CN107466016A - A kind of cell buffer memory device allocation algorithm based on user mobility - Google Patents
A kind of cell buffer memory device allocation algorithm based on user mobility Download PDFInfo
- Publication number
- CN107466016A CN107466016A CN201710936023.7A CN201710936023A CN107466016A CN 107466016 A CN107466016 A CN 107466016A CN 201710936023 A CN201710936023 A CN 201710936023A CN 107466016 A CN107466016 A CN 107466016A
- Authority
- CN
- China
- Prior art keywords
- chromosome
- mrow
- buffer memory
- memory device
- base station
- 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
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/029—Location-based management or tracking services
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a kind of cell buffer memory device allocation algorithm based on user mobility, belong to wireless communication field.The data set analysis of the long-term motion track of customer group is carried out in initial phase first, the system time of user's mobile trajectory data collection is divided into discrete time slot according to certain time interval.User asks a file in each time slot, calculates overall cache hit rate;Then caching assignment problem is converted into an integer programming problem;Using genetic annealing algorithms, the optimal solution of search buffer memory device capacity original allocation problem in solution space.Operated including the fitness function of special optimization design, selection operation, crossover operation etc.;If solution has restrained, now distribution of the device file between small-cell base station, and distribute buffer memory device between small-cell base station accordingly is exported, obtains optimal buffer memory device allocative decision, the cache hit rate performance of user is improved and has effectively saved equipment and be laid to this.
Description
Technical field
The invention belongs to wireless communication field, specifically a kind of cell buffer memory device distribution based on user mobility is calculated
Method.
Background technology
In recent years, as the fast development of mobile network, increasing mobile subscriber are passed through wireless using plurality of devices
Network insertion internet, therefore, the data traffic in mobile network increase substantially, and add the pressure of backbone network data transfer.
Small cell edge caching technology has the advantages of reducing cellular traffic and back haul link flow, can tackle a large amount of mobile data streams
The challenge brought is measured, lifts network performance, optimizes Consumer's Experience.
Because in next generation mobile communication system, small-cell base station disposes comparatively dense, on all small-cell base stations all
The buffer memory device of arrangement enough capacity will cause huge expense, therefore, in the case where total buffer memory device cost is limited, be
System can not ensure to be fitted with enough buffer memory devices at each small-cell base station.How buffer memory device is optimized in small-cell base station
In distribution, i.e. addressing of the buffer memory device between small-cell base station, be a key issue.
Current research all assumes that each small-cell base station when being related to the buffer memory capacity problem of small-cell base station
Carried out on the premise of possessing the buffer memory device of identical capacity, do not consider the situation of buffer memory device uneven distribution;Using
The shortcomings that buffer memory device mean allocation is obvious exactly can not neatly distribute buffer memory device between small-cell base station, and in major part
In reality scene, due to the characteristic of mobile subscriber, seldom there is the small-cell base station installation that user visits with often having a large number of users
The buffer memory device of the identical capacity of small-cell base station of visiting is substantially irrational, it will causes the waste of buffer memory device resource.
Delivered based on M.C.Gonzalez, C.A.Hidalgo, and A.-L.Barabasi et al. in 2008
Correlative study in the texts of Understanding Individual Human Mobility Patterns mono-, the moving rail of the mankind
With 24 hours, for the cycle, and regressive trend was presented in units of this cycle in mark.In addition, in the environment of cell dense deployment,
Same user can typically access several base stations, and complicated topological structure causes the assignment problem of buffer memory device to become more multiple
It is miscellaneous, it is necessary to popularity Conjoint Analysis with reference to file.
The content of the invention
The present invention is while excellent in order to maximize the cache hit rate performance of user in the case where buffer memory device total amount is limited
Change distribution of the buffer memory device in small-cell base station, it is proposed that a kind of cell buffer memory device distribution based on user mobility is calculated
Method.
Specifically:Field is distributed for the edge cache capacity of base station in the intensive small subzone network comprising a large amount of mobile subscribers
Scape, by analyzing the data acquisition system of the long-term motion track of user, the popularity distribution of comprehensive popular file, by between small-cell base station
The original allocation problem of buffer memory device and the Placement Problems of focus file are combined, and are translated into an integer programming and are asked
Topic, searches for buffer memory device capacity original allocation in solution space using the Neighborhood-region-search algorithm of Global Genetic Simulated Annealing Algorithm thought and asks
The optimal solution of topic.
Concretely comprise the following steps:
Step 1: for some user group in mobile network, the long-term shifting of each user's history in the colony is analyzed
Dynamic rail mark, and the time that all motion tracks occupy is discretized into time slot at regular intervals;
Motion track record is position of each user where in different time groove.
Step 2: it is directed in each time slot, file service of each user to one unit-sized of network request;
The probability of different files is asked to meet the distribution of file popularity;
Step 3: existed according to the location of user in each time slot, file popularity, buffer memory device and device file
Distribution between small-cell base station, calculate overall cache hit rate.
The mathematical expectation E of the cache hit rate of total usertotal(χ) is as follows:
T is the discrete time slot set of consecutive hours top-stitching division;For the set of service user;For user's request
The set of popular file;Z (f) is the file popularity size of user;I is indicator function, when its expression formula is true, I=1;
No I=0.For small set of cells;Represent user u the location of in time slot τ;For the position of all users
Put set, binary variable γp,cRepresent whether user is in base station c coverage in p positions, if in γp,c=1;It is no
Then, γp,c=0;Binary variable factor χc,fRepresent whether store file f in the c of base station, if χc,f=1;Otherwise, χc,f=
0;
Meanwhile mathematical expectation Etotal(χ) will meet following constraints:
Wherein, C1 represents the All Files stored on the c of base stationIt is less than the caching that can be installed equal to the base station
Equipment maximum capacity CSmax;The buffer memory device maximum capacity CS of all base stationsmaxIt is worth all same.
C2 represents the memory capacity summation that system is distributed for all base stationsBeing less than can distribute equal to system
Total buffer memory device capacity C Stotal;System is that the memory capacity of base station c distribution is CS (c).
Step 4: being based on genetic algorithm and simulated annealing, optimize the mathematical expectation E of overall cache hit ratetotal
(χ), obtain optimum allocation of the buffer memory device on small-cell base station, and optimal placement of the focus file on buffer memory device.
Comprise the following steps that:
Step 401, by mathematical expectation EtotalBelong to the binary variable factor χ of same distribution state in (χ)c,fArrangement
For oneOKThe matrix χ of row according to the population quantity of setting, generates several chromosomes, made at random as hereditary chromosome
For initial population.
File j whether is stored on the element representation base station i that the i-th row jth arranges in matrix χ;When element χ (i, j) is 1,
Illustrate to store file j on the i of base station;Otherwise, non-storage file on the i of base station.
Step 402, current population is set to parent population, calculates the fitness of each chromosome in parent population;
Current Population Size is N, and the chromosome in population is expressed as:χ1,χ2,…χn,...,χN;
S(χn) it is chromosome χnFitness value, be calculated as follows:
χbestFor the chromosome of caused most defect individual in all previous iteration of genetic algorithm;With the increase of iterations
Change.
Step 403, the fitness according to each chromosome, chromosome is randomly choosed using roulette formula, is calculated respectively every
Individual chromosome is chosen to entail the probability of filial generation.
For chromosome χnIt is chosen to entail the probability P of filial generationnIt is calculated as follows:
Step 405, when the probability that chromosome is chosen to entail filial generation meets constraints, select the chromosome to carry out
Subsequent operation;
Constraints is:
R is to utilize the real number of random number seed generation, 0 < r < 1;
For chromosome χn, work as probability PnWhen meeting above-mentioned constraints, chromosome χnIt is selected;
Two step 406, selection chromosomes for meeting condition, are intersected and mutation operation, obtain two new chromosomes;
Comprise the following steps that:
Step 4061, system generate two random numbers
Step 4062, utilize the two chromosome χ chosena,χb, judgeValue whether be 1, if
It is then to exchange χa(u, v) and χbThe value of (u, v), into step 4063;Otherwise, return to step 4061;
Chromosome χ after step 4063, checking exchangeaWith χbWhether constraints is met, if it is, into step 4064
Carry out mutation operation;Otherwise, return to step 4061;
Constraints is as follows:
The All Files stored in C3 representing matrixs χ on i-th of base station is less than the caching that can be installed equal to the base station
Equipment maximum capacity CSmax;Namely chromosome χaWith χbThe All Files stored in corresponding matrix on each base station meets small
In equal to CSmax;
It is assignable equal to system total that C4 represents that the memory capacity summation that system is all base station distribution in matrix χ is less than
Buffer memory device capacity C Stotal;Namely system is matrix χaWith χbIn the distribution of all base stations memory capacity summation meet it is small
In equal to CStotal。
Step 4064, system produce four random numbers
Step 4065, for the chromosome χ after exchangeaJudge whetherIf it is, exchange
χa(u, v) and χaThe value of (u ', v ') is as chromosome χaVariation result χa', otherwise, return to step 4064.
For the chromosome χ after exchangebEqually operated, obtain chromosome χb'Final variation result.
Step 407, two new chromosome χ to variation gaineda'With χb', disturbance operation is performed respectively, and judges disturbance knot
Whether fruit meets Metropolis criterions, if it is, into step 408, otherwise, disturbance behaviour is re-started to disturbance result
Make.
Disturbance operation is specific as follows:
Step 4071, system produce four random numbers
Step 4072, for mutated chromosome χa'Judge whetherIf it is, exchange
χa'(u1,v1) and χa(u2,v2) value as treating mutated chromosome χa'Disturbance result, otherwise, return to step 4064.
For the chromosome χ after variationb'Equally operated, obtain chromosome χb'Final disturbance result.
Step 408, judge whether current disturbance number reaches iteration step length, if it is, by chromosome χa'With χb'Disturbance
As a result progeny population is put into, into step 409;Otherwise, return to step 407;
Step 409, judge whether progeny population is equal to adaptive Population Size, if it is, into step 410;Otherwise, hold
Row step 406, continue selective staining body and operated.
Step 410, judge that there is maximum E in current progeny populationtotalWhether the chromosome of (χ) has restrained, if it is,
Terminate and export optimal result, otherwise, return to step 402 carries out new round iteration:
Optimal result is how to arrange buffer memory device on small-cell base station and focus text how is placed on buffer memory device
Part.
Step 5: arranging buffer memory device on small-cell base station according to optimal result, and placed on buffer memory device
Focus file.
The advantage of the invention is that:
1), a kind of cell buffer memory device allocation algorithm based on user mobility, it is possible to achieve hit to user cache
The lifting of rate.According to simulation result as can be seen that distribution of the present invention to small-cell base station buffer memory device, effectively improves use
The cache hit rate performance at family.
2) a kind of, cell buffer memory device allocation algorithm based on user mobility, can save buffer memory device cost, lead to
The shift position preference of analysis user is crossed, the buffer memory device for the small-cell base station for seldom having user to visit is reduced, so as to save
The cost of buffer memory device is laid by operator.
Brief description of the drawings
Fig. 1 is the long-term motion track schematic diagram of user group's history in mobile network of the present invention;
Fig. 2 is a kind of cell buffer memory device allocation algorithm flow chart based on user mobility of the present invention;
Fig. 3 is the flow for the mathematical expectation that the present invention optimizes overall cache hit rate based on Global Genetic Simulated Annealing Algorithm
Figure;
Fig. 4 is present invention figure compared with the change that cache hit rate under mean allocation algorithm is distributed with the popularity of file;
Fig. 5 is that the present invention illustrates with cache hit rate under mean allocation algorithm with the relation pair ratio of buffer memory capacity saturation degree
Figure.
Embodiment
Below in conjunction with drawings and examples, the present invention is described in further detail.
A kind of cell buffer memory device allocation algorithm based on user mobility of the present invention, with edge cache technology
In small subzone network, first, the data set analysis of the long-term motion track of customer group is carried out in initial phase, user is moved
The system time of track data collection is divided into discrete time slot according to certain time interval.User please in each time slot
A file is sought, and according to the popularity distribution of user present position, comprehensive popular file in each time slot, by cell base
The original allocation problem of buffer memory device and focus file are combined in the Placement Problems of buffer memory device between standing, and calculate overall caching
Hit rate;Then, caching assignment problem is converted into an integer programming problem;Using genetic annealing algorithms, with device file
Distribution variable between small-cell base station is chromosome, and the optimal of buffer memory device capacity original allocation problem is searched in solution space
Solution, to optimize overall cache hit rate.Including the fitness function of special optimization design, selection operation, crossover operation, change
ETTHER-OR operation, adaptive crossover and mutation probability mechanism, disturbance operation, cooling function, adaptive population quantity mechanism;If Xie Yishou
Hold back, be then considered as and be optimal distribution, export now distribution of the device file between small-cell base station, and accordingly in small-cell base station
Between distribute buffer memory device, obtain optimal buffer memory device allocative decision, improve the cache hit rate performance of user and effectively save
Equipment is laid to this.
As shown in Fig. 2 concretely comprise the following steps:
Step 1: for some user group in mobile network, the long-term shifting of each user's history in the colony is analyzed
Dynamic rail mark, and the time that all motion tracks occupy is discretized into time slot at regular intervals;
Motion track record is position of each user where in different time groove, as shown in Figure 1.
Step 2: it is directed in each time slot, file service of each user to one unit-sized of network request;
The probability of different files is asked to meet the distribution of file popularity;
Step 3: existed according to the location of user in each time slot, file popularity, buffer memory device and device file
Distribution between small-cell base station, calculate overall cache hit rate.
The mathematical expectation E of the cache hit rate of total usertotal(χ) is as follows:
T is the discrete time slot set of consecutive hours top-stitching division;For the set of service user, quantity is For
The set of the popular file of user's request;Quantity isIt is equal in magnitude;Z (f) is the file popularity size of user;I is finger
Show function, when its expression formula is true, I=1;No I=0.For small set of cells, quantity is Represent user u in the time
The location of during groove τ;For the location sets of all users, binary variable γP, cRepresent user p positions whether
In coverage in base station c, if in γp,c=1;Otherwise, γp,c=0;Binary variable factor χc,fRepresent in the c of base station
Whether file f is stored, if χc,f=1;Otherwise, χc,f=0;
Meanwhile mathematical expectation Etotal(χ) will meet following constraints:
Wherein, C1 represents the All Files stored on the c of base stationIt is less than the caching that can be installed equal to the base station
Equipment maximum capacity CSmax;The buffer memory device maximum capacity CS of all base stationsmaxIt is worth all same.
C2 represents the memory capacity summation that system is distributed for all base stationsBeing less than can distribute equal to system
Total buffer memory device capacity C Stotal;System is that the memory capacity of base station c distribution is CS (c);Unit is a popular file
Size.
Step 4: being based on genetic algorithm and simulated annealing, optimize the mathematical expectation E of overall cache hit ratetotal
(χ), obtain optimum allocation of the buffer memory device on small-cell base station, and optimal placement of the focus file on buffer memory device.
As shown in figure 3, comprise the following steps that:
Step 401, by mathematical expectation EtotalBelong to the binary variable factor χ of same distribution state in (χ)c,fArrangement
For oneOKThe matrix χ of row according to the population quantity of setting, generates several chromosomes, made at random as hereditary chromosome
For initial population.
For the mathematical expectation E of cache hit ratetotal(χ), χ are independent variable;Pass through adjustment
To optimize Etotal(χ);χc,fIt is binary variable, meets the chromosomal foci characteristic in genetic algorithm, without introducing extra coding
Mode.In view of formula (2) and the constraints of formula (3), the present invention uses the hereditary chromosome of matrix form, will belonged to same
The χ variables of distribution state are arranged as oneOKThe matrix of row, is represented with χ.In matrix χ the i-th row jth arrange element χ (i,
J) represent whether store file j on the i of base station;When element χ (i, j) is 1, illustrate to store file j on the i of base station;Otherwise, base
Stand non-storage file on i.
The constraints of formula (2) and formula (3) is converted into:
The All Files stored in C3 representing matrixs χ on i-th of base station is less than the caching that can be installed equal to the base station
Equipment maximum capacity CSmax;Namely chromosome χaWith χbThe All Files stored in corresponding matrix on each base station meets small
In equal to CSmax;
It is assignable equal to system total that C4 represents that the memory capacity summation that system is all base station distribution in matrix χ is less than
Buffer memory device capacity C Stotal;Namely system is matrix χaWith χbIn the distribution of all base stations memory capacity summation meet it is small
In equal to CStotal。
Step 402, current population is set to parent population, calculates the fitness of each chromosome in parent population;
Current Population Size is N, and each chromosome in population is expressed as:χ1,χ2,…χn,...,χN;
S(χn) it is chromosome χnIdeal adaptation angle value, be calculated as follows:
χbestFor the chromosome of caused most defect individual in all previous iteration of genetic algorithm;It is a dynamic value, with repeatedly
The increase of generation number and change.
Step 403, the fitness according to each chromosome, chromosome is randomly choosed using roulette formula, is calculated respectively every
Individual chromosome is chosen to entail the probability of filial generation.
When evolving in per generation, current optimal base is added into filial generation first because of χbest;Then it is directed to chromosome χnIt is chosen heredity
Probability P to filial generationnIt is calculated as follows:
Step 405, when the probability that chromosome is chosen to entail filial generation meets constraints, select the chromosome to carry out
Subsequent operation;
Constraints is:
R is to utilize the real number of random number seed generation, 0 < r < 1;
For chromosome χn, work as probability PnWhen meeting above-mentioned constraints, chromosome χnIt is selected;
Two step 406, selection chromosomes for meeting condition, carry out the intersection and mutation operation of genetic algorithm, obtain two
Individual new chromosome;
Comprise the following steps that:
Step 4061, system generate two random numbers
Step 4062, utilize the two chromosome χ chosen from parent populationa,χb, judgeValue be
No is 1, if it is, exchanging χa(u, v) and χbThe value of (u, v), into step 4063;Otherwise, return to step 4061;
Chromosome χ after step 4063, checking exchangeaWith χbWhether formula (4) and the constraints of (5) are met, if it is,
Mutation operation is carried out into step 4064;Otherwise, return to step 4061;
Step 4064, system produce four random numbers
Step 4065, for the chromosome χ after exchangeaJudge whetherIf it is, exchange
χa(u, v) and χaThe value of (u ', v ') is as chromosome χaVariation result χa', otherwise, return to step 4064.
For the chromosome χ after exchangebEqually operated, obtain chromosome χb'Final variation result.
Corresponding to chromosome χa, crossover probability PcrossWith mutation probability PmutateSetting it is as follows:
Wherein, FavgFor the fitness average value of current population, Pc1For the upper bound of the crossover probability of setting;Pc2For setting
The lower bound of crossover probability, Pm1For the upper bound of the mutation probability of setting;Pm2For the lower bound of the mutation probability of setting.Take P hereinc1
=0.9, Pc2=0.5, Pm1=0.2, Pm2=0.02.And when carrying out crossover operation to two chromosomes, crossover probability is elected as
Two chromosomes correspond to the smaller value of crossover probability.
Step 407, two new chromosome χ to variation gaineda'With χb', disturbance operation is performed respectively, and judges disturbance knot
Whether fruit meets Metropolis criterions, if it is, into step 408, otherwise, disturbance behaviour is re-started to disturbance result
Make.
After the crossover operation for having performed genetic algorithm every time produces new individual with mutation operation, in fixed step-length L,
Random change chromosome χ is operated using disturbance, when chromosome χ temperature is T, calculates solid energy G (χ)=1/F now
(χ), with temperature T decline, in next iteration, solid is in a new state χ ' using disturbing function, calculate this
When solid energy G (χ ').And according to Metropolis criterions and probability PMetropolisChoose whether to receive new state χ ' works
For current solid state.In addition, the disturbance operation that the present invention uses is identical with the mutation operation of the foregoing description.
Probability PMetropolisIt is calculated as follows:
Wherein, Δ G=G (χ ')-G (χ), i.e., the difference of new solid energy and old solid energy.
In simulated annealing link, initial temperature T0For:
Wherein, FminAnd FmaxThe minimum value and maximum of fitness value respectively in initial population, s are one adjustable normal
Number.
Cooling function makes system Current Temperatures T change during each iteration, and the present invention is in index cooling function
On the basis of improve, if it is T ' to set generation temperature, have
Wherein, the present embodiment takes α=0.95, K=1.1, M=20.
Disturbance operation is specific as follows:
Step 4071, system produce four random numbers
Step 4072, for mutated chromosome χa'Judge whetherIf it is, exchange
χa'(u1,v1) and χa(u2,v2) value as treating mutated chromosome χa'Disturbance result, otherwise, return to step 4064.
For the chromosome χ after variationb'Equally operated, obtain chromosome χb'Final disturbance result.
Step 408, judge whether current disturbance number reaches iteration step length, if it is, by chromosome χa'With χb'Disturbance
As a result progeny population is put into, into step 409;Otherwise, return to step 407;
Step 409, judge whether progeny population is equal to adaptive Population Size, if it is, into step 410;Otherwise, hold
Row step 406, continue selective staining body and operated.
Step 410, judge that there is maximum E in current progeny populationtotalWhether the chromosome of (χ) has restrained, if it is,
Terminate and export optimal result, otherwise, return to step 402 carries out new round iteration:
Optimal result is how to arrange buffer memory device on small-cell base station and focus text how is placed on buffer memory device
Part.
Based on genetic algorithm and simulated annealing, optimize the mathematical expectation E of overall cache hit ratetotal(χ's) is total
Body algorithm flow is summarized as follows:
Step 1:Input initialization population invariable number Smax, adaptive minimum population number Smin, auto-adaptive parameter M, initial temperature system
Number s, move back warm factor alpha, scrambling COEFFICIENT K, simulated annealing step-length L, crossover probability Pc1With Pc2, mutation probability Pm1With Pm2。
Step 2:Initialize population and the fitness value of each individual is calculated according to fitness function.And counted according to formula (13)
Calculate initial temperature T0。
Step 3:Current population is set to father group, starts to evolve, first current optimal base because of χbestIt is put into progeny population.
Step 4:Two chromosomes are selected according to selection operation, record its fitness value, and according to crossover probability Pc1With Pc2、
Mutation probability Pm1With Pm2Calculate adaptive crossover and mutation probability PcrossAnd Pmutate, and cross and variation behaviour is carried out according to this probability
Make.
Step 5:The chromosome of gained is done into disturbance operation, calculates new adaptive value, and determine according to Metropolis criterions
It is fixed whether to receive new state.
Step 6:If if current iteration number is less than simulated annealing step-length L, return to step 5, otherwise new chromosome is put
Enter in progeny population.
Step 7:If current progeny population less than, return to step 4, otherwise update current progeny population fitness value, and
Current optimal base is updated because of χbest。
Step 8:Judge current optimal solution χbestWhether restrain, if not converged, return to step 3, otherwise, output are current
Optimal solution χbestAnd terminate algorithm.
At algorithm initial stage, initial population quantity is set as a higher value Smax, can so ensure the gene diversity at initial stage
Property, and in genetic algorithm iterative process, if undergoing M genetic iteration and χbestDo not change, then SmaxReduce 1, until subtracting
It is small to arrive minimum population quantity SminUntill.Thus can reduce the algorithm later stage carried out in the solution space in the absence of optimal solution it is unnecessary
The wasted amount of calculation of search, improve the execution efficiency of algorithm.
Step 5: arranging buffer memory device on small-cell base station according to optimal result, and placed on buffer memory device
Focus file.
Embodiment:
The present embodiment considers following scene:Using 92 experiment users of (KAIST) in Korea Advanced Institute of Science and Technology at 1 month
Interior motion track is as mobile trajectory data source., will be adjacent slight using equally distributed small-cell base station deployment way
Distance between area base station is set to 40 meters, and the covering radius of each small-cell base station signal is set to 100 meters.Each base station can at most pacify
The size of the buffer memory device capacity of dress is 5 file units.In terms of user file request, it is assumed that file popularity coincidence coefficient
It is distributed for α Zipf, and the sum of popular file set is 100.In addition, it is analysis system cache hit rate and total buffer memory capacity
CStotalThe relation of size, it is total buffer memory capacity CS to define buffer memory capacity saturation degree ηtotalWith small-cell base station quantityThe ratio between,
Unit is file, i.e.,
In order to prove set forth herein buffer memory device distribution mechanism performance, selected the buffer memory device of mean allocation to distribute
Mechanism is contrasted.
Mean allocation algorithm:Total buffer memory device is averagely allocated to each small-cell base station, does not consider user mobility.
As shown in figure 4, setting buffer memory capacity saturation degree η=1.As seen from the figure, with Zipf breadth coefficients α increase, two
The cache hit rate performance of kind algorithm also increases therewith, and under different Zipf breadth coefficients α, the caching of genetic annealing algorithms
Hit rate performance suffers from necessarily being lifted compared with mean allocation algorithm.
The cache hit rate performance pair of genetic annealing algorithms and mean allocation algorithm under different buffer memory capacity saturation degrees
Than, as shown in figure 5, wherein, setting Zipf breadth coefficients α=1.It can be found that with the increase of buffer memory capacity saturation degree, two kinds
The cache hit rate performance of algorithm has lifting.Also, compared to mean allocation algorithm, genetic annealing algorithms are in low buffer memory capacity
Performance boost in the case of saturation degree 120%, and the performance boost 80% in the case of high level cache capacity saturation degree.This
It is because genetic annealing algorithms adjust the distribution of buffer memory device according to the long-term mobile trajectory data of user group, in total caching
In the case that place capacity is certain, more buffer memory devices are distributed to the more small-cell base station by access chance, and
Simultaneously in the case where user can access the intensive scene of multiple small-cell base stations, cache contents are carried out with reference to the distribution of file popularity
Optimization.As can be seen here, in the case where buffer memory capacity saturation degree is relatively low, genetic annealing algorithms are compared to mean allocation algorithm
There is more preferable cache hit rate performance boost effect.
The present invention in the distribution of small-cell base station buffer memory device, to optimize distribute, and uses by the mobility for introducing user
Solved based on the algorithm model of genetic simulated annealing, so as to which decision-making goes out optimal allocative decision, improve the cache hit of user
Rate performance, save equipment and be laid to this.
Claims (5)
1. a kind of cell buffer memory device allocation algorithm based on user mobility, it is characterised in that concretely comprise the following steps:
Step 1: for some user group in mobile network, the long-term moving rail of each user's history in the colony is analyzed
Mark, and the time that all motion tracks occupy is discretized into time slot at regular intervals;
Step 2: it is directed in each time slot, file service of each user to one unit-sized of network request;
The probability of different files is asked to meet the distribution of file popularity;
Step 3: according to the location of user in each time slot, file popularity, buffer memory device and device file slight
Distribution between area base station, calculate overall cache hit rate;
The mathematical expectation E of the cache hit rate of total usertotal(χ) is as follows:
T is the discrete time slot set of consecutive hours top-stitching division;For the set of service user;For the prevalence of user's request
The set of file;Z (f) is the file popularity size of user;I is indicator function, when its expression formula is true, I=1;No I=
0;For small set of cells;Represent user u the location of in time slot τ; For the position collection of all users
Close, binary variable γp,cRepresent whether user is in base station c coverage in p positions, if in γp,c=1;Otherwise,
γp,c=0;Binary variable factor χc,fRepresent whether store file f in the c of base station, if χc,f=1;Otherwise, χc,f=0;
Step 4: being based on genetic algorithm and simulated annealing, optimize the mathematical expectation E of overall cache hit ratetotal(χ),
Obtain optimum allocation of the buffer memory device on small-cell base station, and optimal placement of the focus file on buffer memory device;
Comprise the following steps that:
Step 401, by mathematical expectation EtotalBelong to the binary variable factor χ of same distribution state in (χ)c,fIt is arranged as one
It is individualOKThe matrix χ of row, according to the population quantity of setting, generates several chromosomes, as first at random as hereditary chromosome
Beginning population;
File j whether is stored on the element representation base station i that the i-th row jth arranges in matrix χ;When element χ (i, j) is 1, explanation
File j is stored on the i of base station;Otherwise, non-storage file on the i of base station;
Step 402, current population is set to parent population, calculates the fitness of each chromosome in parent population;
Current Population Size is N, and the chromosome in population is expressed as:χ1,χ2,…χn,...,χN;
S(χn) it is chromosome χnFitness value, be calculated as follows:
<mrow>
<mi>S</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>&chi;</mi>
<mi>n</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mrow>
<mo>(</mo>
<mfrac>
<mrow>
<msub>
<mi>E</mi>
<mrow>
<mi>t</mi>
<mi>o</mi>
<mi>t</mi>
<mi>a</mi>
<mi>l</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>&chi;</mi>
<mi>n</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>E</mi>
<mrow>
<mi>t</mi>
<mi>o</mi>
<mi>t</mi>
<mi>a</mi>
<mi>l</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<msup>
<mi>&chi;</mi>
<mrow>
<mi>b</mi>
<mi>e</mi>
<mi>s</mi>
<mi>t</mi>
</mrow>
</msup>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
χbestFor the chromosome of caused most defect individual in all previous iteration of genetic algorithm;Change with the increase of iterations;
Step 403, the fitness according to each chromosome, chromosome is randomly choosed using roulette formula, calculates each dye respectively
Colour solid is chosen to entail the probability of filial generation;
For chromosome χnIt is chosen to entail the probability P of filial generationnIt is calculated as follows:
<mrow>
<msub>
<mi>P</mi>
<mi>n</mi>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<mi>S</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>&chi;</mi>
<mi>n</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>n</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<mi>S</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>&chi;</mi>
<mi>n</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
Step 405, when the probability that chromosome is chosen to entail filial generation meets constraints, select the chromosome to carry out follow-up
Operation;
Constraints is:
R is to utilize the real number of random number seed generation, 0 < r < 1;
For chromosome χn, work as probability PnWhen meeting above-mentioned constraints, chromosome χnIt is selected;
Two step 406, selection chromosomes for meeting condition, are intersected and mutation operation, obtain two new chromosomes;
Step 407, two new chromosome χ to variation gaineda'With χb', disturbance operation is performed respectively, and judges that disturbing result is
It is no to meet Metropolis criterions, if it is, into step 408, otherwise, disturbance operation is re-started to disturbance result;
Step 408, judge whether current disturbance number reaches iteration step length, if it is, by chromosome χa'With χb'Disturbance result
Progeny population is put into, into step 409;Otherwise, return to step 407;
Step 409, judge whether progeny population is equal to adaptive Population Size, if it is, into step 410;Otherwise, step is performed
Rapid 406, continue selective staining body and operated;
Step 410, judge that there is maximum E in current progeny populationtotalWhether the chromosome of (χ) has restrained, if it is, terminating
And optimal result is exported, otherwise, return to step 402 carries out new round iteration:
Optimal result is how to arrange buffer memory device on small-cell base station and focus file how is placed on buffer memory device;
Step 5: arranging buffer memory device on small-cell base station according to optimal result, and focus is placed on buffer memory device
File.
A kind of 2. cell buffer memory device allocation algorithm based on user mobility as claimed in claim 1, it is characterised in that
What the motion track described in step 1 recorded is position of each user where in different time groove.
A kind of 3. cell buffer memory device allocation algorithm based on user mobility as claimed in claim 1, it is characterised in that
Mathematical expectation E described in step 3total(χ) will meet following constraints:
Wherein, C1 represents the All Files stored on the c of base stationIt is less than the buffer memory device that can be installed equal to the base station
Maximum capacity CSmax;The buffer memory device maximum capacity CS of all base stationsmaxIt is worth all same;
C2 represents the memory capacity summation that system is distributed for all base stationsIt is less than assignable equal to system total
Buffer memory device capacity C Stotal;System is that the memory capacity of base station c distribution is CS (c).
A kind of 4. cell buffer memory device allocation algorithm based on user mobility as claimed in claim 1, it is characterised in that
Described step 406 concrete operations are as follows:
Step 4061, system generate two random numbers
Step 4062, utilize the two chromosome χ chosena,χb, judgeValue whether be 1, if it is,
Then exchange χa(u, v) and χbThe value of (u, v), into step 4063;Otherwise, return to step 4061;
Chromosome χ after step 4063, checking exchangeaWith χbWhether constraints is met, if it is, being carried out into step 4064
Mutation operation;Otherwise, return to step 4061;
Constraints is as follows:
The All Files stored in C3 representing matrixs χ on i-th of base station is less than the buffer memory device that can be installed equal to the base station
Maximum capacity CSmax;Namely chromosome χaWith χbThe All Files that is stored in corresponding matrix on each base station meets to be less than etc.
In CSmax;
C4 represents that the memory capacity summation that system is all base station distribution in matrix χ is less than assignable equal to system total delay
Deposit place capacity CStotal;Namely system is matrix χaWith χbIn the memory capacity summation of all base stations distribution meet to be less than etc.
In CStotal;
Step 4064, system produce four random numbers
Step 4065, for the chromosome χ after exchangeaJudge whetherIf it is, exchange χa
(u, v) and χaThe value of (u ', v ') is as chromosome χaVariation result χa', otherwise, return to step 4064;
For the chromosome χ after exchangebEqually operated, obtain chromosome χb'Final variation result.
A kind of 5. cell buffer memory device allocation algorithm based on user mobility as claimed in claim 1, it is characterised in that
Disturbance operation described in step 407 is specific as follows:
First, system produces four random numbers
Then, for mutated chromosome χa'Judge whetherIf it is, exchange χa'(u1,v1)
With χa(u2,v2) value as treating mutated chromosome χa'Disturbance result, otherwise, re-start mutation operation;
For the chromosome χ after variationb'Equally operated, obtain chromosome χb'Final disturbance result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710936023.7A CN107466016B (en) | 2017-10-10 | 2017-10-10 | A kind of cell buffer memory device distribution method based on user mobility |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710936023.7A CN107466016B (en) | 2017-10-10 | 2017-10-10 | A kind of cell buffer memory device distribution method based on user mobility |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107466016A true CN107466016A (en) | 2017-12-12 |
CN107466016B CN107466016B (en) | 2019-11-05 |
Family
ID=60554224
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710936023.7A Active CN107466016B (en) | 2017-10-10 | 2017-10-10 | A kind of cell buffer memory device distribution method based on user mobility |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107466016B (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108112039A (en) * | 2018-02-05 | 2018-06-01 | 东南大学 | A kind of isomery cellular network cache policy based on re-transmission and user's movement |
CN108174395A (en) * | 2018-01-15 | 2018-06-15 | 南京邮电大学 | Node B cache management method and system based on transfer action assessment learning framework |
CN108540549A (en) * | 2018-04-03 | 2018-09-14 | 西南交通大学 | A kind of ambulant network edge caching selection method of user oriented |
CN108551472A (en) * | 2018-03-20 | 2018-09-18 | 南京邮电大学 | A kind of content caching optimization method based on edge calculations |
CN108738048A (en) * | 2018-04-25 | 2018-11-02 | 杭州电子科技大学 | A kind of maximization fairness base station Initiative Inventory Management method based on genetic algorithm |
CN108769729A (en) * | 2018-05-16 | 2018-11-06 | 东南大学 | Caching arrangement system based on genetic algorithm and caching method |
CN109413694A (en) * | 2018-09-10 | 2019-03-01 | 北京邮电大学 | A kind of cell caching method and device based on content popularit prediction |
CN109714790A (en) * | 2019-01-23 | 2019-05-03 | 南京邮电大学 | A kind of edge cooperation caching optimization method based on user mobility prediction |
CN112468597A (en) * | 2020-12-11 | 2021-03-09 | 深圳市知小兵科技有限公司 | Data center resource allocation method and device based on artificial intelligence |
CN117785949A (en) * | 2024-02-28 | 2024-03-29 | 云南省地矿测绘院有限公司 | Data caching method, electronic equipment, storage medium and device |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101141399A (en) * | 2007-10-16 | 2008-03-12 | 武汉烽火网络有限责任公司 | Dual-threshold port queue control method for sharing storage exchange supporting multicast |
CN102855153A (en) * | 2012-07-27 | 2013-01-02 | 华中科技大学 | Flow compilation optimization method oriented to chip multi-core processor |
CN106912079A (en) * | 2017-02-20 | 2017-06-30 | 北京邮电大学 | Federated user accesses selection and resource allocation methods in one kind caching heterogeneous network |
-
2017
- 2017-10-10 CN CN201710936023.7A patent/CN107466016B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101141399A (en) * | 2007-10-16 | 2008-03-12 | 武汉烽火网络有限责任公司 | Dual-threshold port queue control method for sharing storage exchange supporting multicast |
CN102855153A (en) * | 2012-07-27 | 2013-01-02 | 华中科技大学 | Flow compilation optimization method oriented to chip multi-core processor |
CN106912079A (en) * | 2017-02-20 | 2017-06-30 | 北京邮电大学 | Federated user accesses selection and resource allocation methods in one kind caching heterogeneous network |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108174395B (en) * | 2018-01-15 | 2020-10-20 | 南京邮电大学 | Base station cache management method and system based on transfer action evaluation learning framework |
CN108174395A (en) * | 2018-01-15 | 2018-06-15 | 南京邮电大学 | Node B cache management method and system based on transfer action assessment learning framework |
CN108112039A (en) * | 2018-02-05 | 2018-06-01 | 东南大学 | A kind of isomery cellular network cache policy based on re-transmission and user's movement |
CN108551472A (en) * | 2018-03-20 | 2018-09-18 | 南京邮电大学 | A kind of content caching optimization method based on edge calculations |
CN108551472B (en) * | 2018-03-20 | 2021-01-19 | 南京邮电大学 | Content cache optimization method based on edge calculation |
CN108540549A (en) * | 2018-04-03 | 2018-09-14 | 西南交通大学 | A kind of ambulant network edge caching selection method of user oriented |
CN108540549B (en) * | 2018-04-03 | 2020-12-22 | 西南交通大学 | User mobility-oriented network edge cache selection method |
CN108738048A (en) * | 2018-04-25 | 2018-11-02 | 杭州电子科技大学 | A kind of maximization fairness base station Initiative Inventory Management method based on genetic algorithm |
CN108738048B (en) * | 2018-04-25 | 2021-02-26 | 杭州电子科技大学 | Active storage method of maximized fairness base station based on genetic algorithm |
CN108769729B (en) * | 2018-05-16 | 2021-01-05 | 东南大学 | Cache arrangement system and cache method based on genetic algorithm |
CN108769729A (en) * | 2018-05-16 | 2018-11-06 | 东南大学 | Caching arrangement system based on genetic algorithm and caching method |
CN109413694A (en) * | 2018-09-10 | 2019-03-01 | 北京邮电大学 | A kind of cell caching method and device based on content popularit prediction |
CN109714790A (en) * | 2019-01-23 | 2019-05-03 | 南京邮电大学 | A kind of edge cooperation caching optimization method based on user mobility prediction |
CN112468597A (en) * | 2020-12-11 | 2021-03-09 | 深圳市知小兵科技有限公司 | Data center resource allocation method and device based on artificial intelligence |
CN112468597B (en) * | 2020-12-11 | 2021-05-28 | 深圳市知小兵科技有限公司 | Data center resource allocation method and device based on artificial intelligence |
CN117785949A (en) * | 2024-02-28 | 2024-03-29 | 云南省地矿测绘院有限公司 | Data caching method, electronic equipment, storage medium and device |
CN117785949B (en) * | 2024-02-28 | 2024-05-10 | 云南省地矿测绘院有限公司 | Data caching method, electronic equipment, storage medium and device |
Also Published As
Publication number | Publication date |
---|---|
CN107466016B (en) | 2019-11-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107466016A (en) | A kind of cell buffer memory device allocation algorithm based on user mobility | |
Yang et al. | Deep neural network for resource management in NOMA networks | |
CN104158855B (en) | Information Mobile Service combination based on genetic algorithm calculates discharging method | |
CN102014412B (en) | Wireless network telephone traffic balanced optimization method based on antenna parameter adjustment | |
CN105050192B (en) | Interference management method based on virtual subdistrict in dense network | |
CN109862573B (en) | LTE hybrid networking self-planning method based on multi-target particle swarm | |
CN108667653A (en) | Cached configuration method and device based on cluster in super-intensive network | |
CN107659973A (en) | Super-intensive network cluster dividing method based on density K means algorithms | |
CN102026240A (en) | Scrambling code optimizing method based on measurement report | |
CN102448071B (en) | Cognitive network power distribution method based on interference temperature | |
CN109391511A (en) | It is a kind of based on can outward bound network intelligence communication resource allocation policy | |
Liu et al. | Resource consumption for supporting federated learning in wireless networks | |
CN112887943B (en) | Cache resource allocation method and system based on centrality | |
CN103984828A (en) | Uniform-temperature core mapping method and system for three-dimensional network on chip | |
CN106060876A (en) | Load balancing method for heterogeneous wireless network | |
CN104159279A (en) | Base station based on energy efficiency norm and relay on/off selection system and method | |
CN109981340A (en) | The method that mist calculates joint optimization of resources in network system | |
CN108600365A (en) | A kind of Wireless Heterogeneous Networks caching method based on sequence study | |
CN104486806A (en) | Carrying efficiency-based heterogeneous network combined carrying method and device | |
CN108260193A (en) | Federated resource distribution method based on channel aggregation in heterogeneous network | |
CN110753365A (en) | Heterogeneous cellular network interference coordination method | |
CN107026450A (en) | B2G and V2G power distribution networks power control system and method based on genetic algorithm | |
CN107171712B (en) | Method for selecting transmitting terminal transmitting antenna in large-scale multi-input multi-output system | |
CN113301576B (en) | Cellular network resource allocation method based on improved genetic algorithm | |
CN103945388A (en) | User network accessing method in heterogeneous network based on genetic algorithm |
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 |