CN109088944A - Cache contents optimization algorithm based on subgradient descent method - Google Patents
Cache contents optimization algorithm based on subgradient descent method Download PDFInfo
- Publication number
- CN109088944A CN109088944A CN201811085116.4A CN201811085116A CN109088944A CN 109088944 A CN109088944 A CN 109088944A CN 201811085116 A CN201811085116 A CN 201811085116A CN 109088944 A CN109088944 A CN 109088944A
- Authority
- CN
- China
- Prior art keywords
- iteration
- optimization
- descent method
- distribution
- probability
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/56—Provisioning of proxy services
- H04L67/568—Storing data temporarily at an intermediate stage, e.g. caching
-
- 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
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/06—Testing, supervising or monitoring using simulated traffic
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The present invention provides a kind of cache contents optimization algorithms based on subgradient descent method, Successful transmissions probability is derived first with the knowledge of random geometry, then to maximize Successful transmissions probability as optimization aim, the constraint of corresponding content caching distribution is set, Optimized model is established；Objective function is rewritten as Lagrangian form, introduces the antithesis factor；Reuse nested iterations algorithm twice, optimization content caching distribution.Content caching distribution after the algorithm optimization is compared with the distribution of existing content caching, greatly improve Successful transmissions probability, solves the problems, such as the demand for services amount explosive growth for being now based on content, meanwhile it avoiding identical content and being repeated the problem of caused bandwidth and other expenses waste when distribution.
Description
Technical field
The present invention relates to a kind of cache contents optimization algorithms of field of communication technology, more particularly to one kind is in mist wireless access
The cache contents optimization algorithm based on subgradient descent method in network.
Background technique
With flourishing for mobile Internet, mobile data amount and access device number explode, and are based especially on content
Demand for services amount greatly increases.Meanwhile when the identical content of the request of different user, and these contents are repeated quickly and easily as many times as required distribution
When, the waste and other expenses of bandwidth will certainly be caused.In this context, cloud Radio Access Network be it is a kind of very
The promising network architecture, the multi-cell cooperating that can use in the following 5th third-generation mobile communication technology hold to improve network
Amount --- base station passes through digital backhaul link and exchanges signal with content supplier.Multicast and wireless caching can be then effectively reduced
Peak flow, power consumption simultaneously mitigate backhaul load.The cache for being deployed in base station end can be stored in advance with high popularity
Content, and these popular files can be transmitted directly to required user, eliminate content supplier and pass through backhaul link weight
These popular files are transmitted again.
In the related art, Cache-enabled small cell networks:Modeling and
tradeoffs,in International Symposium on Wireless Communications Systems,2014,
P.41, author assumes that each base station is provided with identical caching, i.e., all base stations all store identical most popular file.
Modeling and analysis of content caching in wireless small cell networks,
Vol.57, no.1, pp.56-60, in 2015., author considers uniform random cache distribution.But the former is because of all bases
It stands and has all cached identical popular file, so they cannot provide the diversity of cache file, therefore and waste storage money
Source.Although the random cache design of the latter can provide the diversity of file, how do not illustrate in each base station
On effectively store multiple and different files.And above-mentioned technology all only considered the buffer structure of simple layer, there is no consider
Simultaneously when there is content supplier and base station, it should how cache contents.
Summary of the invention
The purpose of the present invention is to solve the above problem, provide it is a kind of in mist Radio Access Network based on subgradient
The cache contents optimization algorithm of descent method.
Cache contents optimization algorithm based on subgradient descent method, comprising the following steps:
Initially set up Optimized model:
s.t.pn≥0,n∈M
Wherein, pnIt is the probability that file n is buffered in a base station, M@{ 1,2 ..., N } indicates N (N > 1) a file
Set, and q (p) is the Successful transmissions probability of network.The above problem is a non-convex optimization problem, n variable simultaneously by etc.
The limitation of formula constraint and inequality constraints is difficult to obtain closed solutions.
So next, objective function is rewritten as Lagrangian form, the introducing antithesis factor, to eliminate equation about
Beam；And then, nested an iteration, the maximum value of above-mentioned Lagrangian is solved using gradient descent method, and obtains phase
The p for making Lagrangian obtain maximum value answerednValue；Inequality constraints is eliminated by max function:
Cache contents optimization algorithm based on subgradient descent method is to obtain suboptimum by using two step iterative algorithms
Solution.In first iteration subalgorithm, the maximum value in above-mentioned Lagrangian descent method is iteratively solved；In second iteration
In algorithm, it then follows following rule iteration optimization antithesis factors:
In the t times iteration, we use μ (t) iteration obtained after (t-1) secondary iteration to update, finally, two straton iteration
Algorithm is cyclically updated iteration, until meeting condition: | pn(t+1)-pn(t) | after < δ (δ be can threshold value set by the user)
It is solved to updated optimization
Beneficial effects of the present invention:
(1) algorithm in the present invention, solves the problems, such as the demand for services amount explosive growth for being now based on content, together
When, it avoids identical content and is repeated the problem of caused bandwidth and other expenses waste when distribution.
(2) optimization algorithm in the present invention has universality, is suitable under a variety of backgrounds.
(3) compared with the content caching distribution obtained after the algorithm optimization in the present invention is distributed with existing content caching,
Greatly improve Successful transmissions probability.
Generally speaking, algorithm process through the invention obtains a preferably content caching and is distributed, is highly suitable for
It is the condition favourable that next generation wireless communication technology is further greatly developed in complicated, large-scale communication system.
Detailed description of the invention
Fig. 1 is system model of the invention；
Fig. 2 is the flow chart of the cache contents optimization algorithm based on subgradient descent method
Specific embodiment
The present invention provides a kind of cache contents optimization algorithms based on subgradient descent method, first with random geometry
Knowledge derives Successful transmissions probability, then to maximize Successful transmissions probability as optimization aim, sets corresponding content caching
The constraint of distribution, establishes Optimized model；Objective function is rewritten as Lagrangian form, introduces the antithesis factor；It reuses
Nested iterations algorithm twice, optimization content caching distribution.Content caching distribution and existing content caching after the algorithm optimization
Distribution compares, and greatly improves Successful transmissions probability, solves the demand for services amount explosive growth for being now based on content
The problem of, meanwhile, it avoids identical content and is repeated the problem of caused bandwidth and other expenses waste when distribution.
Below in conjunction with Figure of description, the present invention is described further.
Fig. 1 is system model figure of the invention, i.e. the present invention is the Successful transmissions probability derived under this network architecture
Algorithm.This system model is the consideration network performance in mist Radio Access Network.System model is by a content provider, and one
Portion of base stations and a part of mobile subscriber composition.Content provider and base station, it is all logical by wireless channel between base station and user
Letter.Base station and mobile subscriber obey independent homogeneous poisson process distribution, and density is respectively λbAnd λu.Base station and content provide
The total bandwidth of person is W and W' respectively, while content provider and each base station are equipped with a transmitting antenna, send function accordingly
Rate is respectively PcpAnd P.Meanwhile each mobile subscriber is equipped with a receiving antenna.Finally, considering the random spy of channel fading
Property, for large-scale fading, the decay factor for transmitting signal is D-α, wherein D is transmission range, and α > 2 is path loss index.
For multipath fading, Rayleigh fading is considered.
For file, M@{ 1,2 ..., N } represents the set of file N > 1.Content provider possesses all texts
Part, and each base station can store a file in advance, it is assumed herein that All Files are all independence and size is identical.Accordingly
Ground, for all mobile subscribers, file popularity is a@(an)n∈M, whereinIt is each use
The probability of family randomly demand file n ∈ M, it is assumed that a1≥a2...≥aN。
Each base station randomly caches a file, enables pnIndicate that file n is buffered in the probability in a base station, accordingly
Ground, p-nIndicate that file n is not buffered in base station.Enable pnMeet:
0≤pn≤1,n∈M
Therefore, content caching is distributed as p@(pn)n∈M。
Below from the angle analysis problem of user, with a typical user u0For, as user u0Randomly demand file n
When, it can be serviced first by the base station for having cached file n nearest from user；But works as in the base station around this user, do not have
Any one node B cache file n, then the user is by by the idle base station nearest from it, (" free time " refers to the base station at this time not
Service any user) transfer service, i.e., by the transfer base station to content provider demand file n, then by the transfer base station
The mobile subscriber for requesting file n is given in forwarding.When transmission rate reaches a critical value, i.e., corresponding channel capacity is greater than
Critical value, file n just can be in user u0Place correctly decodes, this process is known as Successful transmissions, corresponding probability be known as at
Function transmission probability.
After careful deriving analysis signal-to-noise ratio, the expression formula for transmission probability of succeeding is as follows:
Wherein,Refer to the probability of the no any user's request of the caching of base station itself, Beta
Function isIts complementary not exclusively Beta function is
pnIndicate that file n is buffered in the probability in a base station,
p-nIndicate that file n is not buffered in base station；Independent homogeneous poisson process distribution is obeyed in base station, and density is respectively
λb；
anIt is the probability of each user randomly demand file n；
R is the communication radius of user；
D is the distance between the base station of user and service user；
α is path loss index；
The transmission power of content provider and each base station is respectively PcpAnd P；
N is the sum of All Files；
N0It is additive white Gaussian noise.
It is illustrated in figure 2 a kind of flow chart of the cache contents optimization algorithm based on subgradient descent method under embodiment:
1. step establishes Optimized model:
Process as shown in Figure 2, initially sets up Optimized model:
s.t.pn≥0,n∈M
Wherein, pnIndicate that file n is buffered in the probability in a base station, q (p) is Successful transmissions probability, M@1,2 ...,
N } represent the set of file N > 1；Because the above problem is a non-convex optimization problem, n variable simultaneously by equality constraint and
The limitation of inequality constraints, so being difficult to obtain closed solutions.Therefore, subsequent Optimization Steps should eliminate in each iteration equation about
Beam and inequality constraints.
2. step (updates pn(t)):
pnIt (t) is p in the t times iterationnValue.Updating pn(t) before, it is necessary first to rewrite objective function, eliminate equation about
Objective function is write into the form of Lagrangian by beam again:
Wherein, μ (t) is the antithesis factor.
The sub- optimization problem to Lagrangian is solved again, updates p in this processn(t)
The maximum value of above-mentioned Lagrangian is solved using gradient descent method, and is made Lagrange accordingly
Function obtains the p of maximum valuenValue, inequality constraints is then eliminated by max function:
The above process is to update pn(t) process.
3. step (updates μ (t)):
μ (t) is the antithesis factor of above-mentioned Lagrangian.This algorithm is to be obtained by using two step iterative algorithms time
Excellent solution.In first iteration subalgorithm, that is, step 2. in, iteratively being solved with gradient descent method keeps above-mentioned glug bright
The corresponding p of day function acquirement maximum valuen；In second iteration subalgorithm, that is, in this step, it then follows following rules change
The generation optimization antithesis factor:
Wherein,It is subgradient, and c is then step-length, real number, can arbitrarily be set by user.In the t times iteration
In, we use μ (t) iteration obtained after (t-1) secondary iteration to update pn (t).
4. step (judges cycling condition):
Two layers of subiteration algorithm is cyclically updated iteration, until meeting condition: | pn(t+1)-pn(t) | (δ is can be by user by < δ
The threshold value of setting) after obtain updated optimization solution
Basic principles and main features and advantage of the invention have been shown and described above.The technical staff of the industry should
Understand, the present invention is not limited to the above embodiments, and the above embodiments and description only describe originals of the invention
Reason, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes and improvements
It all fall within the protetion scope of the claimed invention.The claimed scope of the invention is by appended claims and its equivalent
Boundary.
Claims (2)
1. a kind of cache contents optimization algorithm based on subgradient descent method, which is characterized in that initially set up system model, utilize
The knowledge of random geometry derives Successful transmissions probability, and then to maximize Successful transmissions probability as optimization aim, establishes phase
The constraint for the content caching distribution answered, establishes Optimized model；Objective function is rewritten as Lagrangian form, introduces antithesis
The factor；Nested iterations algorithm twice is reused, the content caching distribution optimized.
2. the cache contents optimization algorithm according to claim 1 based on subgradient descent method, which is characterized in that step is 1.
Establish Optimized model:
Initially set up Optimized model:
s.t.pn≥0,n∈M
Wherein, pnIndicate that file n is buffered in the probability in a base station, q (p) is Successful transmissions probability, M@{ 1,2 ..., N } generation
The set of list file N > 1；
2. step (updates pn(t)):
Updating pn(t) before, it is necessary first to rewrite objective function, eliminate equality constraint, it is bright that objective function is write into glug again
The form of day function:
Wherein, μ (t) is the antithesis factor；
The sub- optimization problem to Lagrangian is solved again, updates p in this processn(t)
The maximum value of above-mentioned Lagrangian is solved using gradient descent method, and is made Lagrangian accordingly
Obtain the p of maximum valuenValue, inequality constraints is then eliminated by max function:
3. step (updates μ (t)):
Suboptimal solution is obtained by using two step iterative algorithms；
In first iteration subalgorithm, that is, step 2. in, iteratively being solved with gradient descent method makes above-mentioned Lagrange
The corresponding p of function acquirement maximum valuen；In second iteration subalgorithm, that is, in this step, it then follows following rule iteration
Optimize the antithesis factor:
Wherein,It is subgradient, and c is then step-length, real number, can arbitrarily be set by user；In the t times iteration, make
P is updated with μ (t) iteration obtained after (t-1) secondary iterationn(t)；
4. step judges cycling condition:
Two layers of subiteration algorithm is cyclically updated iteration, until meeting condition: | pn(t+1)-pn(t) | (δ is can be by user setting by < δ
Threshold value) after obtain updated optimization solution
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811085116.4A CN109088944B (en) | 2018-09-18 | 2018-09-18 | Cache content optimization method based on sub-gradient descent method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811085116.4A CN109088944B (en) | 2018-09-18 | 2018-09-18 | Cache content optimization method based on sub-gradient descent method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109088944A true CN109088944A (en) | 2018-12-25 |
| CN109088944B CN109088944B (en) | 2020-08-14 |
Family
ID=64841898
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811085116.4A Active CN109088944B (en) | 2018-09-18 | 2018-09-18 | Cache content optimization method based on sub-gradient descent method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109088944B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109451517A (en) * | 2018-12-27 | 2019-03-08 | 同济大学 | A kind of caching placement optimization method based on mobile edge cache network |
| 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 |
| CN109788047A (en) * | 2018-12-29 | 2019-05-21 | 山东省计算中心（国家超级计算济南中心） | A kind of cache optimization method and a kind of storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160156733A1 (en) * | 2014-12-01 | 2016-06-02 | Fujitsu Limited | Content placement in hierarchical networks of caches |
| CN107968835A (en) * | 2017-12-05 | 2018-04-27 | 南京大学 | A kind of Wireless Heterogeneous Networks video cache facility dispositions method based on coding |
| CN108600365A (en) * | 2018-04-20 | 2018-09-28 | 西安交通大学 | A kind of Wireless Heterogeneous Networks caching method based on sequence study |
-
2018
- 2018-09-18 CN CN201811085116.4A patent/CN109088944B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160156733A1 (en) * | 2014-12-01 | 2016-06-02 | Fujitsu Limited | Content placement in hierarchical networks of caches |
| CN107968835A (en) * | 2017-12-05 | 2018-04-27 | 南京大学 | A kind of Wireless Heterogeneous Networks video cache facility dispositions method based on coding |
| CN108600365A (en) * | 2018-04-20 | 2018-09-28 | 西安交通大学 | A kind of Wireless Heterogeneous Networks caching method based on sequence study |
Non-Patent Citations (1)
| Title |
|---|
| WEIQI SUN等: "Joint Optimization of Cache Placement and Bandwidth Allocation in Heterogeneous Networks", 《IEEE ACCESS》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109451517A (en) * | 2018-12-27 | 2019-03-08 | 同济大学 | A kind of caching placement optimization method based on mobile edge cache network |
| CN109451517B (en) * | 2018-12-27 | 2020-06-12 | 同济大学 | Cache placement optimization method based on mobile edge cache network |
| CN109788047A (en) * | 2018-12-29 | 2019-05-21 | 山东省计算中心（国家超级计算济南中心） | A kind of cache optimization method and a kind of storage medium |
| CN109788047B (en) * | 2018-12-29 | 2021-07-06 | 山东省计算中心（国家超级计算济南中心） | Cache optimization method and storage medium |
| 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 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109088944B (en) | 2020-08-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Federated deep reinforcement learning for Internet of Things with decentralized cooperative edge caching | |
| Zhao et al. | A non-orthogonal multiple access-based multicast scheme in wireless content caching networks | |
| Pantisano et al. | Cache-aware user association in backhaul-constrained small cell networks | |
| CN108834080B (en) | Distributed cache and user association method based on multicast technology in heterogeneous network | |
| CN107548102A (en) | The node B cache method of user's time delay is minimized in a kind of edge cache network | |
| CN109088944A (en) | Cache contents optimization algorithm based on subgradient descent method | |
| He et al. | Resource allocation based on graph neural networks in vehicular communications | |
| CN108093435B (en) | Cellular downlink network energy efficiency optimization system and method based on cached popular content | |
| Wang et al. | Distributed edge caching scheme considering the tradeoff between the diversity and redundancy of cached content | |
| CN105873214B (en) | A kind of resource allocation methods of the D2D communication system based on genetic algorithm | |
| CN106851741B (en) | Distributed mobile node file caching method based on social relation in cellular network | |
| CN109673018A (en) | Novel cache contents in Wireless Heterogeneous Networks are placed and content caching distribution optimization method | |
| CN108259628A (en) | Content caching and user-association combined optimization method in isomery cellular network | |
| Zhang et al. | A sleeping mechanism for cache-enabled small cell networks with energy harvesting function | |
| CN110602722A (en) | Design method for joint content pushing and transmission based on NOMA | |
| Wang et al. | Deep reinforcement learning for caching placement and content delivery in UAV NOMA networks | |
| Yen et al. | Cooperative online caching in small cell networks with limited cache size and unknown content popularity | |
| CN110913239B (en) | Video cache updating method for refined mobile edge calculation | |
| CN109348454A (en) | A kind of D2D Cache Communication content sharing method | |
| CN110972227B (en) | Seed node selection method for offloading cellular traffic through opportunistic mobile networks | |
| Do et al. | How to cache in mobile hybrid IoT networks? | |
| CN108541025A (en) | A kind of base station towards Wireless Heterogeneous Networks and the common caching methods of D2D | |
| Kang et al. | Unmanned aerial vehicle allocation and deep learning based content caching in wireless network | |
| CN108810139B (en) | Monte Carlo tree search-assisted wireless caching method | |
| Li et al. | Edge caching for D2D enabled hierarchical wireless networks with deep reinforcement learning |
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 |