CN113709816A - Base station cooperation caching method based on multi-feature user group - Google Patents
Base station cooperation caching method based on multi-feature user group Download PDFInfo
- Publication number
- CN113709816A CN113709816A CN202110907672.0A CN202110907672A CN113709816A CN 113709816 A CN113709816 A CN 113709816A CN 202110907672 A CN202110907672 A CN 202110907672A CN 113709816 A CN113709816 A CN 113709816A
- Authority
- CN
- China
- Prior art keywords
- base station
- user
- group
- matrix
- user group
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 23
- 239000011159 matrix material Substances 0.000 claims abstract description 46
- 230000005540 biological transmission Effects 0.000 claims abstract description 21
- 238000005457 optimization Methods 0.000 claims abstract description 19
- 230000004927 fusion Effects 0.000 claims abstract description 12
- OAICVXFJPJFONN-UHFFFAOYSA-N Phosphorus Chemical compound [P] OAICVXFJPJFONN-UHFFFAOYSA-N 0.000 claims description 4
- 230000005484 gravity Effects 0.000 claims description 3
- 230000003139 buffering effect Effects 0.000 claims 4
- 230000006399 behavior Effects 0.000 description 7
- 239000013598 vector Substances 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000013213 extrapolation Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000135 prohibitive effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
Images
Classifications
-
- 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
- H04W28/08—Load balancing or load distribution
- H04W28/086—Load balancing or load distribution among access entities
- H04W28/0861—Load balancing or load distribution among access entities between base stations
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/02—Hierarchically pre-organised networks, e.g. paging networks, cellular networks, WLAN [Wireless Local Area Network] or WLL [Wireless Local Loop]
- H04W84/04—Large scale networks; Deep hierarchical networks
- H04W84/08—Trunked mobile radio systems
Abstract
The invention provides a base station cooperative caching method based on a multi-feature user group. The method comprises the steps of constructing a plurality of user characteristic similar matrixes by analyzing a base station data set of the mobile internet; combining the plurality of user characteristic similar matrixes with the optimization target to perform similar matrix fusion to obtain a final fused similar matrix; finding the finally fused similar matrixes through a community to obtain a plurality of user groups; counting the times of accessing each base station by a user group, wherein the base station with the most access times being the same user group establishes a base station cluster cooperative cache; and constructing a hit rate function and a transmission cost function according to each base station cluster, constructing constraint conditions, and further solving by using an MCMC algorithm to obtain a cache matrix. The method has the advantages that the zero norm-based characteristic similarity matrix fusion algorithm can fully utilize various characteristics of users to divide user groups, and the optimization target is combined, so that the base station cache has stronger pertinence, higher cache utilization rate and lower transmission cost.
Description
Technical Field
The invention belongs to the technical field of mobile internet, and particularly relates to a base station cooperative caching method based on a multi-feature user group.
Background
In the network information era, the number of terminal users is continuously increased, especially in the background of interconnection of everything, sensors with various data acquisition such as positioning and the like transmit information to a network at no time, and the data volume at the edge of the network is undergoing explosive growth, which brings huge pressure to a transmission link. Overloaded network traffic can cause transmission link congestion and reduce data transmission efficiency, thereby affecting user experience. In fifth generation mobile communications, the shrinking cell area and the dense deployment of wireless access points provide new opportunities for faster data transmission. However, the centralized nature of mobile network architectures and the limited transmission capacity offered by wireless backhaul links make this approach unable to keep up with the rapidly increasing traffic, and deploying an ideal backhaul for each base station becomes impractical because of the prohibitive cost, which may also lead to backhaul congestion and performance degradation.
Thus, it has become a trend that core network functionality is sunk to the network edge, and some services may also be provided by the network edge. In Mobile Edge Computing (MEC), some of the popular network resources may be cached at the base station side so that the cached network content may be delivered directly to the requesting user through multipoint cooperation between base stations without backhaul or core network transmission. The transmission delay of the user for acquiring the resource is greatly improved, and simultaneously, the resource consumption in the transmission process can be reduced. However, the buffer space of the base station is limited, so the idea of sharing space by the base station in cooperation with buffer comes. The cooperative caching indirectly expands the caching space of a single base station, and enlarges the caching file set originally stored by the single base station, so that users served by a plurality of base stations can utilize the caching diversity of the space. When a user makes a request, the requested base station and its cooperating base station may together provide service to the user. How to cache network contents efficiently and improve the hit rate of user requests, so that limited cache space is better utilized, and which contents are cached in which base stations, which is the first problem considered in cooperative caching, and meanwhile, the cooperative caching of the base stations also needs to consider which base stations establish a cooperative relationship. It is easy to think that if the number of cooperative base stations is larger, the cache space shared by them is larger, and the hit rate of the user request is higher. However, it should be noted that maintaining cooperation between base stations and communicating with each other requires some overhead of control transfer and resource scheduling, as well as the transmission cost of the link. In addition, in the large-scale base station cluster, the large-scale users are considered when the caching method is established, and the differences of different user groups in the access behaviors are averaged, so that the caching content of the base station has no pertinence, and the utilization rate of the shared caching space is not high.
Therefore, it is necessary to mine and analyze the access behavior characteristics of the users served by the base station, so that the cooperative base station group mainly serves people with similar access behaviors, and thus the base station cooperative cache can be specific to different user groups. By analyzing the base station data set of the mobile internet, descriptions of different dimensions of each access user can be obtained, such as behavior characteristics of the access position, the access time, the access frequency, the access content, the access duration and the like of the user. If the multiple characteristics of the user are fused, the user behavior can be better described.
Disclosure of Invention
The invention provides a base station cooperative caching method based on multi-feature user groups, aiming at better solving the problem of base station cooperative caching based on multi-user feature description. According to the strategy, the user groups with similar access behaviors under various characteristics are divided, and then the base station cluster with one-to-one cache optimization is built around different user groups, so that the base station cooperative cache has higher pertinence. The cache method optimizes the content placement of the cache space determined by taking the content hit rate and the transmission cost as optimization functions.
A base station cooperation caching method based on a multi-feature user group is characterized by comprising the following specific steps:
step 1, constructing a plurality of user characteristic similar matrixes by analyzing a base station data set of a mobile internet;
step 2, combining the plurality of user characteristic similar matrixes with the optimization target to perform similar matrix fusion to obtain a final fused similar matrix;
step 3, carrying out community discovery on the finally fused similar matrix through a Louvain algorithm so as to obtain a plurality of user groups;
step 4, the base stations to be clustered respectively count the times of the records with the same uid field, and according to the group division result in the step 3, the user group member person corresponding to the uid field is foundd,eThen the number of times the uid field entry is expressed as numd,e. Thereby obtaining user groupdNumber NUM _ G of times of visiting the base stationd=∑e numd,e. The user group with the largest number of visits in each base station is the main service object of the base station. The base stations of which the main service objects are the same user group establish a cooperative relationship and establish base station clusters BS corresponding to U user groups one by one1,BS2…BSU;
Step 5, according to each base station cluster obtained in the step 4, constructing a hit rate function and a transmission cost function, constructing constraint conditions, and further solving by using an MCMC algorithm to obtain a cache matrix;
preferably, the plurality of user feature similarity matrices in step 1 are: s1,S2...SH
Wherein S istRepresenting the t-th user characteristic similarity matrix, and H represents the number of the user characteristic similarity matrices;
preferably, the step 2 of combining the plurality of user characteristic similarity matrices with the optimization target to perform similarity matrix fusion specifically includes:
wherein S isbRepresenting the b-th user characteristic similar matrix, S represents the fused similar matrix to be optimized and solved, | | S | calculation0Is the zero norm, alpha, of the fused similarity matrixbIs the weight factor of the b-th user characteristic similarity matrix. Lambda is more than 0 and is a regularization term | | S | | non-woven phosphor0Represents the specific gravity of sparsity in the formula to be optimized;
solving an optimization target by using a rapid PIHT algorithm so as to obtain a final fused similarity matrix S;
preferably, the plurality of user groups in step 3 are:
group1,group2…groupU
groupd={persond,1,persond,2...persond,e}
wherein U represents the number of user groupsdRepresenting the d-th user group, persond,eRepresenting the e-th user belonging to the d-th user group, wherein each user has a unique uid field identifier in the access record of the base station;
preferably, the hit rate function in step 5 is:
wherein M represents the number of base stations in the base station group, N represents the number of contents owned in the system, phiiRepresentative base station RiThe cluster to which k belongs toiAnd k ≠ i denotes at the base station cluster φiExcept for the local base station RiOther base stations. Pi,jIs the ith row and jth column element of the user request matrix P, which represents the user to the base station RiRequest content cjProbability of, satisfyQi,jIs the ith row and jth column element of the buffer matrix Q, representing the base station RiWhether or not the content c is cachedj,Qi,jWhen 1 denotes the base station RiHas cached content Cj,Qi,jWhen 0 denotes the base station RiHas no cache content Cj。
Step 5, the transmission cost function is:
wherein, 0 < gamma < 1 represents the cost coefficient of the base station relative to directly obtaining the content from the internet and obtaining the content from the cooperative base station. The remaining symbols are as defined in the hit rate function.
And 5, the constraint conditions are as follows:
wherein S isjRepresents content CjIs set as S { S ═ S {, for all content sizes1,S2,...,SN}。ViRepresents a base station RiThe buffer space of (a) is set as V ═ V for each base station1,V2,...,VM}。
The method has the advantages that the zero norm-based characteristic similarity matrix fusion algorithm can fully utilize various characteristics of users to divide user groups, and simultaneously optimizes the cache strategy of the base station by taking the content hit rate and the transmission cost as optimization targets, so that the base station cache has stronger pertinence, higher cache utilization rate and lower transmission cost.
Drawings
FIG. 1: is a flow chart of an embodiment of the present invention.
Detailed Description
The invention is based on the idea of obtaining a plurality of user profiles from collected user data. Considering that the importance degrees of a plurality of characteristics are different in different application scenes, and in order to better fuse the plurality of characteristics, the invention uses a weighted fusion algorithm based on zero norm to perform network fusion of the similarity of the plurality of characteristics.
The specific implementation mode of the invention is a base station cooperative caching method based on a multi-feature user group, which comprises the following specific steps:
step 1, constructing a plurality of user characteristic similar matrixes by analyzing a base station data set of a mobile internet;
step 1, the plurality of user characteristic similarity matrixes are as follows: s1,S2...SH
Wherein S istRepresenting the t-th user characteristic similarity matrix, and H representing the number of the user characteristic similarity matrices
According to the actual application scene, characteristic similarity matrixes such as access positions, access time, access frequency, access contents and duration of the users can be constructed. Furthermore, the description methods of a plurality of characteristic similarity matrixes of users may be different, and common similarity definition methods include euclidean distance, cosine similarity, Jaccard similarity and Pearson similarity. When a certain characteristic similarity matrix of a user is specifically defined, some targeted improvement may need to be made on a traditional definition method, for example, when the similarity of access contents of the user is defined by using the Jaccard similarity, it is considered that some popular contents may be accessed by many people, the influence of the part of contents on the similarity calculation is large, and the calculated Jaccard similarity is high, but this cannot indicate that the access behaviors of any two users are similar. Therefore, when defining the similarity, the access frequency of the part of popular content needs to be correspondingly changed so as to reduce the influence of the similarity on the size of the similarity;
step 2, combining the plurality of user characteristic similar matrixes with the optimization target to perform similar matrix fusion to obtain a final fused similar matrix;
step 2, combining the plurality of user characteristic similar matrixes with the optimization target to perform similar matrix fusion, specifically:
wherein S isbRepresenting the b-th user characteristic similar matrix, S represents the fused similar matrix to be optimized and solved, | | S | calculation0Is the zero norm, alpha, of the fused similarity matrixbIs the weight factor of the b-th user characteristic similarity matrix. Lambda is more than 0 and is a regularization term | | S | | non-woven phosphor0Represents the specific gravity of sparsity in the formula to be optimized;
solving an optimization target by using a rapid PIHT algorithm so as to obtain a final fused similarity matrix S;
the specific fusion steps are as follows:
1. order toSimple row (column) series connection is carried out on the original matrix elements to form vectors, and the operation complexity is greatly reduced. Selecting an initial iteration point x0Selecting parameters and calculatingCorresponding Lipschitz constant L.
2. Let k be 0, perform the following loop:
while (k < max _ iter) do// loop condition, max _ iter is the maximum number of iterations
//xkRepresenting the result of the kth iteration of the vector x to be solved.Represents xkThe ith element of (1).
//ykIs an intermediate variable used for the extrapolation,i∈I(xk) Represents extrapolation only// in subspace I (x)k) Wherein I (x)k) Represents the zero-element index set of vector x.
Gradient information and outer
Whether or not the condition is satisfied, X is to be
Solving for a domain of vectors
yk+1=xkIf the above condition is satisfied, no extrapolated information is received, otherwise yk+1Representing extrapolated information
end if/if statement end flag
k is k +1// this iteration is done, the value of k is incremented by one.
end while// end of cycle
Step 3, carrying out community discovery on the finally fused similar matrix through a Louvain algorithm so as to obtain a plurality of user groups;
step 3, the plurality of user groups are:
group1,group2…groupU
groupd={persond,1,persond,2...persond,e}
wherein U represents the number of user groupsdRepresenting the d-th user group, persond,eRepresenting the e-th user belonging to the d-th user group, wherein each user has a unique uid field identifier in the access record of the base station;
the method comprises the following specific steps:
initialization, each user is regarded as a user group, and the modularity of each user is calculated.
And combining each group with the adjacent groups into a new group, and calculating new global Modularity modulation.
If modulation > modulation, the merged result of step 2 is retained along with the new Modularity.
And when the other groups do not traverse, executing the step 2 in a loop.
And judging whether the modularity is improved in the previous cycle, if so, continuing to enter the cycle, and otherwise, ending the cycle.
Step 4, the base stations to be clustered respectively count the times of the records with the same uid field, and according to the group division result in the step three, the user group member person corresponding to the uid field is foundd,eThen the number of times the uid field entry is expressed as numd,e. Thereby obtaining user groupdNumber NUM _ G of times of visiting the base stationd=∑e numd,e. With the greatest number of visits per base stationThe user group is the main service object of the base station. The base stations of which the main service objects are the same user group establish a cooperative relationship and establish base station clusters BS corresponding to U user groups one by one1,BS2…BSU。
Step 5, according to each base station cluster obtained in the step 4, constructing a hit rate function and a transmission cost function, constructing constraint conditions, and further solving by using an MCMC algorithm to obtain a cache matrix;
step 5 the hit rate function is:
wherein M represents the number of base stations in the base station group, N represents the number of contents owned in the system, phiiRepresentative base station RiThe cluster to which k belongs toiAnd k ≠ i denotes at the base station cluster φiExcept for the local base station RiOther base stations. Pi,jIs the ith row and jth column element of the user request matrix P, which represents the user to the base station RiRequest content CjProbability of, satisfyQi,jIs the ith row and jth column element of the buffer matrix Q, representing the base station RiWhether or not to cache the content Cj,Qi,jWhen 1 denotes the base station RiHas cached content Cj,Qi,jWhen 0 denotes the base station RiHas no cache content Cj。
Pi,j·Qi,jIndicate to base station RiRequested content CjThe probability that a response can be made is called the local hit rate.Is for indicating the content C requested by the userjWhether or not in a cooperative base station group phii (k∈φiAnd k ≠ i indicates that it is a base station other than the own base station). Indicate to base station RiRequested content CjCan be grouped by base station phiiExcept for the home base station RiOther base stations.
Step 5, the transmission cost function is:
wherein, 0 < gamma < 1 represents the cost coefficient of the base station relative to directly obtaining the content from the internet and obtaining the content from the cooperative base station. The remaining symbols are as defined in the hit rate function. Representing the total cost of acquiring content from the cooperating base stations,representing the total cost of obtaining content from the internet.
And 5, the constraint conditions are as follows:
wherein S isjRepresents content CjIs set as S { S ═ S {, for all content sizes1,S2,...,SN}。ViRepresents a base station RiThe buffer space of (a) is set as V ═ V for each base station1,V2,...,VM}。i∈[1,M]Indicating that the total size of the buffer content of each base station cannot exceed the buffer space size of the base station.
The concrete solving steps are as follows:
determining optimization goals for cache policy formulationNote that hit rate and transmission cost are both [0, 1 ]]Insofar, the above multiobjective optimization problem is thus transformed into a consistent form:
the solution is performed using the MCMC algorithm. Let m (q) ═ cost (q) -hit (q) + ζ f (q). F (Q) is a penalty term, ζ is a penalty factor,Qkrepresenting the state of the k-th step markov chain. Definition of QsIs of rankDefining slave State QsTo state QtIs composed ofWherein M isSIs M (Q)s) In short, ω is a normalization factor and ρ 1, 2 are scaling factors to ensure that the values of the probabilities v and q are legal. Initialization state Q0K is the maximum iteration number K, 0, 1, 2.;
when K is less than K, the following process is cycled for sampling:
the state of the Markov chain at the kth time is QkSampling Q*~q(Q*|Qk);Sampling u-Uniform [0, 1 ] from the Uniform distribution](ii) a If it is notThen receive the transfer Qk→Q*I.e. Qk+1=Q*Otherwise, transfer is not accepted, i.e. Qk+1=Qk。
For each Markov chain, the dimensions of the matrices P and Q are M N, which represents a number of base stations of M and a number of content items of N. Thus, in the Markov chain state, the number of contents buffered by each base station is equal and denoted as c. Using a neighborhood search method on the basis of Q (Q)*|Qk) Generating Q*To reduce the spatial complexity of the algorithm, i.e. we randomly select a base station and then publish a content item to cache other content items. Thus, for each iteration state, the number of next possible states is M C (N-c). Q of final output*The solved base station cache placement matrix.
It should be understood that parts of the specification not set forth in detail are well within the prior art.
It should be understood that the above description of the preferred embodiments is given for clarity and not for any purpose of limitation, and that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (5)
1. A base station cooperation caching method based on a multi-feature user group is characterized by comprising the following specific steps:
step 1, constructing a plurality of user characteristic similar matrixes by analyzing a base station data set of a mobile internet;
step 2, combining the plurality of user characteristic similar matrixes with the optimization target to perform similar matrix fusion to obtain a final fused similar matrix;
step 3, carrying out community discovery on the finally fused similar matrix through a Louvain algorithm so as to obtain a plurality of user groups;
step 4, the base stations to be clustered respectively count the times of the records with the same uid field, and according to the group division result in the step 3, the user group member person corresponding to the uid field is foundd,eThen the number of times the uid field entry is expressed as numd,e(ii) a Thereby obtaining user groupdNumber NUM _ G of times of visiting the base stationd=∑enumd,e(ii) a The user group with the most access times in each base station is a main service object of the base station; the base stations of which the main service objects are the same user group establish a cooperative relationship and establish base station clusters BS corresponding to U user groups one by one1,BS2…BSU;
And 5, constructing a hit rate function and a transmission cost function according to each base station cluster obtained in the step 4, constructing constraint conditions, and further solving by using an MCMC algorithm to obtain a cache matrix.
2. The cooperative buffering method for base stations based on multi-feature user group as claimed in claim 1,
step 1, the plurality of user characteristic similarity matrixes are as follows: s1,S2...SH
Wherein S istAnd H represents the number of the user characteristic similarity matrixes.
3. The cooperative buffering method for base stations based on multi-feature user group as claimed in claim 1,
step 2, combining the plurality of user characteristic similar matrixes with the optimization target to perform similar matrix fusion, specifically:
wherein S isbRepresenting the b-th user characteristic similarity matrix, S represents waitingOptimally solved fused similarity matrix, | S | non-woven phosphor0Is the zero norm, alpha, of the fused similarity matrixbIs the weight factor of the b-th user characteristic similarity matrix;
lambda is more than 0 and is a regularization term | | S | | non-woven phosphor0Represents the specific gravity of sparsity in the formula to be optimized;
and solving the optimization target by using a rapid PIHT algorithm so as to obtain a final fused similarity matrix S.
4. The cooperative buffering method for base stations based on multi-feature user group as claimed in claim 1,
step 3, the plurality of user groups are:
group1,group2…groupU
groupd={persond,1,persond,2...persond,e}
wherein U represents the number of user groupsdRepresenting the d-th user group, persond,eRepresenting the e-th user belonging to the d-th user group, each user has a unique uid field identification in the access record of the base station.
5. The cooperative buffering method for base stations based on multi-feature user group as claimed in claim 1,
step 5 the hit rate function is:
wherein M represents the number of base stations in the base station group, N represents the number of contents owned in the system, phiiRepresentative base station RiThe cluster to which k belongs toiAnd k ≠ i denotes at the base station cluster φiExcept for the local base station RiOther base stations of (1); pi,jIs the ith row and jth column element of the user request matrix P, which represents the user to the base station RiRequest content CjThe probability of (a) of (b) being,satisfy the requirement ofQi,jIs the ith row and jth column element of the buffer matrix Q, representing the base station RiWhether or not to cache the content Cj,Qi,jWhen 1 denotes the base station RiHas cached content Cj,Qi,jWhen 0 denotes the base station RiHas no cache content Cj;
Step 5, the transmission cost function is:
wherein, gamma is more than 0 and less than 1, which represents the cost coefficient of the base station relative to the cost coefficient of directly obtaining the content from the internet and obtaining the content from the cooperative base station; the other symbols are consistent with the definition in the hit rate function;
and 5, the constraint conditions are as follows:
wherein S isjRepresents content CjIs set as S { S ═ S {, for all content sizes1,S2,...,SN};ViRepresents a base station RiThe buffer space of (a) is set as V ═ V for each base station1,V2,...,VM}。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110623157 | 2021-06-04 | ||
CN202110623157X | 2021-06-04 |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113709816A true CN113709816A (en) | 2021-11-26 |
CN113709816B CN113709816B (en) | 2024-03-22 |
Family
ID=78651928
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110907672.0A Active CN113709816B (en) | 2021-06-04 | 2021-08-09 | Base station collaborative caching method based on multi-feature user groups |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113709816B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107104703A (en) * | 2017-04-17 | 2017-08-29 | 南京理工大学 | A kind of caching system algorithm design based on energy minimization in power line network |
CN107493328A (en) * | 2017-08-14 | 2017-12-19 | 武汉大学 | A kind of Cooperative caching method of feature based fusion |
CN107592656A (en) * | 2017-08-17 | 2018-01-16 | 东南大学 | Caching method based on base station cluster |
CN111556531A (en) * | 2020-04-28 | 2020-08-18 | 东南大学 | Cooperative cache optimization method in micro-cellular wireless network |
CN112637908A (en) * | 2021-03-08 | 2021-04-09 | 中国人民解放军国防科技大学 | Fine-grained layered edge caching method based on content popularity |
-
2021
- 2021-08-09 CN CN202110907672.0A patent/CN113709816B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107104703A (en) * | 2017-04-17 | 2017-08-29 | 南京理工大学 | A kind of caching system algorithm design based on energy minimization in power line network |
CN107493328A (en) * | 2017-08-14 | 2017-12-19 | 武汉大学 | A kind of Cooperative caching method of feature based fusion |
CN107592656A (en) * | 2017-08-17 | 2018-01-16 | 东南大学 | Caching method based on base station cluster |
CN111556531A (en) * | 2020-04-28 | 2020-08-18 | 东南大学 | Cooperative cache optimization method in micro-cellular wireless network |
CN112637908A (en) * | 2021-03-08 | 2021-04-09 | 中国人民解放军国防科技大学 | Fine-grained layered edge caching method based on content popularity |
Non-Patent Citations (1)
Title |
---|
王源;江昊;吴明;姚冬桂;张毅;羿舒文;汪海;吴静;: "基于用户移动网络接入位置的高效分布式相似矩阵计算方法", 电信科学, no. 05 * |
Also Published As
Publication number | Publication date |
---|---|
CN113709816B (en) | 2024-03-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Elgendy et al. | Joint computation offloading and task caching for multi-user and multi-task MEC systems: reinforcement learning-based algorithms | |
Song et al. | Learning-based content caching and sharing for wireless networks | |
Zhang et al. | Joint service caching, computation offloading and resource allocation in mobile edge computing systems | |
CN108667653B (en) | Cluster-based cache configuration method and device in ultra-dense network | |
Chen et al. | Efficient privacy preserving data collection and computation offloading for fog-assisted IoT | |
Qu et al. | Context-aware online client selection for hierarchical federated learning | |
He et al. | Resource allocation based on graph neural networks in vehicular communications | |
Zhang et al. | Cooperative edge caching based on temporal convolutional networks | |
Zheng et al. | MEC-enabled wireless VR video service: A learning-based mixed strategy for energy-latency tradeoff | |
Khan et al. | On the application of agglomerative hierarchical clustering for cache-assisted D2D networks | |
CN108521640B (en) | Content distribution method in cellular network | |
Lei et al. | Deep learning based proactive caching for effective wsn-enabled vision applications | |
Jiang et al. | Research on new edge computing network architecture and task offloading strategy for Internet of Things | |
CN116321307A (en) | Bidirectional cache placement method based on deep reinforcement learning in non-cellular network | |
Wu et al. | Data transmission scheme based on node model training and time division multiple access with IoT in opportunistic social networks | |
CN110062356B (en) | Cache copy layout method in D2D network | |
CN113472420B (en) | Satellite network cache placement method based on regional user interest perception | |
CN109412661B (en) | User clustering method under large-scale MIMO system | |
CN108810139B (en) | Monte Carlo tree search-assisted wireless caching method | |
Hong et al. | Optimization of cluster resource indexing of Internet of Things based on improved ant colony algorithm | |
Jiang et al. | Double DQN based computing offloading scheme for fog radio access networks | |
Nguyen et al. | Utility optimization for blockchain empowered edge computing with deep reinforcement learning | |
Li et al. | A smart content caching and replacement scheme for UAV-assisted fog computing network | |
CN113709816A (en) | Base station cooperation caching method based on multi-feature user group | |
Qin et al. | Adaptive in-network collaborative caching for enhanced ensemble deep learning at edge |
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 |