CN108600365A - A kind of Wireless Heterogeneous Networks caching method based on sequence study - Google Patents

A kind of Wireless Heterogeneous Networks caching method based on sequence study Download PDF

Info

Publication number
CN108600365A
CN108600365A CN201810362003.8A CN201810362003A CN108600365A CN 108600365 A CN108600365 A CN 108600365A CN 201810362003 A CN201810362003 A CN 201810362003A CN 108600365 A CN108600365 A CN 108600365A
Authority
CN
China
Prior art keywords
base station
small base
file
user
request
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.)
Granted
Application number
CN201810362003.8A
Other languages
Chinese (zh)
Other versions
CN108600365B (en
Inventor
任品毅
张晨曦
杜清河
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.)
Xian Jiaotong University
Original Assignee
Xian Jiaotong University
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 Xian Jiaotong University filed Critical Xian Jiaotong University
Priority to CN201810362003.8A priority Critical patent/CN108600365B/en
Publication of CN108600365A publication Critical patent/CN108600365A/en
Application granted granted Critical
Publication of CN108600365B publication Critical patent/CN108600365B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/06—Testing, supervising or monitoring using simulated traffic
    • 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
    • H04L67/06—Protocols specially adapted for file transfer, e.g. file transfer protocol [FTP]
    • 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
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
    • H04L67/1004—Server selection for load balancing
    • H04L67/1008—Server selection for load balancing based on parameters of servers, e.g. available memory or workload
    • 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
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
    • H04L67/1004—Server selection for load balancing
    • H04L67/1014—Server selection for load balancing based on the content of a request
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/10—Flow control between communication endpoints
    • H04W28/14—Flow control between communication endpoints using intermediate storage

Abstract

The invention discloses a kind of Wireless Heterogeneous Networks caching methods based on sequence study.A macro base station and the cooperative system model of Multiple Small Cell Sites are initially set up, each small base station coverage area is separated from each other, and each area's intra domain user is with different probability demand files, small base station can store certain popular file equipped with buffer memory device, it is preferentially found on buffer memory device when user's demand file, then user is given from base station to macro base station request Concurrency when can not find.For the situation that user's demand file probability is unknown, by recording request history of each small base station to different files, obtain a request history list, classified to small base station using clustering method according to the list, and request history is learnt using sequence study, to obtain the cache contents of each small base station, simulation result shows, this method can achieve the purpose that reduce backhaul link flow compared with existing method in the cache hit rate under effectively improving popularity unknown situation.

Description

A kind of Wireless Heterogeneous Networks caching method based on sequence study
Technical field
The invention belongs to wireless network caching technology fields, and in particular to a kind of Wireless Heterogeneous Networks based on sequence study Caching method.
Background technology
Caching technology is that the small-sized base with certain caching function is introduced in conventional cellular communication network in wireless network It stands, user can obtain hot content by its assistance.One complete wireless network caching system includes mainly three portions Point, respectively：Hot content prediction, caching Placement Strategy and caching distribution policy.It is existing also main to the systematic research Concentrate on these three parts.Wherein in order to determine that the content of caching, content popularit are a critically important indexs, and it is existing Cache policy of the cache policy with popularity when unknown when research is also classified into known to popularity.In practice, file popularity An exactly index changed always, and be difficult to be obtained ahead of time, so many research is all solved using the method for machine learning Cache policy problem when certainly popularity is unknown.
Invention content
It is existing to overcome the purpose of the present invention is to provide a kind of Wireless Heterogeneous Networks caching method based on sequence study The problem of technology, the present invention can reach reduction backhaul link in the cache hit rate under effectively improving popularity unknown situation The purpose of flow.
In order to achieve the above objectives, the present invention adopts the following technical scheme that：
A kind of Wireless Heterogeneous Networks caching method based on sequence study, includes the following steps：
Step 1：Macro base station and the cooperative system model of Multiple Small Cell Sites are established, each small base station coverage area is mutual It separates, and each area's intra domain user, with different probability demand files, small base station is equipped with the caching list that can store popular file Member, when user's demand file first into region small base station requests file, then small base station is enterprising in own cache unit Row is searched, if this document is cached by small node B cache unit, small base station directly transmits this document to user, otherwise small base station The solicited message is forwarded to macro base station, macro base station sends the file to user；
Step 2：For the situation that user's demand file probability is unknown, different files are asked by recording each small base station History is sought, a request history list is obtained, is classified to small base station using clustering method according to the list, and utilize sequence Study learns request history, to obtain the cache contents of each small base station, system cache hit rate is made to maximize.
Further, user finds the probability of this document in demand file in small node B cache equipment in step 1, i.e., System cache hit rate is expressed as：
Wherein, M indicates small base station number, uiRepresent the number of users in i-th small base station range, the total N in system file library A file, with pi.nRepresent the probability i.e. file popularity that user in i-th small base station range asks n-th of file, xi,n=1 generation Otherwise n-th of file of table i-th small node B cache is equal to 0.
Further, step 2 specifically includes following steps：
Step 2.1：Macro base station is collected each small base station in a period of time and is asked to the request historical records of different files, formation Record list R is sought, wherein representing in a period of time request number of times of the user to each file in small base station range per a line R (i)；
Step 2.2：Classified to small base station according to request record list R, obtain k small base station clusters, each cluster forms request Record matrix Hi；
Step 2.3：Each cluster is according to its request record matrix HiEach small base station in cluster is obtained using Ranking Algorithm Ranking results vector Pi；
Step 2.4：Each small base station caches respective file according to ranking results vector in a manner of successively decreasing, until caching is single Member is filled with.
Further, classified to small base station using k-means algorithms according to request record list R in step 2.2.
Further, step 2.2 specifically includes following steps：
Step 2.2.1：K rows are randomly selected as initial barycenter from request record matrix R, and corresponding small base station is divided For k clusters；
Step 2.2.2：Computation requests record in matrix it is remaining per a line at a distance from each barycenter, and by remaining small base Station is divided into the cluster minimum with its distance；
Step 2.2.3：Each cluster forms new request record matrix Hi, and the average value for calculating all the points in the matrix is made For new barycenter；
Step 2.2.4：Step 2.2.2 and step 2.2.3 is repeated, until the change value of each cluster barycenter is less than previously given Threshold delta when, the request of output category result and each cluster records matrix Hi。
Further, step 2.3 specifically includes following steps：
2.3.1：According to request record matrix H in clusteri, the top-one probability of each single item is calculated, formula is as follows：
Wherein, hi,jRepresent request record matrix HiIn be located at the i-th row, jth row element；
2.3.2：Corresponding cross entropy loss function is obtained according to top-one probability, form is as follows：
Wherein I is label matrix, Ii,jWhether requested j-th of the file of user in i-th of base station range is represented, it is requested It is then 1, is otherwise 0；And assume Hi=UTV, Ui,ViThe i-th row of respectively U, V, g (x)=1/ (1+exp (- x)) are to rate Function, λ are regularization coefficient；
2.3.3：Model parameter U, V is updated to cross entropy loss function derivation, and using gradient descent method, more new formula is such as Under：
Wherein η is learning rate；
2.3.4：Stop update when the convergence of the value of loss function, and obtain model parameter U, V, and exports each small base station Ranking results vector Pi, calculation formula is as follows：
Compared with prior art, the present invention has technique effect beneficial below：
The present invention is directed to cache problem of the Wireless Heterogeneous Networks under popularity unknown situation, it is proposed that one kind is based on sequence The Wireless Heterogeneous Networks caching method of study, and influence of the geographical location to content popularit in base station range is considered, first Sub-clustering is carried out to base station using the method for cluster, Ranking Algorithm is then utilized to determine the cache contents of base station in each cluster, And it need to only utilize the content requests record of base station that can determine that cache contents, emulation show that this method can effective learning Content prevalence Information is spent, base station is made to maintain higher cache hit rate, achievees the purpose that reduce backhaul link flow.
Description of the drawings
Fig. 1 is the Wireless Heterogeneous Networks caching method procedure chart based on sequence study；
Fig. 2 is small base station classified part flow chart；
Fig. 3 is file ordering partial process view；
Fig. 4 is influence simulation result diagram of the zipf distributed constants to flow system flow.
Specific implementation mode
Present invention is further described in detail below in conjunction with the accompanying drawings：
The present invention proposes a kind of Wireless Heterogeneous Networks caching method based on sequence study.In being cached for wireless network The unknown situation of user's demand file probability obtains one and asks by recording request history of each small base station to different files Seek history list, classified to small base station using clustering method according to the list, and using sequence study to request history into Row study, to obtain the cache contents of each small base station, simulation result shows that this method can be compared with existing method The cache hit rate under popularity unknown situation is effectively improved, achievees the purpose that reduce backhaul link flow.
One wireless network caching system includes a macro base station and M small base stations, and each small base station coverage area is mutual Independent and small base station is equipped with the ability that buffer unit has the popular file of caching, and user can be obtained by caching in small base station range Take file.
User's demand file process description is as follows：
User's small base station requests file into region first, then small base station looked on own cache unit It looks for, if this document is cached by small node B cache unit, small base station directly transmits this document to user, and otherwise small base station should Solicited message is forwarded to macro base station, and macro base station sends the file to user.
The probability that this document is found in small node B cache equipment when system cache hit rate is defined as user's demand file, It can be expressed as：
Wherein, uiRepresent the number of users in i-th small base station range.System file library is total to N number of file, with pi.nIt represents User asks probability, that is, file popularity of n-th of file in i-th small base station range.xi,n=1, which represents i-th small base station, delays N-th of file has been deposited, has otherwise been equal to 0.
In order to keep cache hit rate of the system when content popularit is unknown as big as possible, we have proposed one kind based on row The Wireless Heterogeneous Networks caching method of sequence study.
Wireless Heterogeneous Networks caching method process based on sequence study is as follows：
1, macro base station collects each small base station in a period of time and forms request record to the request historical record of different files Matrix R, wherein representing in a period of time request number of times of the user to each file in small base station range per a line R (i).
2, classified to small base station using k-means algorithms according to request record matrix R, obtain k small base station clusters, each cluster Form request record matrix Hi。
3, each cluster obtains the sequence knot of each small base station in cluster according to its request record matrix using Ranking Algorithm Fruit vector Pi。
4, each small base station caches respective file according to ranking results vector in a manner of successively decreasing, until buffer unit is deposited It is full.
This method design includes the classification of small base station and file ordering two parts, and medium and small base station classified part flow is as follows：
1) k rows, are randomly selected as initial barycenter from request record matrix R, and corresponding small base station is divided into k clusters.
2), remaining in computation requests record matrix to be divided at a distance from each barycenter, and by remaining small base station per a line With its distance in minimum cluster.
3), each cluster forms new request record matrix Hi, and calculate according to the matrix average value conduct of all row sums New barycenter.
4) it, repeats second step and third walks, until the change value of each cluster barycenter is less than a previously given threshold delta When, the request of output category result and each cluster records matrix Hi。
File ordering part flow is as follows：
1), according to request record matrix H in clusteri, the top-one probability of each single item is calculated, formula is as follows：
Wherein hi,jRepresent request record matrix HiIn be located at the i-th row, jth row element.
2) corresponding cross entropy loss function, is obtained according to top-one probability, form is as follows：
Wherein I is label matrix, Ii,jWhether requested j-th of the file of user in i-th of base station range is represented, it is requested It is then 1, is otherwise 0.And assume Hi=UTV, Ui,ViThe i-th row of respectively U, V.G (x)=1/ (1+exp (- x)) is to rate Function.λ is regularization coefficient
3), to cross entropy loss function derivation, and model parameter U, V is updated using gradient descent method, more new formula is as follows：
Wherein η is learning rate.
4), stop update when the value of loss function restrains, and obtain model parameter U, V, and export each small base station Ranking results vector Pi, calculation formula is as follows：
By Fig. 1 it can be seen that the Wireless Heterogeneous Networks caching method complete procedure based on sequence study.
By Fig. 2 it can be seen that the small base station classified part entire flow based on k-means algorithms.
By Fig. 3 it can be seen that the partially complete flow of file ordering based on Ranking Algorithm
By Fig. 4 it can be seen that, when using institute's extracting method of the present invention, system cache hit rate with zipf parameters raising It gradually rises, and close with optimum performance, compares the caching determining method based on collaborative filtering, can effectively promote file prevalence Spend the cache hit rate under unknown situation.
Therefore in summary, the Wireless Heterogeneous Networks caching method proposed by the present invention based on sequence study can make base station It makes decision cache contents in popularity unknown situation, and maintains higher cache hit rate, reach and reduce backhaul link flow Purpose.
The above content is a further detailed description of the present invention in conjunction with specific preferred embodiments, and it cannot be said that The specific implementation mode of the present invention is only limitted to this, for those of ordinary skill in the art to which the present invention belongs, is not taking off Under the premise of from present inventive concept, several simple deduction or replace can also be made, the present invention is all shall be regarded as belonging to and is carried The protection domain that claims of friendship determine.

Claims (6)

1. a kind of Wireless Heterogeneous Networks caching method based on sequence study, which is characterized in that include the following steps：
Step 1：Macro base station and the cooperative system model of Multiple Small Cell Sites are established, each small base station coverage area is divided mutually It opens, and each area's intra domain user, with different probability demand files, small base station is equipped with the buffer unit that can store popular file, When user's demand file first into region small base station requests file, then small base station looked on own cache unit It looks for, if this document is cached by small node B cache unit, small base station directly transmits this document to user, and otherwise small base station should Solicited message is forwarded to macro base station, and macro base station sends the file to user；
Step 2：For the situation that user's demand file probability is unknown, the request of different files is gone through by recording each small base station History is obtained a request history list, is classified to small base station using clustering method according to the list, and utilizes sequence study Request history is learnt, to obtain the cache contents of each small base station, system cache hit rate is made to maximize.
2. a kind of Wireless Heterogeneous Networks caching method based on sequence study according to claim 1, which is characterized in that step User finds the probability of this document in demand file in small node B cache equipment in rapid 1, i.e. system cache hit rate indicates For：
Wherein, M indicates small base station number, uiThe number of users in i-th small base station range is represented, system file library is total to N number of text Part, with pi.nRepresent the probability i.e. file popularity that user in i-th small base station range asks n-th of file, xi,n=1 represents N-th of file of i small node B caches, is otherwise equal to 0.
3. a kind of Wireless Heterogeneous Networks caching method based on sequence study according to claim 1, which is characterized in that step Rapid 2 specifically include following steps：
Step 2.1：Macro base station collects each small base station in a period of time and forms request note to the request historical record of different files List R is recorded, wherein representing in a period of time request number of times of the user to each file in small base station range per a line R (i)；
Step 2.2：Classified to small base station according to request record list R, obtain k small base station clusters, each cluster forms request record Matrix Hi；
Step 2.3：Each cluster is according to its request record matrix HiThe sequence of each small base station in cluster is obtained using Ranking Algorithm Result vector Pi；
Step 2.4：Each small base station caches respective file according to ranking results vector in a manner of successively decreasing, until buffer unit quilt It is filled with.
4. a kind of Wireless Heterogeneous Networks caching method based on sequence study according to claim 3, which is characterized in that step Classified to small base station using k-means algorithms according to request record list R in rapid 2.2.
5. a kind of Wireless Heterogeneous Networks caching method based on sequence study according to claim 4, which is characterized in that step Rapid 2.2 specifically include following steps：
Step 2.2.1：K rows are randomly selected as initial barycenter from request record matrix R, and corresponding small base station is divided into k Cluster；
Step 2.2.2：Computation requests record remaining every a line in matrix and divide at a distance from each barycenter, and by remaining small base station Enter in the cluster minimum with its distance；
Step 2.2.3：Each cluster forms new request record matrix Hi, and the average value of all the points in the matrix is calculated as new Barycenter；
Step 2.2.4：Step 2.2.2 and step 2.2.3 is repeated, until the change value of each cluster barycenter is less than previously given door When limit value δ, the request of output category result and each cluster records matrix Hi。
6. a kind of Wireless Heterogeneous Networks caching method based on sequence study according to claim 4, which is characterized in that step Rapid 2.3 specifically include following steps：
2.3.1：According to request record matrix H in clusteri, the top-one probability of each single item is calculated, formula is as follows：
Wherein, hi,jRepresent request record matrix HiIn be located at the i-th row, jth row element；
2.3.2：Corresponding cross entropy loss function is obtained according to top-one probability, form is as follows：
Wherein I is label matrix, Ii,jWhether requested j-th of the file of user in i-th of base station range is represented, it is requested, be 1, it is otherwise 0；And assume Hi=UTV, Ui,ViThe i-th row of respectively U, V, g (x)=1/ (1+exp (- x)) be to rate function, λ is regularization coefficient；
2.3.3：Model parameter U, V is updated to cross entropy loss function derivation, and using gradient descent method, more new formula is as follows：
Wherein η is learning rate；
2.3.4：Stop update when the convergence of the value of loss function, and obtain model parameter U, V, and exports the row of each small base station Sequence result vector Pi, calculation formula is as follows：
CN201810362003.8A 2018-04-20 2018-04-20 Wireless heterogeneous network caching method based on sequencing learning Active CN108600365B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810362003.8A CN108600365B (en) 2018-04-20 2018-04-20 Wireless heterogeneous network caching method based on sequencing learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810362003.8A CN108600365B (en) 2018-04-20 2018-04-20 Wireless heterogeneous network caching method based on sequencing learning

Publications (2)

Publication Number Publication Date
CN108600365A true CN108600365A (en) 2018-09-28
CN108600365B CN108600365B (en) 2020-05-22

Family

ID=63613789

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810362003.8A Active CN108600365B (en) 2018-04-20 2018-04-20 Wireless heterogeneous network caching method based on sequencing learning

Country Status (1)

Country Link
CN (1) CN108600365B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109088944A (en) * 2018-09-18 2018-12-25 同济大学 Cache contents optimization algorithm based on subgradient descent method
CN109673018A (en) * 2019-02-13 2019-04-23 同济大学 Novel cache contents in Wireless Heterogeneous Networks are placed and content caching distribution optimization method
CN112020081A (en) * 2019-05-30 2020-12-01 韩国高等科学技术学院 Active caching method using machine learning in small cellular network based on multipoint cooperation

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106331083A (en) * 2016-08-19 2017-01-11 北京邮电大学 Heterogeneous network selection method considering content delivery energy consumption
CN107277159A (en) * 2017-07-10 2017-10-20 东南大学 A kind of super-intensive network small station caching method based on machine learning
CN107592656A (en) * 2017-08-17 2018-01-16 东南大学 Caching method based on base station cluster
CN107909108A (en) * 2017-11-15 2018-04-13 东南大学 Edge cache system and method based on content popularit prediction

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106331083A (en) * 2016-08-19 2017-01-11 北京邮电大学 Heterogeneous network selection method considering content delivery energy consumption
CN107277159A (en) * 2017-07-10 2017-10-20 东南大学 A kind of super-intensive network small station caching method based on machine learning
CN107592656A (en) * 2017-08-17 2018-01-16 东南大学 Caching method based on base station cluster
CN107909108A (en) * 2017-11-15 2018-04-13 东南大学 Edge cache system and method based on content popularit prediction

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
CHENXI ZHANG,PINYI REN: "A Contextual Multi-Armed Bandit Approach to Caching in Wireless Small Cell Network", 《2017 9TH INTERNATIONAL CONFERENCE ON WIRELESS COMMUNICATIONS AND SIGNAL PROCESSING(WCSP)》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109088944A (en) * 2018-09-18 2018-12-25 同济大学 Cache contents optimization algorithm based on subgradient descent method
CN109088944B (en) * 2018-09-18 2020-08-14 同济大学 Cache content optimization method based on sub-gradient descent method
CN109673018A (en) * 2019-02-13 2019-04-23 同济大学 Novel cache contents in Wireless Heterogeneous Networks are placed and content caching distribution optimization method
CN109673018B (en) * 2019-02-13 2020-09-15 同济大学 Novel content cache distribution optimization method in wireless heterogeneous network
CN112020081A (en) * 2019-05-30 2020-12-01 韩国高等科学技术学院 Active caching method using machine learning in small cellular network based on multipoint cooperation

Also Published As

Publication number Publication date
CN108600365B (en) 2020-05-22

Similar Documents

Publication Publication Date Title
Jiang et al. User preference learning-based edge caching for fog radio access network
CN107909108A (en) Edge cache system and method based on content popularit prediction
CN107592656B (en) Caching method based on base station clustering
Thar et al. DeepMEC: Mobile edge caching using deep learning
CN104394236B (en) Node combines the distributed collaborative caching method perceived with message status
CN105812834B (en) Video recommendations server, recommended method and pre-cache method based on clustering information
WO2023168824A1 (en) Mobile edge cache optimization method based on federated learning
CN108600365A (en) A kind of Wireless Heterogeneous Networks caching method based on sequence study
CN108667653A (en) Cached configuration method and device based on cluster in super-intensive network
CN107277159B (en) Ultra-dense network small station caching method based on machine learning
CN113115340B (en) Popularity prediction-based cache optimization method in cellular network
CN113283177B (en) Mobile perception caching method based on asynchronous federated learning
CN111491331B (en) Network perception self-adaptive caching method based on transfer learning in fog computing network
Jiang et al. A novel caching policy with content popularity prediction and user preference learning in fog-RAN
CN104426979A (en) Distributed buffer scheduling system and method based on social relations
CN104754063B (en) Local cloud computing resource scheduling method
Yan et al. Distributed edge caching with content recommendation in fog-rans via deep reinforcement learning
CN108521640B (en) Content distribution method in cellular network
Sun et al. A DQN-based cache strategy for mobile edge networks
CN110062356B (en) Cache copy layout method in D2D network
CN108733921A (en) Coiling hot point of transformer temperature fluctuation range prediction technique based on Fuzzy Information Granulation
Liu et al. Deep reinforcement learning for reactive content caching with predicted content popularity in three-tier wireless networks
CN105138536B (en) Mobile social networking data fragmentation method based on Directed Hypergraph
CN113918829A (en) Content caching and recommending method based on federal learning in fog computing network
CN107493328B (en) A kind of Cooperative caching method based on Fusion Features

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