CN107734482A - The content distribution method unloaded based on D2D and business - Google Patents
The content distribution method unloaded based on D2D and business Download PDFInfo
- Publication number
- CN107734482A CN107734482A CN201710811308.8A CN201710811308A CN107734482A CN 107734482 A CN107734482 A CN 107734482A CN 201710811308 A CN201710811308 A CN 201710811308A CN 107734482 A CN107734482 A CN 107734482A
- Authority
- CN
- China
- Prior art keywords
- user
- cluster
- content
- mrow
- msub
- 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
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W52/00—Power management, e.g. TPC [Transmission Power Control], power saving or power classes
- H04W52/02—Power saving arrangements
- H04W52/0203—Power saving arrangements in the radio access network or backbone network of wireless communication networks
- H04W52/0206—Power saving arrangements in the radio access network or backbone network of wireless communication networks in access points, e.g. base stations
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The present invention discloses a kind of content distribution method unloaded based on D2D and business, applied to the 5th Generation Mobile Communication System, belongs to wireless communication field.Implementation step is：(1) popular content set is built；(2) user preference vector is built；(3) user's cluster preference vector is built；(4) user's cluster is arbitrarily chosen；(5) the compromise model of unloading rate and energy consumption is built；(6) popular content is distributed；(7) residual memory space size is calculated；(8) judge whether residual memory space size is more than 0；(9) content cluster to be distributed is selected；(10) judge whether to complete the content distribution of all user's clusters；(11) content distribution to all users is completed.The present invention can take into account energy consumption compared with conventional contents distribution method while ensureing that BTS service unloading rate is high, and lift user experience quality.
Description
Technical field
The invention belongs to wireless communication field, further relates to a kind of base in wireless network resource distribution technique field
In the content distribution method that terminal-pair terminal D 2D and business are unloaded.The present invention can be according to user to content visitation frequency number
According to finding the point of interest of user, considering that D2D technologies and ensureing unload BTS service efficiency high and the low situation of base station energy consumption
Under, base station selected rational strategy distributes the content of its needs for user, solves the problems, such as personalized recommendation to user, to base station
Solution business unloads and power saving.
Background technology
With the fast development of mobile internet, under the scene of intensive user and huge service request, to network pipe
Reason and information transfer bring huge challenge, and the infrastructure of base station and rare frequency spectrum resource are difficult to the industry for bearing overload
Business request.In face of above mentioned problem, the research for unloading BTS service receives significant attention.D2D communications are one kind in cellular network
Novel Communication mode, the communication that can not realize terminal and terminal between short distance user by base station is realized, there is transmission
The high advantage low with energy expenditure of speed.Therefore base station can distribute content to user in portfolio hour, and then work as user
Directly with other users can not communicate using D2D technologies by base station and content is received and dispatched when having business demand so that
User terminal can unload the partial service of base station, greatly reduce the business load of base station, reduce content transmitting-receiving
During energy expenditure.For content distribution method, the content on the one hand distributed needs the interest for meeting user, the opposing party
The distribution method of face content need to ensure to unload the efficiency high of BTS service and energy consumption is low.Therefore suitable content distribution method is extremely
Close important, it, which affects user and accessed, distributes the probability of content, and then affects the unloading efficiency of BTS service, and existing
Content distribution method based on business unloading often only considered the unloading efficiency of lifting base station and have ignored the interest of user with
The energy expenditure of base station.The present invention accesses user index of the frequency as reflection user interest of content, distributes it for user
Popular content interested, the Quality of experience of user is improved, improve the unloading performance of BTS service, and in content point
While high unloading rate is ensure that in the case of hair, the energy that is consumed when having saved content distribution by base station.
Paper " the Optimal Caching Placement for D2D Assisted that Jun Rao et al. deliver at it
A kind of content distribution method based on D2D is proposed in Wireless Caching Networks " (IEEE ICC 2016).Should
The implementation steps of method are：Step 1, the popularity of content is established as strange husband's distributed model.Step 2, establishing object function is
Unloading rate, solution variable are the function model of distribution policy.Step 3, the solution tried to achieve according to step 2 distributes content for user.Should
Weak point is existing for method, and the interest of user is have ignored when establishing function model, causes the content of distribution to be not necessarily
User's content interested, and energy consumption is not accounted for, and then have influence on the efficiency of user equipment unloading BTS service, base
The energy expenditure stood is big.
Patent document " a kind of content recommendation device towards inferior mobile content dissemination system that Xiamen University applies at it
And its method " (application number：CN201510554452.9, application publication number：CN105246101A disclosed in) a kind of towards secondary
Deng the content recommendation method of mobile content dissemination system.The implementation steps of this method are, step 1, special according to zone user behavior
Seek peace zone flow feature, choose qualified content caching into region buffer unit.Step 2, user initiates content clothes
Business request, it is that user recommends region cache contents according to user characteristics and current network conditions.Step 3, user has chosen caching
After content, the cache contents are transmitted using inferior flow, and the transmission of inferior flow is controlled according to current network conditions.The party
Weak point is existing for method, due to content caching is then first distributed content into user again into region buffer unit, step
It is rapid complicated, a large amount of Internet resources are consumed, and the behavioural characteristic of user is only accounted for without the popularity in view of content,
So that the caching of content has blindness, user's probability interested in the content distributed is low, poor user experience.
The content of the invention
The purpose of the present invention is that it is deposited when unloading the content distribution in the case of BTS service for existing subscriber terminal equipment
The defects of Consumer's Experience is low and unloading rate exists with energy consumption and performance and deficiency, it is proposed that one kind is unloaded based on D2D and business
Content distribution method.The user preference vector structure that the access frequency that the present invention is internally held by user is built carries out content
Distribution, content selection problem is effectively overcome to negative effect caused by Distribution Results.
To achieve these goals, specific implementation step of the invention is as follows：
(1) popular content set is built：
(1a) according to the following formula, calculates the decision threshold of content popularit：
Wherein, ε represents the decision threshold of content popularit, and N represents the sum of content in properties collection, and ∑ represents summation behaviour
Make, t represents content number, and ∈ represents to belong to symbol, and { U } represents properties collection, ktRepresent t-th of content in properties collection
The frequency accessed by user in base station service area, y represent the total number of users in base station service area；
Each content is less than decision threshold ε content as non-prevalence by (1b) by user's access frequency in base station service area
Content, deleted from properties collection, the properties collection after non-popular content will be deleted as popular content set；
(2) user preference vector is built：
Each content of the same content type label set in popular content set as a cluster, is obtained Y by (2a)
Individual different types of popular content cluster；
(2b) arbitrarily chooses a user from the user in base station service area, and selected user is accessed in each prevalence
Hold cluster in each content frequency carry out sum operation, obtain user's access frequency, by the type label of popular content cluster with
User's access frequency forms user preference vector；
(2c) judges whether to have chosen all users in base station service area, if so, obtaining all user preference vectors, holds
Row step (3), otherwise, perform step (2b)；
(3) user's cluster preference vector is built：
(3a) by distributed big data processing platform Hadoop run K- means clustering algorithms, to base station service area
In all users preference vector carry out cluster operation, obtain user's cluster corresponding with difference preference's classification；
(3b) arbitrarily chooses user's cluster, respectively that each user in selected user's cluster is popular for any one
The access frequency of content cluster carries out sum operation, user's cluster access frequency is obtained, by the type label of popular content cluster and user
Cluster access frequency forms key-value pair, and user's cluster preference vector is formed by key-value pair；
Key-value pair in initial user cluster preference vector is accessed the frequency of content cluster from big to small by (3c) according to user's cluster
Arrangement, obtain user's cluster preference vector；
(3d) judges whether to have chosen all user's clusters in base station service area, if so, obtain all user's cluster preferences to
Amount, step (4) is performed, otherwise, perform step (3b)；
(4) user's cluster is arbitrarily chosen；
(5) the compromise model of unloading rate and energy consumption is built：
The popular content cluster that (5a) will be ordered as corresponding to the 1st key-value pair in the preference vector of selected user's cluster, make
For content cluster to be distributed；
(5b) according to the following formula, calculates each content in content cluster to be distributed and accessed by the user in selected user's cluster respectively
Probability：
Wherein, pjUser in user's cluster selected by expression for j-th of content in content cluster to be distributed access probability,
fjUser's cluster selected by expression accesses the frequency of j-th of content in content cluster to be distributed, and user's cluster selected by F expressions, which accesses, to be treated
Distribute the frequency of content cluster；
(5c) according to the following formula, the density of user in calculation base station service area：
Wherein, λ represents the density of user in base station service area, and π represents pi, and R represents base station service radius；
(5d) according to the following formula, the preference similarity of user in user's cluster selected by calculating：
Wherein, α represent selected by user's cluster user preference similarity, e represents that the index using natural constant e the bottom of as is grasped
Make,The standard deviation of user's access frequency in user's cluster selected by expression, σ represent selected by user's access frequency in user's cluster
Average value；
(5e) according to the following formula, builds the compromise model of unloading rate and energy consumption：
Wherein, min represents to find a function minimum Value Operations, g (dj) represent that solution is djFunction, djRepresent jth in content cluster
Individual content is sent to the ratio that the number of users in user's cluster takes number of users in the cluster of family,|
Expression set defines symbol, and ∩ represents to take intersection operation, and C represents content cluster to be distributed, the user of user's cluster selected by H expressions
Sum, SjRepresent that the memory space of all terminal devices in j-th of content occupied terminal cluster tool in content cluster to be distributed is big
Small, terminal device set is made up of the terminal device for possessing storage capacity that each user possesses in user's cluster, and M represents user
The storage size of all terminal devices in the terminal device set of cluster, r represent terminal-to-terminal service D2D communication radius, x tables
Show x-th of user in selected user's cluster, L represents selected user's cluster, and η represents base station power amplifier efficiency, ptTable
Show base station's transmission power, pcThe loss of base station loop power is represented, W represents channel width, log2Represent to grasp for the logarithm at bottom with 2
Make, γxRepresent the Signal to Interference plus Noise Ratio of x-th of user in user's cluster；
(6) popular content is distributed：
(6a) utilizes Multipurpose Optimal Method, obtains compromise solution；
J-th of content in content cluster to be distributed is distributed in user's cluster by (6b) base station simultaneouslyIndividual user, wherein,
H represents the number of users in user's cluster；
The content received is stored in its terminal device by the user in (6c) user's cluster；
(7) according to the following formula, in computing terminal cluster tool all terminal devices residual memory space size：
Wherein, B represents the size of all terminal device residual memory spaces in terminal device set；
(8) judge whether residual memory space size is more than 0, if so, then performing step (9), otherwise, perform step
(10)；
(9) content cluster to be distributed is selected：
According to the sequence of key-value pair in user's cluster preference vector, successively using the popular content cluster corresponding to key-value pair as treating
Step (5) is performed after distributing content cluster；
(10) judge whether to complete the content distribution of all user's clusters, if so, then performing step (11), otherwise, perform step
Suddenly (4)；
(11) content distribution to all users is completed.
The present invention has the advantage that compared with prior art：
1st, present invention employs when preparing to distribute content, popular content is built to the access frequency of content using user
Popular content, the user of similar users preference is distributed to reference to clustering method, is overcome existing by set and user preference vector
Technology is not in the case where considering user interest, and content of the user to distribution probability interested is low, poor user experience, and then
The shortcomings that causing BTS service to unload poor performance so that the present invention distributes content for user's popular content interested, improves
The Quality of experience of user, improve the unloading performance of BTS service.
2nd, present invention employs the compromise model of joint energy consumption and unloading rate, carries out popular content and is distributed to similar users
The user of preference, overcome prior art and only consider the performance of unloading rate when carrying out content distribution and ignore base station energy
Consumption, the shortcomings that causing base station energy consumption big so that the present invention has saved content while high unloading rate is ensured, for base station and divided
The energy consumed during hair.
Brief description of the drawings
Fig. 1 is the flow chart of the present invention；
Fig. 2 is the scene graph of the embodiment of the present invention；
Fig. 3 is the simulation result figure that the unloading rate of the invention with art methods changes with user density；
Fig. 4 is the simulation result figure that the energy expenditure of the invention with art methods changes with user density.
Embodiment
Below in conjunction with accompanying drawing, the present invention will be described in detail.
Reference picture 1, the specific steps of the present invention are described in further detail.
Step 1, popular content set is built.
1st step, according to the following formula, calculate the decision threshold of content popularit：
Wherein, ε represents the decision threshold of content popularit, and N represents the sum of content in properties collection, and ∑ represents summation behaviour
Make, t represents content number, and ∈ represents to belong to symbol, and { U } represents properties collection, ktRepresent t-th of content in properties collection
The frequency accessed by user in base station service area, y represent the total number of users in base station service area.
2nd step, each content is less than decision threshold ε content as non-streaming by user's access frequency in base station service area
Row content, is deleted from properties collection, will delete the properties collection after non-popular content as popular content set.
Step 2, user preference vector is built.
1st step, each content of the same content type label set in popular content set as a cluster, obtains
Obtain Y different types of popular content clusters.
2nd step, a user is arbitrarily chosen from the user in base station service area, selected user is accessed each popular
The frequency of each content in content cluster carries out sum operation, user's access frequency is obtained, by the type label of popular content cluster
User preference vector is formed with user's access frequency.
3rd step, judge whether to have chosen all users in base station service area, if so, all user preference vectors are obtained,
Step 3 is performed, otherwise, performs the step of this step the 2nd.
Step 3, user's cluster preference vector is built.
1st step, by running K- means clustering algorithms in distributed big data processing platform Hadoop, base station is serviced
The preference vector of all users in area carries out cluster operation, obtains user's cluster corresponding with difference preference's classification.
Described the step of K- means clustering algorithms are run in distributed big data processing platform Hadoop, is as follows：
A. all user preference vectors are formed into user preference vector set；
B. initial center of the K user preference vector distribution as K class is randomly selected from user preference vector set；
C. using split processes of cutting into slices, all user preference vectors in user preference vector set are grouped；
D. using map processes are mapped, according to the result of section split processes packet, calculate in user preference vector set
Each user preference vector arrives the centre distance of K initial center respectively, according to minimum centers-distance from principle, forms K and gathers
Class, while determine the customer center of K cluster；
E. using shuffle processes of shuffling, classification of shuffling is carried out to K cluster；
F. stipulations reduce processes are utilized, recalculate the cluster centre of K cluster after classification of shuffling；
G. judge to recalculate the corresponding customer center of obtained cluster centre it is whether equal, if so, completing cluster
Operation, performs the step of this step the 2nd, otherwise, performs step D.
2nd step, user's cluster is arbitrarily chosen, respectively flow each user in selected user's cluster for any one
The access frequency of row content cluster carries out sum operation, user's cluster access frequency is obtained, by the type label and use of popular content cluster
Family cluster access frequency composition key-value pair, user's cluster preference vector is formed by key-value pair.
3rd step, by the key-value pair in initial user cluster preference vector according to user's cluster access content cluster frequency from greatly to
Minispread, obtain user's cluster preference vector.
4th step, judge whether to have chosen all user's clusters in base station service area, if so, obtaining all user's cluster preferences
Vector, step 4 is performed, otherwise, performs the step of this step the 2nd.
Step 4, user's cluster is arbitrarily chosen.
Step 5, the compromise model of unloading rate and energy consumption is built.
1st step, the popular content cluster corresponding to the 1st key-value pair will be ordered as in the preference vector of selected user's cluster,
As content cluster to be distributed.
2nd step, according to the following formula, each content in content cluster to be distributed is calculated respectively and is visited by the user in selected user's cluster
The probability asked：
Wherein, pjUser in user's cluster selected by expression for j-th of content in content cluster to be distributed access probability,
fjUser's cluster selected by expression accesses the frequency of j-th of content in content cluster to be distributed, and user's cluster selected by F expressions, which accesses, to be treated
Distribute the frequency of content cluster.
3rd step, according to the following formula, the density of user in calculation base station service area：
Wherein, λ represents the density of user in base station service area, and π represents pi, and R represents base station service radius.
4th step, according to the following formula, the preference similarity of user in user's cluster selected by calculating：
Wherein, α represent selected by user's cluster user preference similarity, e represents that the index using natural constant e the bottom of as is grasped
Make,The standard deviation of user's access frequency in user's cluster selected by expression, σ represent selected by user's access frequency in user's cluster
Average value.
5th step, according to the following formula, build the compromise model of unloading rate and energy consumption：
Wherein, min represents to find a function minimum Value Operations, g (dj) represent that solution is djFunction, djRepresent jth in content cluster
Individual content is sent to the ratio that the number of users in user's cluster takes number of users in the cluster of family,
Expression set defines symbol, and ∩ represents to take intersection operation, and C represents content cluster to be distributed, the user of user's cluster selected by H expressions
Sum, SjRepresent that the memory space of all terminal devices in j-th of content occupied terminal cluster tool in content cluster to be distributed is big
Small, terminal device set is made up of the terminal device for possessing storage capacity that each user possesses in user's cluster, and M represents user
The storage size of all terminal devices in the terminal device set of cluster, r represent terminal-to-terminal service D2D communication radius, x tables
Show x-th of user in selected user's cluster, L represents selected user's cluster, and η represents base station power amplifier efficiency, ptTable
Show base station's transmission power, pcThe loss of base station loop power is represented, W represents channel width, log2Represent to grasp for the logarithm at bottom with 2
Make, γxRepresent the Signal to Interference plus Noise Ratio of x-th of user in user's cluster.
Step 6, popular content is distributed.
1st step, using Multipurpose Optimal Method, obtain compromise solution.
The step of described utilization Multipurpose Optimal Method, is as follows：
A. efficiency coefficient method is utilized, is to d in the model by the compromise model conversation of unloading rate and energy consumptionjConstrained list
Objective optimisation problems；
B. method of Lagrange multipliers is utilized, the single-object problem of Prescribed Properties is solved, obtains unloading rate and energy consumption
D in compromise modeljCompromise solution
J-th of content in content cluster to be distributed is distributed in user's cluster by the 2nd step, base station simultaneouslyIndividual user, its
In, H represents the number of users in user's cluster.
The content received is stored in its terminal device by the 3rd step, the user in user's cluster.
Step 7, according to the following formula, in computing terminal cluster tool all terminal devices residual memory space size：
Wherein, B represents the size of all terminal device residual memory spaces in terminal device set.
Step 8, judge whether residual memory space size is more than 0, if so, then performing step 9, otherwise, perform step 10.
Step 9, content cluster to be distributed is selected.
According to the sequence of key-value pair in user's cluster preference vector, successively using the popular content cluster corresponding to key-value pair as treating
Step 5 is performed after distributing content cluster.
Step 10, judge whether to complete the content distribution of all user's clusters, if so, then performing step 11, otherwise, perform step
Rapid 4.
Step 11, the content distribution to all users is completed.
The effect of the present invention can be further illustrated by following simulation result.
1. simulated conditions：
Simulated environment of the present invention is Hadoop 1.2.1 and Matlab R2014b.
2. emulate data and parameter：
Present invention emulation is the unloading rate and energy expenditure performance of contrast institute's extracting method of the present invention and art methods.
Wherein art methods are described as follows：Without the step 2 in institute's extracting method of the present invention, 3,4,9,10, and will
The formula of the 5th step is replaced as follows in step 5：
Wherein, DjRatio of j-th of the content distribution in properties collection to total number of users in base station service area is represented,T represents properties collection, and G represents total number of users in base station service area, U tables
Show the storage size of all subscriber terminal equipments in base station service area,β represents institute in base station service area
There is user preference similarity,The standard deviation of all user content access frequencys in base station service area is represented, δ represents base station clothes
The average value of all user content access frequencys in business area,qjRepresent that all users are to content in base station service area
The access probability of j-th of content, X in setjRepresent that all users access j-th of content in properties collection in base station service area
Frequency, E represents that all users in base station service area access the total number of frequencies of all the elements in properties collections.
The formula that unloading rate is calculated in art methods emulation is as follows：
Wherein, offl represents the unloading rate size of art methods.
The formula that energy expenditure is calculated in art methods emulation is as follows：
Wherein, eng represents the energy expenditure size of art methods, and Q represents the collection of all users in base station service area
Close.Unaccounted parameter is identical with the present invention above.
The parameter setting of prior art is as shown in table 1：
The art methods parameter setting list of table 1
| Parameter | Numerical value |
| D2D communication radius r | 15m |
| Content number to be distributed | 30 |
| Total storage size U of user equipment | 40G |
| Each content size S-phase is same | 1G |
| Number of users G | 20 |
| User density λ | 50,100,200,500,1000,2000 people are per sq-km |
| Bandwidth W | 20MHz |
| Base station power amplifier efficiency eta | 0.2 |
| Base station's transmission power pt | 50W |
| Base station loop power loss pc | 10W |
| Signal to Interference plus Noise Ratio γx | 50dB |
The formula that unloading rate is calculated in institute's extracting method emulation of the present invention is as follows：
Wherein, OFFL represents the unloading rate size of institute's extracting method of the present invention.
The formula that energy expenditure is calculated in institute's extracting method emulation of the present invention is as follows：
Wherein, ENG represents the energy expenditure size of institute's extracting method of the present invention.
The parameter setting of institute's extracting method of the present invention is as shown in table 2：
The institute's extracting method simulation parameter of the present invention of table 2 sets list
| Parameter | Numerical value |
| D2D communication radius r | 15m |
| Content number to be distributed | 30 |
| Total storage size M of user equipment | Each cluster is respectively：8,16,6,10 |
| Content size S | 1G |
| The number of users H of each user's cluster | Each cluster is respectively 4,8,3,5 |
| User density λ | 50,100,200,500,1000,2000 people are per sq-km |
| Bandwidth W | 20MHz |
| Base station power amplifier efficiency eta | 0.2 |
| Base station's transmission power pt | 50W |
| Base station loop power loss pc | 10W |
| Signal to Interference plus Noise Ratio γx | 50dB |
Institute's extracting method and art methods simulating scenes of the present invention are assumed to the residential quarter that single macro base station is serviced, and use
PPP distributions, the rule by cell density according to every people of sq-km 50,100,200,500,1000,2000 are obeyed in the distribution at family
Choose, the performance of the unloading rate and energy consumption under different user density is compared during to emulate.
User accesses data is as shown in table 3 used by institute's extracting method of the present invention and art methods, the first row in table 3
Video attribute, the separated film from youku.com website are represented, first row represents user label, the element representation institute in form
The user for the user label being expert at accesses the frequency of the video type of column, 0 to 300 generated at random from Matlab
Between integer.Last row represents 4 cluster labels after Hadoop is clustered, and the row of identical label represents same cluster.
The user of table 3 accesses video content frequency data list
3. analysis of simulation result：
By institute's extracting method of the present invention considers unloading rate Offloading and energy consumption Energy under user preference Interests
It is compromise, therefore IOE methods are named as in this emulation, the IOE methods described in simulation result represent the side of carrying of the invention
Method.
The emulation 1 of the present invention is to be directed to the size variation of unloading rate under different user density, and by carried IOE methods and now
There is a superiority that technical method is contrasted the extracting method so as to reflect, simulation result is as shown in Figure 3.Abscissa in Fig. 3 is
Number of users per sq-km, reflects the size of user density, and ordinate is unloading rate, and reflection user equipment can unload base
The performance size for business of standing.Weight ratio in legend represents IOE methods when carrying out the multiple-objection optimization of unloading rate and energy consumption
Weight, such as 0.3:0.7 represents that the weight that unloading rate accounts for when carrying out multiple-objection optimization is 0.3, and the weight shared by energy consumption is
0.7.Four curves are shared in Fig. 3, the curve being represented by dotted lines represents unloading rate under art methods to be changed with user density
Figure, the curve represented with solid line represent under IOE methods that unloading rate is with user density variation diagram, wherein the solid line with diamond indicia
Expression IOE methods are 0.3 in weight ratio：Unloading rate represents IOE with user density variation diagram with the solid line that asterisk marks when 0.7
Method is 0.5 in weight ratio：Unloading rate represents IOE methods with user density variation diagram with the solid line of triangular marker when 0.5
It is 0.7 in weight ratio：Unloading rate is with user density variation diagram when 0.3.As shown in figure 3, three curves of IOE methods are all located at
On curve representated by control methods, illustrate that the unloading rate of carried IOE methods is high compared with the unloading rate of control methods, than existing
Technical method performance is more preferable, and weight shared by IOE method unloading rates it is bigger curve location it is higher, unloading rate is bigger.
The emulation 2 of the present invention is Energy Expenditure Levels when being directed to distribution content in base station under different user density, and by institute
The energy expenditure for putting forward IOE methods and art methods is contrasted, as a result as shown in Figure 4.Abscissa is every square in Fig. 4
The number of users of km, reflects the size of user density, and ordinate is energy expenditure when content is distributed in base station.Power in legend
Again than representing weight of the IOE methods when carrying out the multiple-objection optimization of unloading rate and energy consumption, such as 0.3:0.7 represents more in progress
The weight that unloading rate accounts for during objective optimization is 0.3, and the weight shared by energy consumption is 0.7.Four curves are shared in Fig. 4, with dotted line
The curve of expression represents under art methods that base station energy expenditure is with user density variation diagram, the curve generation represented with solid line
Base station energy expenditure is with user density variation diagram under Table I OE methods, wherein representing that IOE methods are being weighed with the solid line of diamond indicia
Again than for 0.3：Base station energy expenditure represents that IOE methods are being weighed with user density variation diagram with the solid line that asterisk marks when 0.7
Again than for 0.5：Base station energy expenditure represents that IOE methods exist with user density variation diagram with the solid line of triangular marker when 0.5
Weight ratio is 0.7：Base station energy expenditure is with user density variation diagram when 0.3.As shown in figure 4, the energy consumption curve of IOE methods is equal
Under art methods curve, illustrate that the energy expenditure of carried IOE methods is low compared with control methods, compare art methods
Performance is more preferable, and weight shared by IOE method energy expenditures it is bigger curve location it is lower, energy expenditure is smaller.
Claims (3)
1. a kind of content distribution method unloaded based on terminal-to-terminal service D2D and business, it is characterised in that comprise the following steps：
(1) popular content set is built：
(1a) according to the following formula, calculates the decision threshold of content popularit：
<mrow>
<mi>&epsiv;</mi>
<mo>=</mo>
<mfrac>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>t</mi>
<mo>&Element;</mo>
<mo>{</mo>
<mi>U</mi>
<mo>}</mo>
</mrow>
<mi>N</mi>
</munderover>
<msub>
<mi>k</mi>
<mi>t</mi>
</msub>
</mrow>
<mi>y</mi>
</mfrac>
</mrow>
Wherein, ε represents the decision threshold of content popularit, and N represents the sum of content in properties collection, and ∑ represents sum operation, t
Content number is represented, ∈ represents to belong to symbol, and { U } represents properties collection, ktRepresent t-th of content in properties collection by base station
The frequency that user accesses in service area, y represent the total number of users in base station service area；
Each content is less than decision threshold ε content as in non-prevalence by (1b) by user's access frequency in base station service area
Hold, deleted from properties collection, the properties collection after non-popular content will be deleted as popular content set；
(2) user preference vector is built：
Each content of the same content type label set in popular content set as a cluster, is obtained Y not by (2a)
The popular content cluster of same type；
(2b) arbitrarily chooses a user from the user in base station service area, and selected user is accessed into each popular content cluster
In each content frequency carry out sum operation, user's access frequency is obtained, by the type label of popular content cluster and user
Access frequency forms user preference vector；
(2c) judges whether to have chosen all users in base station service area, if so, obtaining all user preference vectors, performs step
Suddenly (3), otherwise, step (2b) is performed；
(3) user's cluster preference vector is built：
(3a) by distributed big data processing platform Hadoop run K- means clustering algorithms, in base station service area
The preference vector of all users carries out cluster operation, obtains user's cluster corresponding with difference preference's classification；
(3b) arbitrarily chooses user's cluster, respectively by each user in selected user's cluster for any one popular content
The access frequency of cluster carries out sum operation, obtains user's cluster access frequency, and the type label of popular content cluster and user's cluster are visited
Ask that into key-value pair, user's cluster preference vector is formed by key-value pair for group of frequencies；
The frequency that (3c) accesses the key-value pair in initial user cluster preference vector according to user's cluster content cluster arranges from big to small,
Obtain user's cluster preference vector；
(3d) judges whether to have chosen all user's clusters in base station service area, if so, obtaining all user's cluster preference vectors, holds
Row step (4), otherwise, perform step (3b)；
(4) user's cluster is arbitrarily chosen；
(5) the compromise model of unloading rate and energy consumption is built：
The popular content cluster that (5a) will be ordered as corresponding to the 1st key-value pair in the preference vector of selected user's cluster, as treating
Distribute content cluster；
(5b) according to the following formula, calculate respectively each content in content cluster to be distributed accessed by the user in selected user's cluster it is general
Rate：
<mrow>
<msub>
<mi>p</mi>
<mi>j</mi>
</msub>
<mo>=</mo>
<mfrac>
<msub>
<mi>f</mi>
<mi>j</mi>
</msub>
<mi>F</mi>
</mfrac>
</mrow>
Wherein, pjUser in user's cluster selected by expression is for the access probability of j-th of content in content cluster to be distributed, fjTable
Show that selected user's cluster accesses the frequency of j-th of content in content cluster to be distributed, user's cluster selected by F expressions is accessed in be distributed
Hold the frequency of cluster；
(5c) according to the following formula, the density of user in calculation base station service area：
<mrow>
<mi>&lambda;</mi>
<mo>=</mo>
<mfrac>
<mi>y</mi>
<mrow>
<msup>
<mi>&pi;R</mi>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
</mrow>
Wherein, λ represents the density of user in base station service area, and π represents pi, and R represents base station service radius；
(5d) according to the following formula, the preference similarity of user in user's cluster selected by calculating：
Wherein, α represent selected by user's cluster user preference similarity, e represents the index operation using natural constant e the bottom of as,
The standard deviation of user's access frequency in user's cluster selected by expression, σ represent selected by user's cluster user's access frequency be averaged
Value；
(5e) according to the following formula, builds the compromise model of unloading rate and energy consumption：
<mrow>
<mi>min</mi>
<mi> </mi>
<mi>g</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>d</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mrow>
<mo>(</mo>
<mo>-</mo>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>&Element;</mo>
<mi>C</mi>
</mrow>
</munder>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<msup>
<mi>&pi;r</mi>
<mn>2</mn>
</msup>
<msub>
<mi>&lambda;d</mi>
<mi>j</mi>
</msub>
</mrow>
</msup>
<mo>)</mo>
</mrow>
<msub>
<mi>&alpha;p</mi>
<mi>j</mi>
</msub>
<mo>,</mo>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>&Element;</mo>
<mi>L</mi>
</mrow>
</munder>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>&Element;</mo>
<mi>C</mi>
</mrow>
</munder>
<mfrac>
<mrow>
<msub>
<mi>d</mi>
<mi>j</mi>
</msub>
<msub>
<mi>HS</mi>
<mi>j</mi>
</msub>
<mrow>
<mo>(</mo>
<mfrac>
<mn>1</mn>
<mi>&eta;</mi>
</mfrac>
<msub>
<mi>p</mi>
<mi>t</mi>
</msub>
<mo>+</mo>
<msub>
<mi>p</mi>
<mi>c</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>W</mi>
<mi> </mi>
<msub>
<mi>log</mi>
<mn>2</mn>
</msub>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>+</mo>
<msub>
<mi>&gamma;</mi>
<mi>x</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
Wherein, min represents to find a function minimum Value Operations, g (dj) represent that solution is djFunction, djRepresent in content cluster j-th
Hold the ratio that the number of users being sent in user's cluster takes number of users in the cluster of family,|
Expression set defines symbol, and ∩ represents to take intersection operation, and C represents content cluster to be distributed, and the user of user's cluster is total selected by H expressions
Number, SjThe storage size of all terminal devices in j-th of content occupied terminal cluster tool in content cluster to be distributed is represented,
Terminal device set is made up of the terminal device for possessing storage capacity that each user possesses in user's cluster, and M represents user's cluster
The storage size of all terminal devices in terminal device set, r represent terminal-to-terminal service D2D communication radius, x represent selected by
X-th of user in the cluster of family is taken, L represents selected user's cluster, and η represents base station power amplifier efficiency, ptRepresent base station
Transimission power, pcThe loss of base station loop power is represented, W represents channel width, log2Represent the log operations bottom of for, γ with 2xTable
Show the Signal to Interference plus Noise Ratio of x-th of user in user's cluster；
(6) popular content is distributed：
(6a) utilizes Multipurpose Optimal Method, obtains compromise solution；
J-th of content in content cluster to be distributed is distributed in user's cluster by (6b) base station simultaneouslyIndividual user, wherein, H is represented
Number of users in user's cluster；
The content received is stored in its terminal device by the user in (6c) user's cluster；
(7) according to the following formula, in computing terminal cluster tool all terminal devices residual memory space size：
<mrow>
<mi>B</mi>
<mo>=</mo>
<mi>M</mi>
<mo>-</mo>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>&Element;</mo>
<mi>C</mi>
</mrow>
</munder>
<msub>
<mi>HS</mi>
<mi>j</mi>
</msub>
<msubsup>
<mi>d</mi>
<mi>j</mi>
<mo>*</mo>
</msubsup>
</mrow>
Wherein, B represents the size of all terminal device residual memory spaces in terminal device set；
(8) judge whether residual memory space size is more than 0, if so, then performing step (9), otherwise, perform step (10)；
(9) content cluster to be distributed is selected：
According to the sequence of key-value pair in user's cluster preference vector, successively using the popular content cluster corresponding to key-value pair as to be distributed
Step (5) is performed after content cluster；
(10) judge whether to complete the content distribution of all user's clusters, if so, then performing step (11), otherwise, perform step
(4)；
(11) content distribution to all users is completed.
2. according to the content distribution method unloaded based on terminal-to-terminal service D2D and business described in claim 1, its feature exists
In as follows the step of operation K- means clustering algorithms in distributed big data processing platform Hadoop described in step (3a)：
The first step, all user preference vectors are formed into user preference vector set；
Second step, randomly selected from user preference vector set the distribution of K user preference vector as K class it is initial in
The heart；
3rd step, using split processes of cutting into slices, all user preference vectors in user preference vector set are grouped；
4th step, using map processes are mapped, according to the result of section split processes packet, calculate in user preference vector set
Each user preference vector arrives the centre distance of K initial center respectively, is clustered according to minimum centers-distance from principle, formation K,
The customer center of K cluster is determined simultaneously；
5th step, using shuffle processes of shuffling, classification of shuffling is carried out to K cluster；
6th step, using stipulations reduce processes, recalculate the K cluster centre clustered after classification of shuffling；
7th step, judge to recalculate the corresponding customer center of obtained cluster centre it is whether equal, if so, completing cluster
Operation, otherwise, perform the 4th step.
3. according to the content distribution method unloaded based on terminal-to-terminal service D2D and business described in claim 1, its feature exists
It is as follows in, the utilization Multipurpose Optimal Method described in step (6a) the step of：
The first step, it is to d in the model by the compromise model conversation of unloading rate and energy consumption using efficiency coefficient methodjConstrained list
Objective optimisation problems；
Second step, using method of Lagrange multipliers, the single-object problem of Prescribed Properties is solved, obtains unloading rate and energy consumption
D in compromise modeljCompromise solution
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710811308.8A CN107734482B (en) | 2017-09-11 | 2017-09-11 | The content distribution method unloaded based on D2D and business |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710811308.8A CN107734482B (en) | 2017-09-11 | 2017-09-11 | The content distribution method unloaded based on D2D and business |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107734482A true CN107734482A (en) | 2018-02-23 |
| CN107734482B CN107734482B (en) | 2019-10-11 |
Family
ID=61205918
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710811308.8A Active CN107734482B (en) | 2017-09-11 | 2017-09-11 | The content distribution method unloaded based on D2D and business |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107734482B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108650657A (en) * | 2018-04-20 | 2018-10-12 | 京东方科技集团股份有限公司 | Content distribution method, content updating method and relevant apparatus in a kind of car networking |
| CN108769958A (en) * | 2018-06-14 | 2018-11-06 | 重庆邮电大学 | M2M communication joint cluster based on energy optimization and resource allocation methods |
| CN109413149A (en) * | 2018-09-19 | 2019-03-01 | 上海哔哩哔哩科技有限公司 | Information distribution control method, system, server and computer readable storage medium |
| WO2020063201A1 (en) * | 2018-09-25 | 2020-04-02 | 中兴通讯股份有限公司 | Method and apparatus for creating network slice |
| CN113781181A (en) * | 2021-09-16 | 2021-12-10 | 中山大学 | Recommendation method for assisting user interest modeling based on use of item popularity |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101789182A (en) * | 2010-02-05 | 2010-07-28 | 北京工业大学 | Traffic signal control system and method based on parallel simulation technique |
| WO2013038056A1 (en) * | 2011-09-15 | 2013-03-21 | Nokia Corporation | Offloading traffic to device-to-device communications |
| CN105553612A (en) * | 2015-12-10 | 2016-05-04 | 上海华为技术有限公司 | Transmission method of D2D (device-to-device) communication links, base station and terminal |
| CN105992124A (en) * | 2015-02-12 | 2016-10-05 | 北京邮电大学 | Inter-cluster communication method based on D2D (Device-to-Device) |
| CN106535284A (en) * | 2016-12-12 | 2017-03-22 | 西安电子科技大学 | Power control method in D2D communication based on full duplex relay |
-
2017
- 2017-09-11 CN CN201710811308.8A patent/CN107734482B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101789182A (en) * | 2010-02-05 | 2010-07-28 | 北京工业大学 | Traffic signal control system and method based on parallel simulation technique |
| WO2013038056A1 (en) * | 2011-09-15 | 2013-03-21 | Nokia Corporation | Offloading traffic to device-to-device communications |
| CN105992124A (en) * | 2015-02-12 | 2016-10-05 | 北京邮电大学 | Inter-cluster communication method based on D2D (Device-to-Device) |
| CN105553612A (en) * | 2015-12-10 | 2016-05-04 | 上海华为技术有限公司 | Transmission method of D2D (device-to-device) communication links, base station and terminal |
| CN106535284A (en) * | 2016-12-12 | 2017-03-22 | 西安电子科技大学 | Power control method in D2D communication based on full duplex relay |
Non-Patent Citations (1)
| Title |
|---|
| 石玉: "异构蜂窝网络的内容分发技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108650657A (en) * | 2018-04-20 | 2018-10-12 | 京东方科技集团股份有限公司 | Content distribution method, content updating method and relevant apparatus in a kind of car networking |
| US10911917B2 (en) | 2018-04-20 | 2021-02-02 | Boe Technology Group Co., Ltd. | Content delivery method and content update method for internet of vehicles |
| CN108769958A (en) * | 2018-06-14 | 2018-11-06 | 重庆邮电大学 | M2M communication joint cluster based on energy optimization and resource allocation methods |
| CN108769958B (en) * | 2018-06-14 | 2020-11-06 | 重庆邮电大学 | M2M communication joint clustering and resource allocation method based on energy consumption optimization |
| CN109413149A (en) * | 2018-09-19 | 2019-03-01 | 上海哔哩哔哩科技有限公司 | Information distribution control method, system, server and computer readable storage medium |
| WO2020063201A1 (en) * | 2018-09-25 | 2020-04-02 | 中兴通讯股份有限公司 | Method and apparatus for creating network slice |
| US11606255B2 (en) | 2018-09-25 | 2023-03-14 | Zte Corporation | Method and apparatus for creating network slices |
| CN113781181A (en) * | 2021-09-16 | 2021-12-10 | 中山大学 | Recommendation method for assisting user interest modeling based on use of item popularity |
| CN113781181B (en) * | 2021-09-16 | 2024-03-05 | 中山大学 | Recommendation method for assisting user interest modeling based on using item popularity |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107734482B (en) | 2019-10-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107734482B (en) | The content distribution method unloaded based on D2D and business | |
| CN107995660A (en) | Support Joint Task scheduling and the resource allocation methods of D2D- Edge Servers unloading | |
| Wang et al. | CachinMobile: An energy-efficient users caching scheme for fog computing | |
| CN107105453B (en) | Cut-in method is selected based on the heterogeneous network of analytic hierarchy process (AHP) and evolutionary game theory | |
| CN111475274A (en) | Cloud collaborative multi-task scheduling method and device | |
| CN107295619A (en) | A kind of base station dormancy method based on user's connection matrix in edge cache network | |
| Khan et al. | On the application of agglomerative hierarchical clustering for cache-assisted D2D networks | |
| CN110234155A (en) | A kind of super-intensive network insertion selection method based on improved TOPSIS | |
| CN108153830A (en) | Active cache control method based on study in cordless communication network | |
| CN113115340B (en) | Popularity prediction-based cache optimization method in cellular network | |
| CN110418367A (en) | A kind of 5G forward pass mixture of networks edge cache low time delay method | |
| CN104426979A (en) | Distributed buffer scheduling system and method based on social relations | |
| CN108521640B (en) | Content distribution method in cellular network | |
| Li et al. | Dynamic spectrum access for internet-of-things based on federated deep reinforcement learning | |
| CN106921998A (en) | A kind of energy mix is powered the access control system and method for mobile communications network | |
| CN112887943B (en) | Cache resource allocation method and system based on centrality | |
| Liu et al. | Mobility-aware task offloading and migration schemes in scns with mobile edge computing | |
| CN105682176A (en) | Node incentive method based on buying-selling model and two-layer optimization | |
| Fan et al. | Task caching and computation offloading for muti-user mobile edge computing network | |
| Xu et al. | Social-aware D2D caching content deployment strategy over edge computing wireless networks | |
| CN107018528A (en) | Wireless network virtual method based on equivalent capacity in a kind of single cell LTE A systems | |
| Zhang et al. | A bankruptcy game for optimize caching resource allocation in small cell networks | |
| Wu et al. | Graph-based edge-user collaborative caching with social attributes | |
| CN111866952A (en) | D2D probabilistic cache placement method based on user group preference concentration difference | |
| He et al. | Research on streaming media cache optimization based on mobile edge computing |
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 |