calculation formula of each element of the QoS matrix is as follows:
Figure BDA0002105349590000071
wherein N isθIs the number of QoS data in the theta group time, tbAnd tuTime of day values representing the two endpoints of the theta set of times.
Further, the specific method of step 3.1.1) is as follows:
similarity between the target user and each of the other users is performed according to the following formula:
Figure BDA0002105349590000072
1,2,., m and e ≠ v;
wherein u iseIs a target user, uvIn order to be the other users, the user can select the user,
Figure BDA0002105349590000073
representing users u calculated based on QoS data in the theta-th group of timeseWith user uvThe similarity of (2); sθ(ue)、Sθ(uv) Respectively represent users ueAnd user uvA set of services invoked during the theta set of times; r'θ(ue,sj) Representing a target user ueInvoking service s within the theta set of timesjAverage of all QoS data of r'θ(uv,sj) Individual user uvInvoking service s within the theta set of timesjAverage of all QoS data of (a);
Figure BDA0002105349590000081
respectively represent users uv、ueSet S of services invoked during the theta set of timesθ(ue)∩Sθ(uv) An average value of the corresponding QoS data;
Figure BDA0002105349590000082
Figure BDA0002105349590000083
N(Sθ(ue)∩Sθ(uv) Is S)θ(ue)∩Sθ(uv) The number of services contained therein.
Further, in step 3.1.2), the method for calculating the weight of the inter-user similarity obtained by calculating the QoS data in the θ -th group of time includes:
Figure BDA0002105349590000084
delta is a parameter;
and obtaining the comprehensive inter-user similarity between the target user and each user based on the weight obtained by calculation:
Figure BDA0002105349590000085
1,2,., m and e ≠ v.
Further, in step 3.1.3), the QoS prediction value based on the user is calculated as follows:
Figure BDA0002105349590000086
wherein s iseServing the target; SU (u)e) Representation and user ueAnd the top N users with the maximum similarity.
Further, in step 3.2.1), a specific calculation procedure for calculating the service similarity between the target service and each other service based on the QoS data of each group of time is as follows:
Figure BDA0002105349590000091
1,2, n and e ≠ f;
wherein s iseFor the target service, sfFor the purpose of other services, it is known to provide,
Figure BDA0002105349590000092
representing a target service s calculated based on QoS data for a theta set of timeseAnd service sfThe similarity of (2); r'θ(ui,se) Representing user uiInvoking service s within the theta set of timeseAverage of all QoS data of r'θ(ui,sf) Representing user uiInvoking service s within the theta set of timesfAverage of all QoS data of (a); u shapeθ(se) And Uθ(sf) Respectively representing the invocation of service s within the theta set of timeseAnd sfA set of users of (1);
Figure BDA0002105349590000093
respectively representing the user sets U in the theta group timeθ(se)∩Uθ(sf) Invoking a service sf、seAverage value of QoS data of time:
Figure BDA0002105349590000094
Figure BDA0002105349590000095
N(Uθ(se)∩Uθ(sf) Is a set Uθ(se)∩Uθ(sf) The number of users in (1).
Further, in step 3.2.2), different weights are assigned to the k inter-service similarity degrees calculated according to the following formula:
Figure BDA0002105349590000096
delta is a parameter;
based on the distributed weight, the similarity between the target service and each other service is obtained as follows:
Figure BDA0002105349590000101
1,2, n and e ≠ f.
Further, in step 3.2.3), the QoS prediction based on service is calculated as follows:
Figure BDA0002105349590000102
wherein, SS(s)e) Representation and service seThe set of top M services with the greatest similarity.
Further, the specific calculation of step S3.3 is performed as follows:
Figure BDA0002105349590000103
wherein the content of the first and second substances,
Figure BDA0002105349590000104
represents a QoS prediction value based on collaborative filtering,
Figure BDA0002105349590000105
indicating a user-based QoS prediction value,
Figure BDA0002105349590000106
the lambda is a parameter factor, and the lambda is more than or equal to 0 and less than or equal to 1.
The invention has the beneficial effects that: the invention considers the dynamic characteristic of QoS, introduces a time perception mechanism and obtains really similar users or services based on collaborative filtering, thereby improving the QoS prediction accuracy.
Drawings
FIG. 1 is a schematic flow chart of a method of example 1 of the present invention;
FIG. 2 is a schematic diagram of a QoS matrix in embodiment 1 of the present invention;
fig. 3 is a schematic diagram of a network topology according to embodiment 2 of the present invention.
Detailed Description
The present invention will be further described with reference to the accompanying drawings, and it should be noted that the present embodiment is based on the technical solution, and the detailed implementation and the specific operation process are provided, but the protection scope of the present invention is not limited to the present embodiment.
Example 1
The embodiment provides an internet of things service QoS prediction method based on collaborative filtering, as shown in FIG. 1, including the following steps:
s1, in order to ensure the timeliness of QoS data used in QoS prediction, firstly filtering to obtain QoS data of each user in the latest T time; the filter calculation formula is as follows:
r(ui,sj)＝{r(ui,sj,t)|tcurrent-T≤t≤tcurrent}，i＝1,2,...,m，j＝1,2,...,n；
wherein r (u)i,sjT) represents user uiSelecting service s at a certain time T within T timejQoS data of r (u)i,sj) Representing user uiSelecting service s during time TjAll QoS data of (a); t is tcurrentRepresenting the current time, wherein m is the number of users and n is the number of services;
s2, dividing time T into k groups T1，T2，…，T k1 st group time T1Closest to the current time; the time span of each group is T/k, and each element in the QoS matrix is the average value of QoS data in each group of time; the calculation formula is as follows:
Figure BDA0002105349590000111
wherein N isθIs the number of QoS data in the theta group time, tbAnd tuTime of day values representing the two endpoints of the theta set of times.
S3, as shown in FIG. 2, the QoS prediction is to predict the time T based on the QoS matrix in the latest T time1The missing QoS value of (2).
The QoS prediction method proposed in this embodiment first targets k groups of time periods (T)1，T2，…，Tk) The QoS data of (2) respectively calculates the similarity of the user and the service, and further obtains the final user and service similarity by combining the k groups of similarities. And selecting the first N users and services which are most similar to the target user, fusing the cooperative prediction methods based on the users and the services, and finally predicting to obtain the missing QoS. CF (Collaborative Filtering) is simply to use the preferences of groups with mutual interests and common experiences to infer the information of interest to users, and due to the good speed and robustness of CF, CF is widely used. The method of the embodiment is based on a CF algorithm, and introduces a time perception mechanism, thereby realizing QoS prediction.
S3.1 user-based QoS prediction
3.1.1) similarity calculation of groups
The historical QoS data are correspondingly divided into k groups, and the similarity among users is respectively calculated based on the QoS data in each group of time:
Figure BDA0002105349590000121
1,2,.. m and e ≠ v
Wherein u iseIn the case of the target user,
Figure BDA0002105349590000122
representing users u calculated based on QoS data in the theta-th group of timeseWith user uvThe similarity of (2); sθ(ue)、Sθ(uv) Respectively represent users ueAnd user uvA set of services invoked during the theta set of times;
Figure BDA0002105349590000123
respectively represent users uv、ueSet S of services invoked during the theta set of timesθ(ue)∩Sθ(uv) An average value of the corresponding QoS data;
Figure BDA0002105349590000124
Figure BDA0002105349590000125
N(Sθ(ue)∩Sθ(uv) Is S)θ(ue)∩Sθ(uv) The number of services contained therein.
3.1.2) comprehensive inter-user similarity calculation
The inter-user similarity calculated based on the QoS data of different groups (time periods) has different degrees of reliability, and for this reason, different weights need to be assigned to the calculated inter-k user similarity. The calculation of the weight takes into account two factors: (1) the user similarity calculated by the group closer to the current time has higher reliability; (2) the more the number of the services commonly called among the users is, the higher the reliability of the similarity among the users is. The weight calculation method of the similarity between the users obtained by calculating the QoS data in the theta group time comprises the following steps:
Figure BDA0002105349590000131
δ is a parameter and the default value is 2.
And obtaining the similarity between the comprehensive users based on the weight obtained by calculation:
Figure BDA0002105349590000132
1,2,., m and e ≠ v;
3.1.3) QoS prediction
And based on the calculated similarity between the target user and each user, selecting the first N users with the maximum similarity with the target user, and further obtaining a QoS predicted value based on the users:
Figure BDA0002105349590000133
wherein s iseServing the target; SU (u)e) Representation and user ueThe front N user sets with the maximum similarity;
s3.2 QoS prediction based on service
3.2.1) similarity calculation between services:
respectively calculating the similarity between services based on the QoS data of each group of time:
Figure BDA0002105349590000141
1,2, n and e ≠ f;
wherein the content of the first and second substances,
Figure BDA0002105349590000142
representing a target service s calculated based on QoS data for a theta set of timeseAnd service sfThe similarity of (2); u shapeθ(se) Indicating invocation of service s within the theta set of timeseUser set of Uθ(sf) Indicating invocation of service s within the theta set of timesfA set of users of (1);
Figure BDA0002105349590000143
respectively representing the user sets U in the theta group timeθ(se)∩Uθ(sf) Invoking a service sf、seAverage value of QoS data of time:
Figure BDA0002105349590000144
Figure BDA0002105349590000145
N(Uθ(se)∩Uθ(sf) Is a set Uθ(se)∩Uθ(sf) The number of users in (1);
3.2.2) comprehensive similarity calculation
The inter-service similarity calculated based on the QoS data of different groups (time periods) has different degrees of reliability, and for this reason, different weights need to be assigned to the calculated k inter-service similarities:
Figure BDA0002105349590000146
δ is a parameter and the default value is 2.
And obtaining the similarity between the comprehensive services based on the weight obtained by calculation:
Figure BDA0002105349590000147
1,2, n and e ≠ f;
3.2.3) QoS prediction:
target user obtained based on calculationThe comprehensive service similarity with each user, and the selection and target service seAnd further obtaining QoS predicted values based on the services for the first M services with the maximum similarity:
Figure BDA0002105349590000151
wherein, SS(s)e) Representation and service seA set of top M services with the greatest similarity;
s3.3, calculating the QoS prediction based on collaborative filtering:
Figure BDA0002105349590000152
wherein, λ is a parameter factor, λ is more than or equal to 0 and less than or equal to 1, λ is obtained by training and fitting according to actual data, and λ is more than 0.5 when the number of users is large and the number of services is small; λ <0.5 when the number of users is small and the number of services is small.
Example 2
In this embodiment, by taking the service of the internet of things as an example to compare the existing method with the method described in embodiment 1, a network topology used in this embodiment is shown in fig. 3, and a QoS matrix is shown in table 1.
TABLE 1
Figure BDA0002105349590000153
Figure BDA0002105349590000161
As shown in Table 1, there is a total of U1，U2，U3Three users and S1，S2，S3And the data in the table indicates that the corresponding user invokes the QoS of the corresponding service, and/indicates that the corresponding user does not invoke the corresponding service in the corresponding time period. T is1Is the most recent time group, T2Is a group of a long time period,T3is the farthest time group.
1. In the existing CF-based approach, QoS prediction is as follows:
at T3Time, user U1，U2，U3The QoS set of (A) is as follows:
R(U1)＝{100,80,30}
R(U2)＝{100,78,32}
R(U3)＝{70,50,60}
user U1And user U2The similarity of (A) is as follows:
Figure BDA0002105349590000162
user U1And user U3The similarity of (A) is as follows:
Figure BDA0002105349590000171
at time T2, router R1 was congested, resulting in an increase in service response time through R1 of 30, and at time T3, router R1 continued to remain congested, resulting in an increase in service response time through R1 of 60.
At T1Time, user U1，U2，U3The QoS set of (c) is changed to:
R(U1)＝{160,80,30}
R(U2)＝{100,138,32}
R(U3)＝{130,80,60}
user U1And user U2The similarity of (A) is as follows:
Figure BDA0002105349590000172
user U1And user U3The similarity of (A) is as follows:
Figure BDA0002105349590000173
as shown in FIG. 3, user U1，U2In the same local area network, the similarity should be higher, but the existing method does not consider the time attribute, and the calculated U is obtained1And U2Low degree of similarity, and U1And U3The similarity of the images is greatly increased, which is not in accordance with the reality.
2. The embodiment 1 method considers the time attribute, and the QoS prediction method is as follows:
at T3Time, user U1And user U2At T3Has a similarity of 0.999, user U1And user U3At T3The similarity of (c) was 0.229.
At T2Time, user U1，U2，U3The QoS set of (A) is as follows:
R(U1)＝{130,×,30}
R(U2)＝{×,108,32}
R(U3)＝{×,80,60}
user U1And user U2At T2Has a similarity of 0.5, user U1And user U3At T2The similarity of (a) was 0.5.
At T1Time, user U1，U2，U3The QoS set of (A) is as follows:
R(U1)＝{160,×,30}
R(U2)＝{×,138,32}
R(U3)＝{130,×,60}
user U1And user U2At T1Has a similarity of 0.5, user U1And user U3At T1The similarity of (a) was 0.5.
Finally, user U1And user U2The weighted similarity of (a) is:
Sim(U1,U2)＝0.999*0.73+0.5*0.15+0.5*0.12＝0.864
user U1And user U3The weighted similarity of (a) is:
Sim(U1,U3)＝0.229*0.73+0.5*0.15+0.5*0.12＝0.302
the final result is more practical.
Various corresponding changes and modifications can be made by those skilled in the art based on the above technical solutions and concepts, and all such changes and modifications should be included in the protection scope of the present invention.

Claims (6)

1. An internet of things service QoS prediction method based on collaborative filtering is characterized by comprising the following steps:
s1, filtering to obtain QoS data of each service called by each user in the latest T time;
s2, dividing time T into k groups T1，T2，…，Tk1 st group time T1Closest to the current time; the time span of each group is T/k, each element in the QoS matrix is the average value of all QoS data of a certain service called by a certain user in each group of time;
s3, predicting T based on QoS matrix in latest T time1Missing QoS value of time:
s3.1, QoS prediction based on users:
3.1.1) respectively calculating the similarity between the target user and each other user based on the QoS data in each group of time;
similarity between the target user and each of the other users is performed according to the following formula:
Figure FDA0003425752630000011
1,2,., m and e ≠ v;
wherein u iseIs a target user, uvIn order to be the other users, the user can select the user,
Figure FDA0003425752630000012
representing users u calculated based on QoS data in the theta-th group of timeseWith user uvThe similarity of (2); sθ(ue)、Sθ(uv) Respectively represent users ueAnd user uvA set of services invoked during the theta set of times; r'θ(ue,sj) Representing a target user ueInvoking service s within the theta set of timesjAverage of all QoS data of r'θ(uv,sj) Individual user uvInvoking service s within the theta set of timesjAverage of all QoS data of (a);
Figure FDA0003425752630000013
respectively represent users uv、ueSet S of services invoked during the theta set of timesθ(ue)∩Sθ(uv) An average value of the corresponding QoS data;
Figure FDA0003425752630000021
Figure FDA0003425752630000022
N(Sθ(ue)∩Sθ(uv) Is S)θ(ue)∩Sθ(uv) The number of services contained therein;
3.1.2) for the similarity between the target user and each other user, respectively calculating as follows:
calculating QoS data based on k groups of time to obtain similarity distribution weights among k users, and calculating comprehensive inter-user similarity between a target user and a corresponding user based on the distributed weights;
the weight calculation method of the similarity between the users obtained by calculating the QoS data in the theta group time comprises the following steps:
Figure FDA0003425752630000023
delta is a parameter;
and obtaining the comprehensive inter-user similarity between the target user and each user based on the weight obtained by calculation:
Figure FDA0003425752630000024
1,2,., m and e ≠ v;
3.1.3) based on the calculated comprehensive inter-user similarity between the target user and other users, selecting the first N users with the maximum similarity with the target user, and further obtaining a QoS predicted value based on the users; s3.2, QoS prediction based on service:
3.2.1) respectively calculating the service similarity of the target service and other services based on the QoS data of each group of time;
the specific calculation process for calculating the service similarity between the target service and each of the other services based on the QoS data of each group of time is as follows:
Figure FDA0003425752630000031
1,2, n and e ≠ f;
wherein s iseFor the target service, sfFor the purpose of other services, it is known to provide,
Figure FDA0003425752630000032
representing a target service s calculated based on QoS data for a theta set of timeseAnd service sfThe similarity of (2); r'θ(ui,se) Representing user uiInvoking service s within the theta set of timeseAverage of all QoS data of r'θ(ui,sf) Representing user uiInvoking a service within a theta set of timessfAverage of all QoS data of (a); u shapeθ(se) And Uθ(sf) Respectively representing the invocation of service s within the theta set of timeseAnd sfA set of users of (1);
Figure FDA0003425752630000033
respectively representing the user sets U in the theta group timeθ(se)∩Uθ(sf) Invoking a service sf、seAverage value of QoS data of time:
Figure FDA0003425752630000034
Figure FDA0003425752630000035
N(Uθ(se)∩Uθ(sf) Is a set Uθ(se)∩Uθ(sf) The number of users in (1);
3.2.2) for the similarity between the target service and each service, respectively calculating as follows:
distributing weight values for the similarity among k services obtained by calculating QoS data based on k groups of time, and calculating to obtain the comprehensive similarity among the services between a target service and the corresponding service based on the distributed weight values;
distributing different weights to the k service similarity degrees obtained by calculation according to the following formula:
Figure FDA0003425752630000041
delta is a parameter;
based on the distributed weight, the similarity between the target service and each other service is obtained as follows:
Figure FDA0003425752630000042
1,2, n and e ≠ f;
3.2.3) calculate service-based QoS prediction:
selecting the target service s based on the similarity between the target service and other serviceseThe first M services with the maximum similarity further obtain a QoS predicted value based on the services;
and S3.3, integrating the QoS predicted value based on the user calculated in the step S3.1 and the QoS predicted value based on the service calculated in the step S3.2, and calculating the QoS predicted value based on the collaborative filtering.
2. The method according to claim 1, wherein in step S1, the filter calculation formula is as follows:
r(ui,sj)＝{r(ui,sj,t)|tcurrent-T≤t≤tcurrent}，i＝1,2,...,m，j＝1,2,...,n；
wherein r (u)i,sjT) represents user uiSelecting service s at a certain time T within T timejQoS data of r (u)i,sj) Representing user uiSelecting service s during time TjAll QoS data of (a); t is tcurrentRepresenting the current time, m is the number of users and n is the number of services.
3. The method according to claim 2, wherein in step S2, the calculation formula of each element of the QoS matrix is as follows:
Figure FDA0003425752630000051
wherein N isθIs the number of QoS data in the theta group time, tbAnd tuTime of day values representing the two endpoints of the theta set of times.
4. The method of claim 1, wherein the user-based QoS prediction value is calculated as follows:
Figure FDA0003425752630000052
wherein s iseServing the target; SU (u)e) Representation and user ueAnd the top N users with the maximum similarity.
5. Method according to claim 1, characterized in that in step 3.2.3) the service based QoS prediction is calculated as follows:
Figure FDA0003425752630000053
wherein, SS(s)e) Representation and service seThe set of top M services with the greatest similarity.
6. The method according to claim 1, characterized in that the specific calculation of step S3.3 is performed according to the following formula:
Figure FDA0003425752630000054
wherein the content of the first and second substances,
Figure FDA0003425752630000055
represents a QoS prediction value based on collaborative filtering,
Figure FDA0003425752630000056
indicating a user-based QoS prediction value,
Figure FDA0003425752630000057
the lambda is a parameter factor, and the lambda is more than or equal to 0 and less than or equal to 1.
CN201910550459.1A 2019-06-24 2019-06-24 Internet of things service QoS prediction method based on collaborative filtering Active CN110266539B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910550459.1A CN110266539B (en) 2019-06-24 2019-06-24 Internet of things service QoS prediction method based on collaborative filtering

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910550459.1A CN110266539B (en) 2019-06-24 2019-06-24 Internet of things service QoS prediction method based on collaborative filtering

Publications (2)

Publication Number Publication Date
CN110266539A CN110266539A (en) 2019-09-20
CN110266539B true CN110266539B (en) 2022-03-01

Family

ID=67921014

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910550459.1A Active CN110266539B (en) 2019-06-24 2019-06-24 Internet of things service QoS prediction method based on collaborative filtering

Country Status (1)

Country Link
CN (1) CN110266539B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102130955A (en) * 2011-03-18 2011-07-20 东北大学 System and method for generating alternative service set of composite service based on collaborative filtering
CN102170449A (en) * 2011-04-28 2011-08-31 浙江大学 Web service QoS prediction method based on collaborative filtering
CN102306336A (en) * 2011-06-10 2012-01-04 浙江大学 Service selecting frame based on cooperative filtration and QoS (Quality of Service) perception
EP3349500A1 (en) * 2017-07-26 2018-07-18 smartTECH GmbH Enhancement of a cellular mobile network

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102130955A (en) * 2011-03-18 2011-07-20 东北大学 System and method for generating alternative service set of composite service based on collaborative filtering
CN102170449A (en) * 2011-04-28 2011-08-31 浙江大学 Web service QoS prediction method based on collaborative filtering
CN102306336A (en) * 2011-06-10 2012-01-04 浙江大学 Service selecting frame based on cooperative filtration and QoS (Quality of Service) perception
EP3349500A1 (en) * 2017-07-26 2018-07-18 smartTECH GmbH Enhancement of a cellular mobile network

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于协同过滤的Web服务QoS预测;陈唯;《科技创新与应用》;20141231;第2-3部分 *

Also Published As

Publication number Publication date
CN110266539A (en) 2019-09-20

Similar Documents

Publication Publication Date Title
Yu et al. A Web service QoS prediction approach based on time-and location-aware collaborative filtering
Wang et al. Service composition in cyber-physical-social systems
Li et al. An efficient and reliable approach for quality-of-service-aware service composition
Ma et al. A highly accurate prediction algorithm for unknown web service QoS values
Wang et al. Infedge: A blockchain-based incentive mechanism in hierarchical federated learning for end-edge-cloud communications
Zhu et al. WSP: A network coordinate based web service positioning framework for response time prediction
Zou et al. NDMF: Neighborhood-integrated deep matrix factorization for service QoS prediction
Liu et al. COMP: Online control mechanism for profit maximization in privacy-preserving crowdsensing
Zhang et al. LA-LMRBF: Online and long-term web service QoS forecasting
Qi et al. Personalized QoS prediction via matrix factorization integrated with neighborhood information
Chen et al. Web service QoS prediction: when collaborative filtering meets data fluctuating in big-range
Shi et al. QoS prediction for smart service management and recommendation based on the location of mobile users
Smithamol et al. TMM: trust management middleware for cloud service selection by prioritization
Wu et al. Fedab: Truthful federated learning with auction-based combinatorial multi-armed bandit
WO2022252251A1 (en) Method for predicting quality of service in service network environment
Zong et al. A broker-assisting trust and reputation system based on artificial neural network
CN110266539B (en) Internet of things service QoS prediction method based on collaborative filtering
Faruk et al. A genetic PSO algorithm with QoS-aware cluster cloud service composition
CN102880725A (en) Recommending method based on demand-based service organization
Zhang et al. A probability distribution and location-aware ResNet approach for QoS prediction
Li et al. Cultural distance-aware service recommendation approach in mobile edge computing
Zhang et al. Collective ratings for online communities with strategic users
CN106055665B (en) Context-aware Web service recommendation method and system based on abnormal value elimination
Granat et al. Multicriteria analysis in telecommunications
Kim et al. A trust type based model for managing QoS in web services composition

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