CN115103029A - Low-rank tensor completion QoS prediction method and device based on truncated nuclear norm - Google Patents

Abstract

The invention discloses a low-rank tensor completion QoS prediction method based on a truncated nuclear norm, which comprises the following steps of: obtaining service quality QoS data of a service based on the service called by a user from a mobile edge service platform; preprocessing QoS data to generate a sparse third-order QoS data tensor; establishing a low-rank tensor completion TLTC (transition threshold transport temperature) model based on a truncated nuclear norm by using a three-order QoS data tensor, and predicting QoS data lacking service by using the TLTC model and an alternative direction multiplier method to obtain a complete QoS data tensor; and determining the service recommended to the user according to the complete QoS data tensor. According to the method, missing QoS data is predicted by using a TLTC model, time correlation between different users and different services can be captured better through regularization of truncation kernel norms, the truncation degree of all tensor modes is controlled by introducing a rate parameter, and finally the optimal solution of each variable is solved based on a multiplier alternating direction method, so that good estimation of a target tensor is obtained to predict the missing QoS data, and the prediction precision is guaranteed.

Description

Low-rank tensor completion QoS prediction method and device based on truncated nuclear norm

Technical Field

The invention belongs to the technical field of information, and particularly relates to a low-rank tensor completion QoS prediction method and device based on a truncated nuclear norm.

Background

In recent years, the rapid development of mobile edge computing has led to a dramatic increase in the number of services with similar functionality. In the scenario of service invocation in a mobile edge computing environment, an edge server needs to provide relevant mobile services according to service requests issued by users. Therefore, how to select and recommend high quality services to a user from a large number of candidates is a challenging problem.

As a non-functional attribute, QoS (Quality of Service) is an important index for distinguishing the Quality of several functionally equivalent services. Due to the huge number of candidate mobile services, a user usually calls a limited number of services and obtains corresponding QoS values in the past, but accurate recommendation cannot be made for the user due to the lack of the QoS values, so that predicting the lack values based on historical QoS records is an important step of mobile service related operations.

In the related art, a collaborative filtering method is mostly adopted for predicting the missing QoS value, and the method mainly comprises a field-based collaborative filtering method and a model-based collaborative filtering method. However, in the field-based collaborative filtering method, the user-service QoS data matrix is extremely sparse, and it is difficult to find similar users or services according to the existing QoS data, so that the accuracy of detecting similar neighbors is reduced, and the final prediction effect is influenced; and if the user has not called any service or the service has not been called by any user, the neighbor has no way to determine, and the missing value cannot be predicted by using the method. For the model-based collaborative filtering method, the influence of the time factor on the QoS data is not considered, only the relation between the user and the service is considered, and the method often needs to manually set some parameters, which has a great influence on the final prediction accuracy.

Disclosure of Invention

In order to solve the problems in the prior art, the invention provides a low rank tensor completion QoS prediction method and device based on a truncated nuclear norm. The technical problem to be solved by the invention is realized by the following technical scheme:

in a first aspect, the present invention provides a QoS prediction method for low rank tensor completion based on truncated nuclear norm, including:

obtaining service quality QoS data of a service based on the service called by a user from a mobile edge service platform;

preprocessing the QoS data to generate a sparse third-order QoS data tensor;

establishing a low-rank tensor completion TLTC (transition threshold transport temperature) model based on a truncation kernel norm by using the three-order QoS data tensor, and predicting the QoS data lacking the service by using the TLTC model and an alternative direction multiplier method to obtain a complete QoS data tensor;

and determining the service recommended to the user according to the complete QoS data tensor.

In an embodiment of the present invention, the step of obtaining QoS data of the service based on the service invoked by the user from the mobile edge service platform includes:

determining services called by a user from a mobile edge service platform in a plurality of time intervals of a preset time window;

QoS data of the service is acquired.

In one embodiment of the invention, the QoS data of the service comprises a plurality of quadruplets;

wherein the quadruple comprises: user identification, service identification, invocation time and QoS values, wherein the QoS values comprise throughput and response time.

In one embodiment of the present invention, the TLTC model is:

s.t.:

in the formula,

a representation of the auxiliary tensor is given,

respectively representing the expansion of the tensor in different k modes, wherein k represents the order, k is 1,2,3, alpha _k Representing the weight cutoff parameter α _k ≥0，

The relationship of the function is expressed,

a tensor is represented which is a part of the observed tensor,

representing the truncated nuclear norm.

In a second aspect, the present invention provides a low rank tensor completion QoS prediction apparatus based on a truncated kernel norm, including:

the data acquisition module is used for acquiring the QoS data of the service based on the service called by the user from the mobile edge service platform;

the tensor generation module is used for preprocessing the QoS data to generate a sparse third-order QoS data tensor;

the prediction module is used for establishing a low-rank tensor completion TLTC model based on a truncation nuclear norm by using the three-order QoS data tensor, predicting the QoS data lacking the service by using the TLTC model and an alternative direction multiplier method, and obtaining a complete QoS data tensor;

and the determining module is used for determining the service recommended to the user according to the complete QoS data tensor.

Compared with the prior art, the invention has the beneficial effects that:

1. according to the low-rank tensor completion QoS prediction method and device based on the truncated nuclear norm, aiming at the action of time information, a time dimension is added on the basis of a second-order user-service to form a third-order tensor so as to express a QoS value, and the third-order QoS data tensor can effectively express a complex ternary relation of Qos data.

2. According to the method, missing QoS data is predicted by using a TLTC model, time correlation between different users and different services can be captured better through regularization of truncation kernel norms, while the data correlation hidden by the QoS data is considered, the truncation degree of all tensor modes is controlled by introducing a rate parameter, and finally the optimal solution of each variable is solved based on a multiplier alternating direction method, so that good estimation of a target tensor is obtained to predict the missing QoS data, and the prediction precision is effectively guaranteed.

The present invention will be described in further detail with reference to the drawings and examples.

Drawings

Fig. 1 is a flowchart of a QoS prediction method for low rank tensor completion based on truncated nuclear norm according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a service invocation scenario provided by an embodiment of the present invention;

fig. 3 is a schematic diagram of a truncated-core-norm-based low-rank tensor-completion QoS prediction method according to an embodiment of the present invention;

fig. 4 is another schematic diagram of a truncated norm-based low-rank tensor completion QoS prediction method according to an embodiment of the present invention;

FIG. 5(a) is a MAE experimental result of response times provided by embodiments of the present invention;

FIG. 5(b) is an experimental result of an RMSE with response time provided by embodiments of the present invention;

FIG. 5(c) is a MAE experimental result of throughput provided by embodiments of the present invention;

FIG. 5(d) is a result of an RMSE experiment of throughput provided by embodiments of the present invention;

FIG. 6(a) is another MAE experimental result of response times provided by embodiments of the present invention;

FIG. 6(b) is another RMSE experimental result of response times provided by embodiments of the present invention;

FIG. 6(c) is another MAE experimental result of throughput provided by an embodiment of the present invention;

FIG. 6(d) is another RMSE experimental result of throughput provided by embodiments of the present invention;

fig. 7 is a schematic structural diagram of a low rank tensor completion QoS prediction apparatus based on truncated nuclear norm according to an embodiment of the present invention.

Detailed Description

The present invention will be described in further detail with reference to specific examples, but the embodiments of the present invention are not limited thereto.

With the rapid and efficient development of cloud computing, many services are being transferred to the cloud for storage and application. The MEC (Mobile Edge Computing) is a supplement to cloud Computing, and the Edge server is used as a service carrier of a Mobile user, can provide Mobile service for the Mobile user close to the Edge Computing server, and solves the defects of insufficient storage capacity, high time delay and low bandwidth of the user in the traditional application service. At present, a recommendation technology based on service quality prediction is widely applied to cloud computing services, the problem of information overload is effectively solved, and research related to the application of the MEC field is still less.

In a real cloud environment, the time for a user to invoke a service is uncertain, and therefore, only limited services invoked in the past time can be obtained, which results in sparsity of QoS data. However, it is difficult to make a correct decision when providing high-quality service for users by using sparse QoS data, and therefore predicting missing QoS data has become a key step for providing high-quality service for users.

In view of the above, the present invention provides a method and an apparatus for QoS compensation based on truncated nuclear norm for low rank tensor.

Fig. 1 is a flowchart of a method for low rank tensor completion QoS prediction based on truncated norm according to an embodiment of the present invention. As shown in fig. 1, an embodiment of the present invention provides a QoS prediction method for low rank tensor completion based on truncated nuclear norm, including:

s1, obtaining service quality QoS data of service based on the service called by the user from the mobile edge service platform;

s2, preprocessing the QoS data to generate a sparse third-order QoS data tensor;

s3, establishing a low-rank tensor completion TLTC model based on a truncation kernel norm by using a three-order QoS data tensor, and predicting QoS data with service loss by using the TLTC model and an alternative direction multiplier method to obtain a complete QoS data tensor;

and S4, determining the service recommended to the user according to the complete QoS data tensor.

It should be understood that the number of services invoked by a user in a past time period is limited, and thus there is a lack of Qos values for the invoked services. Fig. 2 is a schematic diagram of a service invocation scenario provided in the embodiment of the present invention. Specifically, in the service invocation scenario shown in fig. 2, three users George, Durant and Harden may invoke the car rental services Server1, Server2 and Server3 provided by the cloud Server at times T1 and T2; taking user George as an example, the user George generates the Qos data of the service after calling the service each time, but since the service called by George at time T1 is Server1, Server2 and Server3, and the service called at time T2 is Server1 and Server2, the Qos data of the Server3 at time T2 is missing, that is, the Qos value of the service called by the user in a certain past period has sparseness.

Optionally, in the step S1, the step of obtaining QoS data of the service based on the service invoked by the user from the mobile edge service platform includes:

s101, determining services called by a user from a mobile edge service platform in a plurality of time intervals of a preset time window;

s102, QoS data of the service is obtained.

Generally, in a mobile service invocation environment, a user invokes a service in an accumulated time, and each time the user invokes the service, the user obtains Qos data of the invoked service. In this embodiment, the QoS data includes a plurality of quadruplets < UserID, ServiceID, TimeID, Value >, each quadruplet contains a plurality of attributes of the corresponding service, where UserID is a user identifier, ServiceID is a service identifier, TimeID is call time, and Value is a QoS Value; illustratively, the QoS value includes throughput, which is the duration between the service user sending a request and receiving a response, and response time, which is defined as the average rate at which messages are successfully delivered per unit time.

Fig. 4 is another schematic diagram of a truncated norm-based low-rank tensor completion QoS prediction method according to an embodiment of the present invention. As shown in fig. 4, after the QoS data is obtained, a sparse third order QoS data tensor is further generated. Specifically, taking the case that a User with User ID 0 calls a service with serviceid 0 when timeids are 0 and 2, and a User with User ID 0 calls a service with serviceid 2 when timeids are 0, 1 and 2 as an example, Qos values obtained by the same service called by the User in a plurality of time intervals can be stored as a time series, and a three-order Qos data tensor is constructed by using these information

Where 0 represents missing QoS data.

In this embodiment, the steps of constructing the third-order QoS data tensor are as follows:

(1) acquiring QoS data < user, service,1, value > of a service called by a user;

(2) according to QoS data<user,service,1,value>Constructing a user-service matrix N ⁽¹⁾ Taking U users at the time interval 1 as matrix rows and S called services as matrix columns;

(3) according to<user,service,1,value>Filling the corresponding QoS value in the matrix N ⁽¹⁾ Performing the following steps;

(4) constructing a matrix N of k from 1 to k at time intervals ⁽¹⁾ ,N ⁽²⁾ ,…,N ^(K) ；

(5) To obtain a third orderQoS data tensor

Where the user-service matrix is each slice of the tensor.

In this embodiment, the TLTC (Truncated nuclear norm Low-rank Tensor compensation) model is:

s.t.:

in the formula,

a representation of the auxiliary tensor is given,

respectively representing the expansion of the tensor in different k modes, wherein k represents the order, and k is 1,2,3 and alpha _k Representing the weight cutoff parameter α _k ≥0，

The relationship of the function is expressed,

a tensor is represented which is a part of the observed tensor,

representing the truncated nuclear norm.

Since the TLTC model is derived based on the matrix truncated kernel norm, the matrix truncated kernel norm will be described below. Illustratively, a given matrix

And a non-negative integer r<min (m, n), a matrix may be definedThe truncated nuclear norm of X is the sum of min (m, n) -r minimum singular values. However, as shown in the following equation, the above process does not take into account the truncation for large singular values 1 to r in the matrix:

wherein, delta _i (X) the ith singular value of the matrix X, following the ordering rule of delta ₁ ≥δ ₂ ≥δ ₃ ≥…≥δ _min{m,n} 。

The definition of the matrix truncated nuclear norm cannot be applied directly to the high-dimensional tensor, and therefore the definition of the tensor truncated nuclear norm can be combined as the basis of the LRTC for TNN minimization, and the truncation of each tensor mode will be automatically assigned if the rate parameter can be set appropriately. Data tensor for third order QoS

Its TNN definition model is as follows:

wherein the truncation of each tensor mode is:

k e {1,2,3, …, d }, d represents the mode of tensor expansion, and the parameter Θ is a generic rate parameter used to control the tensor

Degree of truncation at $ d $ mod. To r _k Should satisfy that ₁ ,α ₂ ,α ₃ ,…,α _d Tensor of representation

Weight parameters when computing TNN (Truncated Nuclear Norm) by unfolding into a matrix.

From the definition of the tensor truncated nuclear norm, the TLTC model for QoS data can be derived:

s.t.:

in order to ensure the dependency of variables of the tensor when the tensor is unfolded, and add a set of additional constraints, the invention introduces an auxiliary tensor

Under specific conditions

k＝1,2,3。

Further, the final transformation of the TLTC model is as follows:

s.t.:

introduction of tensor

After being an intermediate variable, in order to save the observed entry from being lost, the observed entry is passed to the variable

k＝1,2,3。

Optionally, the present embodiment adopts an ADMM (Alternating Direction Method of Multipliers) to optimally solve each variable.

First, the augmented lagrangian function of the TLTC model is derived:

wherein the auxiliary variable

For updating, signs<·,·>The inner product is expressed.

Thus, the ADMM iteratively transforms the original tensor completion problem into the following three sub-problems:

the order of the relationship of these three variables is derived as:

wherein let ρ be _k ρ, k, 1,2,3, i.e. variables

Conversion to:

is a fourth order tensor, with size U × S × T × 3.

Calculating tensors

k is 1,2,3 and tensor

Is given byRespectively as follows:

iteratively, from the above equation:

wherein, fold _k Representing tensor

The k-mode of (a) is unfolded,

and expressing singular value decomposition of the tensor expansion matrix, wherein U is a left singular value of the tensor expansion matrix, and V is a right singular value of the tensor expansion matrix.

Wherein,

and the fixed constraint can ensure the conversion of the observation information in each iteration.

Based on the above analysis, the step of predicting the QoS data with missing service by using the TLTC model and the alternative direction multiplier method to obtain a complete QoS data tensor includes:

(1) initialization

(2) Setting initial parameters rho, alpha _k ,γ,l＝0,

(3) Execution loop While does not converge then:

ρ＝min{1.05×ρ,ρ _max }

For k＝1,2,3then:

updating:

EndFor

updating

Updating

l＝l+1

End While

(4) Get the complete QoS data tensor

In order to verify the effectiveness of the above-mentioned low-rank tensor completion QoS prediction method based on truncated kernel norm, this embodiment uses the existing 7 QoS data interpolation algorithms to compare the prediction performance, and mainly studies two attributes of representative response time and throughput in QoS data.

Specifically, the 7 baseline methods in the prior art are:

UPCC (User-based collaborative filtering method using Pearson Correlation Coefficient) is a classic collaborative filtering method for prediction, the method uses the similarity between similar users to predict QoS data, and the UPCC method uses Pearson Coefficient to calculate the similarity;

an IPCC (Item-based collaborative filtering method using Pearson Correlation Coefficient) method predicts unknown QoS data by using the similarity between similar services, and calculates the similarity by using a Pearson Coefficient;

UIPCC (user-item-based CF method) is a mixing method combining UPCC and IPCC, the similarity between similar services and similar users is utilized to predict unknown QoS data, and the Pearson coefficient is utilized to calculate the similarity;

a probability matrix decomposition model is proposed by a PMF (probabilistic matrix factorization) method, a user service quality matrix is decomposed by using Gaussian hypothesis, and QoS missing data is predicted;

an NTF (Non-negative Tensor Factorization) method provides a Non-negative Tensor decomposition model, and Non-negative constraints are added in the CP decomposition process of the user service time Tensor model;

the WSPred (web service prediction mechanism) method regularizes the objective function by using the average value of QoS in the tensor decomposition process.

The ClUS (A Web Service Reliability prediction) method adopts a k-means clustering algorithm to aggregate QoS data of the past calling Service.

It should be noted that, in this embodiment, an accepted large-scale data set WS-Dream is selected, a third-order QoS data tensor "user-service-time" is constructed after preprocessing an original data set, and the data set records data of two indexes, i.e., throughput and response time when 142 users invoke 4532 services in 64 slots (5-minute intervals). Table 1 shows detailed statistics of Qos attribute data, with an average throughput of 9.609kbps and an average response time of 3.165 s.

TABLE 1

Then, the QoS data set is randomly deleted to ensure the sparsity of the QoS data, as shown in table 2:

TABLE 2

This embodiment sets the density of the training data set between 10% and 30%, with 5% increments, defining the tensor density as the density of the training data set. For example, the training data density value is 10%, which means that 10% is randomly selected from the data set as training data, the remaining 90% is used as test data, and 90% of the test items are predicted by using 10% of the training items. In addition, the same initial assumptions are assigned to all the models described above, and the predicted values and the original values are compared for Error using two indexes, MAE (Mean Absolute Error) and RSME (Root Mean Squared Error) under the same training and testing data set.

In addition, the weight parameter α in the TLTC model _k The truncation rate parameter γ and the learning rate ρ in the ADMM need to be initialized, and these three parameters directly affect the performance of the whole model prediction index. Alternatively, the setting of the truncation rate parameter γ in this embodiment is obtained by cross validation, but since the data is relatively complex and large and validation is performed for each deletion rate, this embodiment also performs local analysis while performing cross validation to select parameters. In particular, the local analysis is a small selection of parameters based on a priori experience, with a truncation rate parameter γ of 0.01 for both throughput and response time. For the weight parameter α _k Since cross validation also requires a large amount of calculation, the method of HaLRTC (High-accuracy Low-Rank sensor Assembly) is used to set alpha _k 。

The learning rate ρ in ADMM determines the convergence of the entire model, with larger values generally slowing down the convergence process and smaller values allowing the model to converge in several iterations. ρ is set to 1 × 10 in each of two QoS attribute data sets of throughput and response time ^-5 And ρ 1 × 10 ^-4 The maximum number of iterations is set to 20 to achieve convergence.

The embodiment further studies the influence of the truncation rate parameter on the prediction accuracy, adjusts the density of the training data set from 10% to 30%, and increases the density by 5% each time, so as to obtain the prediction accuracy of two QoS data, namely the response time and the throughput under different tensor densities. Fig. 5(a) and 5(b) are the MAE and RMSE experimental results, respectively, of response time provided by an embodiment of the present invention, and 5(c) and 5(d) are the MAE and RMSE experimental results, respectively, of throughput provided by an embodiment of the present invention. As shown in fig. 5(a) - (b), for the QoS data set of throughput, tests were performed by using truncation rate parameters of 0.01, 0.1, 0.15, 0.2, and 0.25, respectively, and the number of iterations is selected to 20 (learning rate), obviously, when the integer truncation value of tensor is 0.01, and the tensor density is between 10% and 30%, the MAE value is more stable than other decreases, which indicates that the occurrence of difference value is better handled, and the MAE value is the lowest, indicating that the prediction effect is better when the truncation rate parameter value is 0.01. For the RMSE, the truncation rate parameter of 0.01 is not ideal, but converges faster than other values, and the overall prediction effect is the best. On the other hand, when the truncation rate parameter is 0.1, the RMSE value of the throughput is relatively low, but the MAE value is relatively high.

Referring to fig. 5(c) and 5(d), the test was still performed using truncation rate parameters 0.01, 0.1, 0.15, 0.2, 0.25, and the number of iterations was 50 (learning rate ρ is 1 × 10) ^-5 ) Convergence, it is clear that similar to the Qos data set of throughput, when the truncation rate parameter is 0.01, the MAE and RMSE values are the lowest, and thus the prediction effect is the most desirable.

The experimental result shows that under different tensor densities, the prediction precision of the TLTC model adopted by the invention has smaller MAE value and RMSE value when the truncation rate parameter is 0.01, and the truncation rate parameter is locally verified according to the prior experience, so that the prediction precision is favorably improved, and the calculation cost is saved. As the tensor density increases, the prediction accuracy of the TLTC model gradually increases when the truncation rate parameter is 0.01, which means that higher prediction accuracy can be obtained by providing more QoS data.

Table 3 shows MAE values and RMSE values of 7 baseline methods in the prior art for the truncated-core-norm-based low-rank tensor completion QoS prediction method provided in this embodiment. As can be seen from table 3, the neighborhood-based CF methods UPCC, IPCC, UIPCC and matrix decomposition PMF methods have lower prediction accuracy compared to the WSPred, NTF, TLTC methods using tensor models, because they only use the user-service second-order model relationship and do not consider the more useful user-service ternary relationship and time information in the "user-service-time" model. Illustratively, the TLTC model provided by the invention remarkably improves the prediction accuracy of QoS data, and the response time and the throughput obtain smaller MAE and RMSE values under different QoS data loss densities. The MAE and RMSE values for throughput are much larger than those for response time, since throughput ranges from 0-1000kbps, and response time ranges from only 0-20 s. As the data set density increases from 10% to 30%, the MAE and RMSE values of the TLTC method become smaller because the greater the data set density, the more information is provided for missing QoS value predictions. Therefore, the low-rank tensor completion QoS prediction method based on the truncated nuclear norm has better performance than other methods.

TABLE 3

Furthermore, the prediction accuracy is strongly linked to the lack of density of QoS data tensors. Fig. 6(a) and 6(b) are the results of the MAE and RMSE experiments, respectively, of the response time provided by the embodiment of the present invention, and fig. 6(c) and 6(d) are the results of the MAE and RMSE experiments, respectively, of the throughput provided by the embodiment of the present invention. The density of the training tensor is changed from 10% to 30%, and is increased by 5% in steps, as can be seen in fig. 6(a) and 6 (b):

(1) with the increase of the training density, the performance of the low-rank tensor completion QoS prediction method based on the truncated nuclear norm provided by the invention is enhanced, and the more QoS data, the better the prediction effect.

(2) The test result obtained by adopting the TLTC model is always superior to that obtained by adopting 7 baseline methods, because the baseline method only utilizes the second-order static relation of the user-service model and does not consider the more useful ternary relation and time information of the user and the service in the user-service-time model.

(3) In the WSPred and NTF methods using the tensor model, although time information is also added, the methods use tensor decomposition for prediction, and data loss caused by decomposition is not concerned in the decomposition process, so that the prediction accuracy is reduced. The low-rank tensor completion QoS prediction method based on the truncated nuclear norm provided by the invention utilizes the automatic expansion tensor mode of the truncated nuclear norm, considers the global situation and improves the correlation of local data, thereby obtaining higher prediction precision.

Fig. 7 is a schematic structural diagram of a low rank tensor completion QoS prediction apparatus based on truncated nuclear norm according to an embodiment of the present invention. As shown in fig. 7, an embodiment of the present invention further provides a QoS prediction apparatus for low rank tensor completion based on truncated norm, including:

a data obtaining module 710, configured to obtain quality of service QoS data of a service invoked by a user from a mobile edge service platform;

a tensor generation module 720, configured to preprocess the QoS data, and generate a sparse third-order QoS data tensor;

the predicting module 730 is configured to establish a truncated norm-based low-rank tensor completion TLTC model by using the third-order QoS data tensor, and predict the QoS data lacking in service by using the TLTC model and an alternating direction multiplier method to obtain a complete QoS data tensor;

a determining module 740, configured to determine, according to the complete QoS data tensor, a service recommended for the user.

The beneficial effects of the invention are that:

1. according to the low-rank tensor completion QoS prediction method and device based on the truncated nuclear norm, aiming at the action of time information, a time dimension is added on the basis of a second-order user-service to form a third-order tensor so as to express a QoS value, and the third-order QoS data tensor can effectively express a complex ternary relation of Qos data.

2. According to the method, missing QoS data is predicted by using a TLTC model, time correlation between different users and different services can be captured better through regularization of truncation kernel norms, while the data correlation hidden by the QoS data is considered, the truncation degree of all tensor modes is controlled by introducing a rate parameter, and finally the optimal solution of each variable is solved based on a multiplier alternating direction method, so that good estimation of a target tensor is obtained to predict the missing QoS data, and the prediction precision is effectively guaranteed.

In the description of the present invention, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or to imply that the number of technical features indicated is significant. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present invention, "a plurality" means two or more unless specifically defined otherwise.

In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples described in this specification can be combined and combined by those skilled in the art.

While the present application has been described in connection with various embodiments thereof, other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed application, from a review of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the word "a" or "an" does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

The foregoing is a more detailed description of the present invention in connection with specific preferred embodiments, and it is not to be construed that the specific embodiments of the present invention are limited to those descriptions. It will be apparent to those skilled in the art that various modifications, additions and substitutions can be made without departing from the spirit of the invention.

Claims (5)

1. A low rank tensor completion QoS prediction method based on a truncated kernel norm is characterized by comprising the following steps:

obtaining service quality QoS data of a service based on the service called by a user from a mobile edge service platform;

preprocessing the QoS data to generate a sparse third-order QoS data tensor;

establishing a low-rank tensor completion TLTC (transition threshold) model based on a truncated nuclear norm by using the three-order QoS data tensor, and predicting the QoS data missing the service by using the TLTC model and an alternating direction multiplier method to obtain a complete QoS data tensor;

and determining the service recommended for the user according to the complete QoS data tensor.

2. The truncated core norm-based low rank tensor completion QoS prediction method of claim 1, wherein the step of obtaining QoS data of the service based on the service invoked by the user from a mobile edge service platform comprises:

determining services called by a user from a mobile edge service platform in a plurality of time intervals of a preset time window;

QoS data of the service is acquired.

3. The truncated nuclear norm-based low rank tensor-complemental QoS prediction method of claim 2, wherein the QoS data for the service comprises a plurality of quadruplets;

wherein the quadruple comprises: user identification, service identification, invocation time and QoS values, wherein the QoS values comprise throughput and response time.

4. The truncated nuclear norm-based low rank tensor completion QoS prediction method of claim 1, wherein the TLTC model is:

in the formula,

the auxiliary tensor is represented as a function of,

respectively representing the expansion of the tensor in different k modes, wherein k represents the order, k is 1,2,3, alpha _k Representing the weight cutoff parameter α _k ≥0，

The relationship of the function is expressed,

a tensor is represented which is a part of the observed tensor,

representing the truncated nuclear norm.

5. A low rank tensor completion QoS prediction apparatus based on a truncated kernel norm, comprising:

the data acquisition module is used for acquiring the QoS data of the service based on the service called by the user from the mobile edge service platform;

the tensor generation module is used for preprocessing the QoS data and generating a sparse third-order QoS data tensor;

the prediction module is used for establishing a low-rank tensor completion TLTC model based on a truncation nuclear norm by using the three-order QoS data tensor, predicting the QoS data lacking the service by using the TLTC model and an alternative direction multiplier method, and obtaining a complete QoS data tensor;

and the determining module is used for determining the service recommended to the user according to the complete QoS data tensor.

Images