Summary of the invention
In view of this, an embodiment of the present invention is intended to provide a kind of edge cache management method, personal cloud system and computersReadable storage medium storing program for executing.
In order to achieve the above objectives, the technical solution of the embodiment of the present invention is achieved in that
The embodiment of the invention provides a kind of edge cache management methods, this method comprises:
It obtains user and perceives big data, service-aware big data and network aware big data；
Big data is perceived based on the user to carry out in advance the motion track of different time scales, different spaces granularity userIt surveys, excavates, predicts and be associated with business potential demand based on behavior pattern of the service-aware big data to user；
Behavior pattern based on the different time scales, the motion track of different spaces granularity user and the userWith business potential demand, predict user's future in the business demand of space-time bidimensional；
Based on user's future in the business demand and the network aware big data of space-time bidimensional, setting edge delaysIt deposits.
Wherein, described that movement of the big data to different time scales, different spaces granularity user is perceived based on the userTrack is predicted, including following either type or combinations thereof:
Under idle state or cell re-selection state, user location excavate according to base station relevant location information and pre-It surveys；
Under cell switching state, the relevant parameter of determining user's motion track is got ready based on positioning, be based on the correlationParameter settling time prism；
By matching, being associated with the mobile relevant information of user by the mobile simple space-time trajectory of user, and excavateHereafter semantic information.
Wherein, described to be dug based on behavior pattern of the service-aware big data to user with business potential demandPick, prediction and association, including following either type or combinations thereof:
The user tag of personal cloud is set, and is drawn a portrait to user；
By deep message detection and spiders technology, depth excavation is carried out to user's internet log, thus it is speculated that user'sOnline habit and interest preference excavate user to the potential demand of business；
Multivariate joint probability prediction is carried out based on temporal information, location information and business information.
Wherein, described by deep message detection and spiders technology, depth excavation, packet are carried out to user's internet logIt includes but is not limited to such as under type:
Parse identification user based on address base and often use application type, and by crawler technology realize the automatic identification of label withIt generates；
Deep analysis user's internet behavior, business branch, label or the channel that identification is accessed using user under group；
Identify the content identification of user's access in content type；
Identify user to the access state of a content title.
Wherein, described that multivariate joint probability prediction is carried out based on temporal information, location information and business information, including but not limited toSuch as under type:
Using time series, spatial sequence as independent variable, based on multi-dimensional data model prediction business making in space-time bidimensionalUse mode；
Multidimensional prediction is subjected to dimension-reduction treatment, establishes the relevant trajectory predictions in user and time, place and user respectivelyTraffic forecast relevant to personal preference, business preference；Space-time two is generated based on the trajectory predictions, traffic forecast and major keyTie up contingency table.
In the embodiment of the present invention, it is described based on user's future space-time bidimensional business demand and the network awareEdge cache, including but not limited to following content is arranged in big data:
The size and storage content, the update mode of the edge cache, the edge cache of the edge cache are setWays of distribution, the deployed position of the edge cache, the way to manage between the different edge cache.
Wherein, the edge cache includes two regions, is respectively as follows: based on file popularity, towards the public slow of crowd's needDeposit region, and based on personal preference, towards the personal buffer zone of a need；
When the edge cache is deployed in network center, the public buffer zone is greater than the personal buffer zone；InstituteWhen stating edge cache and being deployed in network edge, the public buffer zone is less than personal buffer zone.
Wherein, it is mobile special to update using the periodic based on user time characteristic and be based on user for the personal buffer zoneProperty event-triggered update the update mode that combines；
It is described individual buffer zone using based on application type, content type, content title and content progress orientation, pointPiece, intelligent ways of distribution.
Wherein, when business demand is lower than the whole network average threshold level, the deployed position of the edge cache is in networkThe heart；
When business demand is higher than the whole network average threshold level, the deployed position of the edge cache is network edge.
Wherein, the way to manage between the difference edge cache includes:
Layering centralized management, and/or distributed collaboration management, and/or hybrid management can be used.
The embodiment of the invention also provides a kind of personal cloud systems, which includes: two or more people's cloud netsMember, individual's cloud network element includes: resource pool, individual's cloud network element further include:
Transceiver perceives big data, service-aware big data and network aware big data for obtaining user；
Processor, for perceiving shifting of the big data to different time scales, different spaces granularity user based on the userDynamic rail mark predicted, excavated based on behavior pattern and business potential demand of the service-aware big data to user,Prediction and association；
It is also used to the row based on the different time scales, the motion track of different spaces granularity user and the userFor mode and business potential demand, predict user's future in the business demand of space-time bidimensional；
It is also used to the business demand and the network aware big data based on user's future in space-time bidimensional, side is setEdge caching.
Wherein, the different personal cloud network elements are set to access layer, and/or convergence layer, and/or core layer；
Each personal cloud network element of different levels constitutes differentiated control, master-slave network framework, and the topological structure of composition isDendrogram；
Each personal cloud network element of same level constitutes equity, autonomous management the network architecture, and the topological structure of composition isStar-plot or cyclic annular figure；
It is interactive or private network is direct-connected by public network between different personal cloud network elements.
Wherein, the personal cloud network element is independent server, and/or is integrated in other nets in addition to personal cloud network elementIn member.
Wherein, the personal cloud network element is located at convergence layer and core layer and the personal cloud network element is independent serverWhen, individual's cloud network element and following interactive interfacing:
The interface of uniform depth packet detection system；The interface of strategy and charging control system；The interface of gateway.
The embodiment of the invention also provides a kind of personal cloud systems, which includes: two or more people's cloud netsMember, individual's cloud network element includes: processor and the memory for storing the computer program that can be run on a processor,
Wherein, the step of processor is for executing the above method when running the computer program.
The embodiment of the invention also provides a kind of computer readable storage mediums, are stored thereon with computer program, the meterThe step of above method is realized when calculation machine program is executed by processor.
Edge cache management method, personal cloud system and computer readable storage medium provided in an embodiment of the present invention, are obtainedTake family perception big data, service-aware big data and network aware big data；Big data is perceived to difference based on the userTime scale, different spaces granularity user motion track predict, based on the service-aware big data to the row of userIt excavated, predicted and is associated with business potential demand for mode；Based on the different time scales, different spaces granularity userMotion track and the user behavior pattern and business potential demand, prediction user's future space-time bidimensional business needIt asks；Based on user's future in the business demand and the network aware big data of space-time bidimensional, edge cache is set.This hairIndividual's cloud system (network element) described in bright embodiment can be deployed to the different layers of network according to the evolution of business needs and the network architectureSecondary, deployed position and deployment way are more flexible；Mobility and personalization of the edge cache based on user class granularity can be tracedThe moving condition of user simultaneously provides edge service whenever and wherever possible, can accurately match individual subscriber demand and go deep into excavating in " long-tail "The value of appearance, the intelligent pipeline for giving full play to operator in wireless big data field are acted on, are finally brought preferably to user" end-pipe-cloud " experience.
Specific embodiment
Present invention is described with reference to the accompanying drawings and examples.
The embodiment of the invention provides a kind of edge cache management methods, as shown in Figure 1, this method comprises:
Step 101: obtaining user and perceive big data, service-aware big data and network aware big data；
Step 102: big data is perceived to the moving rail of different time scales, different spaces granularity user based on the userMark is predicted, is excavated, is predicted with business potential demand based on behavior pattern of the service-aware big data to userAnd association；
Step 103: based on the different time scales, the motion track of different spaces granularity user and the userBehavior pattern and business potential demand, business demand of the prediction user's future in space-time bidimensional；
Step 104: business demand and the network aware big data based on user's future in space-time bidimensional, settingEdge cache.
It is described that big data is perceived to different time scales, different spaces granularity based on the user in the embodiment of the present inventionThe motion track of user is predicted, including following either type or combinations thereof:
Under idle state or cell re-selection state, according to base station relevant location information (such as: GPS longitude and latitude, cell ID,Such as serving cell Cell ID, ECI, 3 signal strengths, orientation and deflection (TA+AoA), the base station position Gong Can etc.) toIt is excavated and is predicted in family position；
Under cell switching state, based on positioning get ready determining user's motion track relevant parameter (as: motion trackStarting point, terminal, path, speed, approach time, residence time etc.), it is based on the relevant parameter settling time prism；
(such as: calendar, weather, map, traffic by the simple space-time trajectory for moving user and the mobile relevant information of userMode, approach/stop, permanent residence etc.) it matched, be associated with, and excavate context semantic information.
In the embodiment of the present invention, it is described based on the service-aware big data to the behavior pattern of user and the potential need of businessIt asks and is excavated, predicted and be associated with, including following either type or combinations thereof:
The user tag of personal cloud is set, and is drawn a portrait to user；
By deep message detection and spiders technology, depth excavation is carried out to user's internet log, thus it is speculated that user'sOnline habit and interest preference excavate user to the potential demand of business；
Multivariate joint probability prediction is carried out based on temporal information, location information and business information.
It is described by deep message detection and spiders technology in the embodiment of the present invention, user's internet log is carried outDepth is excavated, including but not limited to such as under type:
Parse identification user based on address base and often use application type, and by crawler technology realize the automatic identification of label withIt generates；
Deep analysis user's internet behavior, business branch, label or the channel that identification is accessed using user under group；
Identify the content identification of user's access in content type；
Identify user to the access state of a content title.
It is described that multivariate joint probability prediction, packet are carried out based on temporal information, location information and business information in the embodiment of the present inventionIt includes but is not limited to such as under type:
Using time series, spatial sequence as independent variable, based on multi-dimensional data model prediction business making in space-time bidimensionalUse mode；
Multidimensional prediction is subjected to dimension-reduction treatment, establishes the relevant trajectory predictions in user and time, place and user respectivelyTraffic forecast relevant to personal preference, business preference；Space-time two is generated based on the trajectory predictions, traffic forecast and major keyTie up contingency table.
In the embodiment of the present invention, it is described based on user's future space-time bidimensional business demand and the network awareEdge cache, including but not limited to following content is arranged in big data:
The size and storage content, the update mode of the edge cache, the edge cache of the edge cache are setWays of distribution, the deployed position of the edge cache, the way to manage between the different edge cache.
In the embodiment of the present invention, the edge cache includes two regions, is respectively as follows: based on file popularity, towards crowdNeed public buffer zone, and based on personal preference, towards the personal buffer zone of a need；
When the edge cache is deployed in network center, the public buffer zone is greater than the personal buffer zone；InstituteWhen stating edge cache and being deployed in network edge, the public buffer zone is less than personal buffer zone.
In the embodiment of the present invention, the individual buffer zone is updated and is based on using the periodic based on user time characteristicThe event-triggered of user's mobility updates the update mode combined；
It is described individual buffer zone using based on application type, content type, content title and content progress orientation, pointPiece, intelligent ways of distribution.
In the embodiment of the present invention, when business demand is lower than the whole network average threshold level, the deployment position of the edge cacheIt is set to network center；
When business demand is higher than the whole network average threshold level, the deployed position of the edge cache is network edge.
In the embodiment of the present invention, the way to manage between the difference edge cache includes:
Layering centralized management, and/or distributed collaboration management, and/or hybrid management can be used.
The embodiment of the invention also provides a kind of personal cloud systems, which includes: two or more people's cloud netsMember；As shown in Fig. 2, individual's cloud network element includes: resource pool 201, individual's cloud network element further include:
Transceiver 202 perceives big data, service-aware big data and network aware big data for obtaining user；
Processor 203, for perceiving big data to different time scales, different spaces granularity users based on the userMotion track is predicted, is dug based on behavior pattern of the service-aware big data to user with business potential demandPick, prediction and association；
It is also used to the row based on the different time scales, the motion track of different spaces granularity user and the userFor mode and business potential demand, predict user's future in the business demand of space-time bidimensional；
It is also used to the business demand and the network aware big data based on user's future in space-time bidimensional, side is setEdge caching.
In the embodiment of the present invention, the different personal cloud network elements are set to access layer, and/or convergence layer, and/or coreLayer；
Each personal cloud network element of different levels constitutes differentiated control, master-slave network framework, and the topological structure of composition isDendrogram；
Each personal cloud network element of same level constitutes equity, autonomous management the network architecture, and the topological structure of composition isStar-plot or cyclic annular figure；
It is interactive or private network is direct-connected by public network between different personal cloud network elements.
In the embodiment of the present invention, individual's cloud network element be independent server, and/or be integrated in except personal cloud network element itOn other outer network elements.
In the embodiment of the present invention, it is only that individual's cloud network element, which is located at convergence layer and core layer and the personal cloud network element,When vertical server, individual's cloud network element and following interactive interfacing:
The interface of uniform depth packet detection system；The interface of strategy and charging control system；The interface of gateway.
The embodiment of the invention also provides a kind of personal cloud systems, which includes: two or more people's cloud netsMember, individual's cloud network element includes: processor and the memory for storing the computer program that can be run on a processor,
Wherein, the processor is for executing when running the computer program:
It obtains user and perceives big data, service-aware big data and network aware big data；
Big data is perceived based on the user to carry out in advance the motion track of different time scales, different spaces granularity userIt surveys, excavates, predicts and be associated with business potential demand based on behavior pattern of the service-aware big data to user；
Behavior pattern based on the different time scales, the motion track of different spaces granularity user and the userWith business potential demand, predict user's future in the business demand of space-time bidimensional；
Based on user's future in the business demand and the network aware big data of space-time bidimensional, setting edge delaysIt deposits.
When the processor is also used to run the computer program, following either type or combinations thereof is executed:
Under idle state or cell re-selection state, according to base station relevant location information (such as: GPS longitude and latitude, cell ID,Such as serving cell Cell ID, ECI, 3 signal strengths, orientation and deflection (TA+AoA), the base station position Gong Can etc.) toIt is excavated and is predicted in family position；
Under cell switching state, based on positioning get ready determining user's motion track relevant parameter (as: motion trackStarting point, terminal, path, speed, approach time, residence time etc.), it is based on the relevant parameter settling time prism；
(such as: calendar, weather, map, traffic by the simple space-time trajectory for moving user and the mobile relevant information of userMode, approach/stop, permanent residence etc.) it matched, be associated with, and excavate context semantic information.
When the processor is also used to run the computer program, following either type or combinations thereof is executed:
The user tag of personal cloud is set, and is drawn a portrait to user；
By deep message detection and spiders technology, depth excavation is carried out to user's internet log, thus it is speculated that user'sOnline habit and interest preference excavate user to the potential demand of business；
Multivariate joint probability prediction is carried out based on temporal information, location information and business information.
When the processor is also used to run the computer program, executing includes but is not limited to such as under type:
Parse identification user based on address base and often use application type, and by crawler technology realize the automatic identification of label withIt generates；
Deep analysis user's internet behavior, business branch, label or the channel that identification is accessed using user under group；
Identify the content identification of user's access in content type；
Identify user to the access state of a content title.
When the processor is also used to run the computer program, executing includes but is not limited to such as under type:
Using time series, spatial sequence as independent variable, based on multi-dimensional data model prediction business making in space-time bidimensionalUse mode；
Multidimensional prediction is subjected to dimension-reduction treatment, establishes the relevant trajectory predictions in user and time, place and user respectivelyTraffic forecast relevant to personal preference, business preference；Space-time two is generated based on the trajectory predictions, traffic forecast and major keyTie up contingency table.
When the processor is also used to run the computer program, executing includes but is not limited to following content:
The size and storage content, the update mode of the edge cache, the edge cache of the edge cache are setWays of distribution, the deployed position of the edge cache, the way to manage between the different edge cache.
The embodiment of the invention also provides a kind of computer readable storage mediums, are stored thereon with computer program, the meterWhen calculation machine program is run by processor, execute:
It obtains user and perceives big data, service-aware big data and network aware big data；
Big data is perceived based on the user to carry out in advance the motion track of different time scales, different spaces granularity userIt surveys, excavates, predicts and be associated with business potential demand based on behavior pattern of the service-aware big data to user；
Behavior pattern based on the different time scales, the motion track of different spaces granularity user and the userWith business potential demand, predict user's future in the business demand of space-time bidimensional；
Based on user's future in the business demand and the network aware big data of space-time bidimensional, setting edge delaysIt deposits.
When the computer program is run by processor, following either type or combinations thereof is also executed:
Under idle state or cell re-selection state, according to base station relevant location information (such as: GPS longitude and latitude, cell ID,Such as serving cell Cell ID, ECI, 3 signal strengths, orientation and deflection (TA+AoA), the base station position Gong Can etc.) toIt is excavated and is predicted in family position；
Under cell switching state, based on positioning get ready determining user's motion track relevant parameter (as: motion trackStarting point, terminal, path, speed, approach time, residence time etc.), it is based on the relevant parameter settling time prism；
(such as: calendar, weather, map, traffic by the simple space-time trajectory for moving user and the mobile relevant information of userMode, approach/stop, permanent residence etc.) it matched, be associated with, and excavate context semantic information.
When the computer program is run by processor, following either type or combinations thereof is also executed:
The user tag of personal cloud is set, and is drawn a portrait to user；
By deep message detection and spiders technology, depth excavation is carried out to user's internet log, thus it is speculated that user'sOnline habit and interest preference excavate user to the potential demand of business；
Multivariate joint probability prediction is carried out based on temporal information, location information and business information.
When the computer program is run by processor, also executing includes but is not limited to such as under type:
Parse identification user based on address base and often use application type, and by crawler technology realize the automatic identification of label withIt generates；
Deep analysis user's internet behavior, business branch, label or the channel that identification is accessed using user under group；
Identify the content identification of user's access in content type；
Identify user to the access state of a content title.
When the computer program is run by processor, also executing includes but is not limited to such as under type:
Using time series, spatial sequence as independent variable, based on multi-dimensional data model prediction business making in space-time bidimensionalUse mode；
Multidimensional prediction is subjected to dimension-reduction treatment, establishes the relevant trajectory predictions in user and time, place and user respectivelyTraffic forecast relevant to personal preference, business preference；Space-time two is generated based on the trajectory predictions, traffic forecast and major keyTie up contingency table.
When the computer program is run by processor, also executing includes but is not limited to following content:
The size and storage content, the update mode of the edge cache, the edge cache of the edge cache are setWays of distribution, the deployed position of the edge cache, the way to manage between the different edge cache.
Below with reference to scene embodiment, the present invention will be described in detail.
Under existing net and Future network architectures, the demand at edge is sunk to based on content, the embodiment of the present invention proposes aDeployed position, networking mode, implementation, function structure, communication interface, the data with other systems and network element of people's cloud network elementInteractive process etc., specific as follows:
Relative to the deployment strategy of existing content network and edge calculations node fixed single, individual's cloud of the embodiment of the present inventionSystem deployed position in a network and networking mode very flexibly, elasticity, be easily achieved, be specifically deployed in network whichNet topology is organized on layer, which node, using which kind of, depends on business demand, Internet resources, service ability, operation cost etc.Composite factor.Typical case scene and general Arranging principles by the personal cloud system of wireless big data combing, so that it is determined thatThe deployed position and networking mode of edge cache strategy carrying node.
As shown in figure 3, Metropolitan Area Network (MAN) is simplified to access layer, convergence layer, three layers of core layer, then personal Cloud Server is (personalCloud network element) it can be deployed in the interdependent node in any layer, it is specifically including but not limited to:
Access layer: being suitable for the small ranges business scenarios such as house, office, coffee shop, shop, such as:
1) Cellular Networks macro base station (such as BTS/nodeB/eNodeB)
Personal Cloud Server can be deployed at the single base station of access layer, or be autonomous device, or by increasing base station newlyCard or software upgrading mode are integrated in inside of base station；
2) Cellular Networks distribution unit (Distributed Unit, DU)
With in the following 5G/5G+ network, BBU can functionally be divided into CU (Centralized Unit) and DU(Distributed Unit) two parts, personal Cloud Server can be deployed at the DU of access layer, or be autonomous device, or and DUNode is integrated.
Convergence layer: it is suitable for work/enterprise campus, campus, stadiums, large-scale store, rail traffic (along subway) etc.Medium range business scenario, such as:
1) WLAN wireless controller (Access Controller, AC)
Personal Cloud Server can be deployed at the AC of convergence layer, or be autonomous device, or be integrated in AC equipment；
2) access gatewaies such as the small base station of Cellular Networks, Internet of Things, car networking, industry internet
Personal Cloud Server can be deployed in various access gatewaies (such as the room subnetting pass, home gateway, local monitor of convergence layerGateway etc.) at, or be autonomous device, or integrated with gateway；
3) base station controller (BSC/RNC), interchanger
Personal Cloud Server can be deployed at the aggregation node of the multiple base stations of convergence layer, or be autonomous device, or and base stationController, access switch, convergence switch are integrated；
4) Cellular Networks centralized unit (Centralized Unit, CU)
With in the following 5G/5G+ network, BBU can functionally be divided into CU (Centralized Unit) and DU(Distributed Unit) two parts, personal Cloud Server can be deployed at the CU of convergence layer, or be autonomous device, or and CUNode is integrated.
Core layer: being suitable for prefecture-level, provincial, big administrative region (such as East China, North China, south China) a wide range of business scenario,Such as:
1) gateway (SGSN/SGW)
Personal Cloud Server can be deployed at core layer gateway (SGSN/SGW), or be autonomous device, or and districts and citiesGrade anchor point is integrated；
2) egress gateways (GGSN/PGW)
Personal Cloud Server can be deployed at core layer egress gateways (GGSN/PGW), or be autonomous device, or with it is provincialAnchor point is integrated.
Fig. 3 illustrates two kinds of typical personal cloud system networking modes from longitudinally, laterally two dimensions:
For longitudinal dimension, differentiated control is constituted by each carrying network element of different levels (core layer/convergence layer/access layer), master-slave network framework, topological structure is dendrogram, passes through public network between personal cloud service node (personal Cloud Server)Interactive or private network is direct-connected；
For transverse dimensions, each carrying network element by being located on the same floor (such as convergence layer) constitutes P2P equity, autonomous managementThe network architecture, topological structure is star-plot or cyclic annular figure, interactive by public network between personal cloud service node or private network is straightEven.
According to the above-mentioned elaboration to personal cloud network element deployment position, the specific implementation of personal cloud network element and function structure can divideFor but be not limited to following several types:
1) personal cloud network element is separate server
As shown in figure 4, by taking personal cloud network element deployment is in convergence layer and core layer as an example, it can be seen that independent individual's cloud clothesBusiness device (Personalized Cloud Server, PCS) can be both set up on the access ring or convergence ring of transmission network, can alsoTo be set up between SGW, PGW of core net.PCS and the interactive interfacing of other systems and network element are as follows:
The interactive interfacing of PCS and uniform depth packet detection (Deep Packet Inspection, DPI) system
The problem of in view of time delay, by ticket writing (the X Data of Uu/X2 interface (i.e. signaling is thin-skinned adopts)Recording, XDR) data distribution, PCS is given after mirror image handled, the XDR of other interfaces (as adopted firmly, exporting firewall)Data are still handled by unified DPI system；
The interactive interfacing of PCS and strategy and charging control (Policy Charging Control, PCC) system
After PCS generates the corresponding strategies and rule of customer-centric, the strategy and charging control that are issued in PCC systemUnit (Policy and Charging Rule Function, PCRF), user property memory (SubscriptionProfile Repository, SPR) etc. network elements；
The interactive interfacing of PCS and PDN Gateway (PDN Gateway, PGW)
Internet egress gateways PGW is mainly that PCS provides session management and the carrying control, charging authentication, peace of userThe functions such as full control.
For independent personal Cloud Server, from bottom to top, function structure can be divided into software/hardware resource layer, big dataProcess layer, personal cloud strategic layer and application-interface layer etc., as shown in Figure 5.
Software and hardware resources layer: including the unified software and hardware resources such as calculating, storage, transmission, I/O, management；
Big data process layer: including acquisition, parsing, excavation, the prediction etc. to data flow IP five-tuple, QoS flow and IP packet；
Personal cloud strategic layer: based on big data, treated as a result, generating includes user's perception, service-aware, network senseModels/the algorithms such as space-time bidimensional traffic forecast, personalized service and content push including knowing；
Application-interface layer: data information is obtained from related network elements, and by personal cloud policy distribution to related network elements.
2) personal cloud network element and existing network element are integrated
The big data process layer of cloud network element personal in Fig. 5, personal cloud strategic layer are integrated in the form of software/hardware moduleIn existing network element (such as base station, AC, gateway), the information exchange inside new API or hardware interface progress equipment is set.
3) personal cloud network element and future network element are integrated
As network function virtualizes (Network Function Virtualization, NFV), software defined networkTechnologies such as (Software Defined Network, SDN) gradually mature, and the cloud of core net even wireless network has becomeFuture developing trend.Personal cloud network element and MEC, mCDN become a kind of APP application or system service at that time, are present in edge voidIn quasi-ization father of node application layer, as shown in Figure 6.
Individual's cloud system provided in an embodiment of the present invention is excavated by the depth to wireless network mass data, and novelty solvesWireless edge caching in " personalization " and " mobility " two class key problem, by user's perception, service-aware, network aware with whenEmpty track is that tie is dynamically associated, and the more of user's space-time trajectory, personal preference, business demand and Internet resources are covered in settingOverall edge cache policy is tieed up, and is other network aware optimizations based on big data and design (such as network planning, network optimization, framework, associationView, signaling etc.) correlation model and algorithm of " personalization " and " mobility " are provided while meeting, as shown in Figure 7:
Step 701: obtaining user and perceive big data, service-aware big data and network aware big data；
Step 702: perceiving big data according to user and determine whether user has mobility, if so, thening follow the steps703, otherwise, execute step 704；
Step 703: the motion track of different time scales, different spaces granularity user is predicted；
It is specifically including but not limited to following technical scheme:
1) position prediction: at Idle or cell re-selection state, according to GPS longitude and latitude, cell ID (such as serving cell CellID, ECI etc.), 3 signal strengths, the information such as orientation and deflection (TA+AoA), the base station position Gong Can carry out user locationIt excavates and predicts；
2) trajectory predictions: under cell switching state, based on positioning get ready the starting point of determining user's motion track, terminal,The parameters such as path, speed, approach time, residence time, settling time prism；
3) other track related context informations excavate: by by the mobile simple space-time trajectory of user and calendar, weather,The information such as map, mode of transportation, approach/stop, permanent residence are matched, are associated with, and are therefrom excavated above and below hiding, abundantLiterary semantic information, to excavate prediction, personalized edge for individual cloud VIP user tag, user behavior pattern and business demandCache policy etc. provides foundation, as shown in Figure 8.Several matching process include:
Calendar matching: the time of user's trip, space are matched with calendar information, excavate user on weekdays/The periodical trip rule at weekend, festivals or holidays, commemoration day；
Weather matching: the time of user's trip, space are matched with Weather information, excavate user in different weatherThe periodical trip rule of (the especially bad weathers such as sleet, strong wind, haze)；
Map match: two-dimensional surface map and three-dimensional land map, reality involved by auxiliary judgment user trajectory are introducedScene Semantics information；
Mode of transportation matching: based on parameters such as movement speed, residence times, judge that user's walking still rides public transportation means(such as riding, by bus), to speculate that the probability and mode of business may occur for user；
Approach mode matches: analysis the user residence time of each location point and speed on track, to judge userOnly the approach point, do not occur business, or have the stop of long period in the point, excavate the potential business demand of user；
Permanent residence matching: based on the regularity of distribution of user's motion track over time and space, judge that user often accessesPlace and scope of activities (such as house, school or office building), further to portray user characteristics, thus it is speculated that individual subscriber is inclinedGood and potential business demand.
Step 704: determining whether user has personalization according to service-aware big data, if so, thening follow the steps705, otherwise, execute step 707；
Step 705: the behavior pattern of user is excavated, predicted and is associated with business potential demand；
It is specifically including but not limited to following technical scheme:
1) service object of personal cloud, i.e. VIP user tag are defined, is drawn a portrait to user；
2) by deep message detection (Deep Packet Inspection, DPI) and spiders technology, on userNet log carries out depth excavation, thus it is speculated that the online habit and interest preference of user is excavated user to the potential demand of business, can be wrappedIt includes:
Application type: identification user is parsed based on address base and often uses application type, including using major class and applies group, is led toCross automatic identification and generation that crawler technology realizes label.It is " video " using major class by taking video as an example, is " to rise using groupInterrogate video ".
Content type: deep analysis user's internet behavior, business branch that identification is accessed using user under group, label orChannel.By taking video as an example, content type can be subdivided into TV play, film, variety, juvenile etc..
Content title: the particular content ID of user's access in identification content type.By taking video as an example, it can identify that user seesThe video resource title seen, can be specific to a certain collection or a certain portion, such as collection of " Song of Joy 2 " the 1st, " The Bourne Ultimatum " the 5th.
Content progress: user is identified to the access state of some content title, to judge user preference indirectly.With videoFor, it can identify that user watches the time schedule of the TV play or film, F.F. number etc..
3) the multivariate joint probability prediction based on time, place, business
Using time series, spatial sequence as independent variable, based on moulds such as multidimensional markov chain, multidimensional time-series, multi _ dimensional AR MAType predicts business in the use pattern of space-time bidimensional；
By multidimensional prediction dimension-reduction treatment, establish trajectory predictions relevant to time, place respectively, and with personal preference,The relevant traffic forecast of business preference, then the association of space-time bidimensional is generated by major keys such as User ID, service request time, cell IDTable, as shown in table 1:
1 user's space-time bidimensional of table is associated with example
Step 706: predicting that user's future provides relevant mode in the business demand of space-time bidimensional, and for network aware optimizationType/algorithm；
Step 707: the management of user data is carried out using conventional method；
Step 708: business demand and the network aware big data based on user's future in space-time bidimensional, settingEdge cache.
Here, consider user's space-time mobility (mobility) and personal preference (personalization), design edge cache size andThe strategies such as content, update, distribution, deployment, management:
Size and content: being divided into two big regions for personal cloud storage space, respectively based on file popularity, towards crowdNeed public buffer zone, and based on personal preference, towards the personal buffer zone of a need.The size accounting in two big regions takesCertainly in the deployed position of personal cloud, that is, it is deployed in public buffer zone when network center and is greater than personal buffer zone, be deployed in netPublic buffer zone is less than personal buffer zone when network edge, and specific ratio is adjustable；
Update mode: being directed to above-mentioned public buffer zone, main to use and periodically update as conventional contents network classMode；For above-mentioned personal buffer zone, is updated using the periodic based on user time characteristic and be based on user's mobilityEvent-triggered update combine mode, renewal speed faster, resource utilization it is higher；
Ways of distribution: being directed to above-mentioned public buffer zone, main to use and distribution scheduling machine as conventional contents network classSystem；For above-mentioned personal buffer zone, using based on application type, content type, content title and content progress orientation, pointPiece, intelligent ways of distribution, guarantee VIP user experience；
Deployed position: for the different deployed position of above-mentioned personal cloud network element, when business demand is lower than the whole network average threshold(such as the requirement of trough times, QoS/QoE is low, time delay is insensitive, user mobility is irregular, business predictability is weak when horizontalDeng), edge cache tends to the heart (such as core layer) deployment in a network；When business demand is higher than the whole network average threshold level(such as wave crest moment, QoS/QoE require height, delay sensitive, user mobility more rule, business predictability strong), edgeCaching tends to dispose at network edge (such as access layer, convergence layer)；
Way to manage: for the different networking mode of above-mentioned personal cloud network element, layering centralized management can be used (such as settingPeople's cloud super node), distributed collaboration management (such as P2P), hybrid way to manage.
The foregoing is only a preferred embodiment of the present invention, is not intended to limit the scope of the present invention.