CN113472420B - Satellite network cache placement method based on regional user interest perception - Google Patents
Satellite network cache placement method based on regional user interest perception Download PDFInfo
- Publication number
- CN113472420B CN113472420B CN202110727166.3A CN202110727166A CN113472420B CN 113472420 B CN113472420 B CN 113472420B CN 202110727166 A CN202110727166 A CN 202110727166A CN 113472420 B CN113472420 B CN 113472420B
- Authority
- CN
- China
- Prior art keywords
- content
- leo satellite
- cache
- cluster
- satellite node
- 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.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/14—Relay systems
- H04B7/15—Active relay systems
- H04B7/185—Space-based or airborne stations; Stations for satellite systems
Abstract
The invention relates to a satellite network cache placement method based on regional user interest perception, which comprises the steps of extracting a request log of a base station and establishing a content library set; acquiring state information of an LEO satellite node; establishing an LEO satellite node set as a cache space according to the state information; calculating the similarity of the content labels; clustering the content into a plurality of content clusters according to the similarity of the content labels; calculating the interestingness of each content cluster; dividing the cache space into cache subspaces of each content cluster according to the interest degree of the content cluster; determining popularity of the content; determining the number of the selected contents in each content cluster according to the popularity and the cache subspace; placing selected content assemblies in the plurality of content clusters into a content assembly; ranking all the placing contents according to the popularity to obtain a placing content ordered set; and placing the contents in the ordered set of placed contents into the LEO satellite node according to the scheme of shortest placing time. The method and the device realize the reduction of content response time delay and the improvement of content hit rate.
Description
Technical Field
The invention relates to the field of satellite networks, in particular to a satellite network cache placement method based on regional user interest perception.
Background
The sixth generation mobile communication system is a communication system that integrates a terrestrial network with a satellite communication system. In the aspect of communication coverage, the communication scene of 6G can be expanded to sea, land, air and even underwater space. As on-board processing performance continues to improve, research into deploying caching schemes on satellites is receiving increasing attention from researchers. And the cache placement problem in the satellite network directly influences the content request effect.
The satellite network adopts a communication mode based on ICN, and the satellite nodes can realize data caching and local caching content perception. And adopting a network architecture based on the SDN to globally control various network resources.
However, the content has regional differences in the wide-area coverage satellite network, and the content selection of the whole network affects the content request response effect.
Disclosure of Invention
The invention aims to provide a satellite network cache placement method based on regional user interest perception, which can reduce the response time delay of contents and improve the hit rate of the contents.
In order to achieve the purpose, the invention provides the following scheme:
A satellite network cache placement method based on regional user interest perception, the method comprising:
extracting request logs of all base stations in an area and establishing a whole network content library set in the area;
acquiring state information of a LEO satellite node, wherein the state information comprises: inter-satellite link bandwidth information of the LEO satellite and space capacity information of the LEO satellite;
establishing an LEO satellite node set as a cache space according to the state information of the LEO satellite nodes;
calculating the label similarity of the contents in the content library set;
clustering the content in the content library set into a plurality of content clusters according to the content label similarity;
calculating the interest degree of the users in the area to each content cluster;
dividing a cache subspace of each content cluster in the cache space according to the interestingness of the content cluster;
determining the popularity of the content in each of the content clusters;
determining the number of the selected contents in each content cluster according to the popularity and the cache subspace;
placing selected ones of the plurality of content clusters into a content collection;
ranking all the placing contents according to the popularity of each placing content in the placing content set to obtain a placing content ordered set;
And sequentially placing the contents in the placed content ordered set into the LEO satellite node according to the state information of the LEO satellite node according to the scheme with the shortest placing time.
Optionally, the calculating the tag similarity of the content in the content library set specifically includes:
using a formula
Calculating the label similarity of the contents in the content library set; wherein, Sim(s)a,sb) Is the label similarity of content a and content b, saIs content a, sbIs content b, ljFor the jth tag, the number of tags,
the weight of the jth label in content a,
is the weight of the jth label in the content b, and m is the total number of labels.
Optionally, the calculating the interest level of the user in the area in each content cluster specifically includes:
using formulas
Is calculated toThe interest degree of each content cluster of users in the region; wherein the content of the first and second substances,
the interest degree of the users in the area to the g-th content cluster, q is the number of base stations in the area, u1、u2And uqThe number of users under the first base station, the second base station and the q-th base station respectively, omega is the total number of users in the area,
and
respectively short-term interest degrees of users under a first base station, a second base station and a q-th base station in the area in the g-th content cluster, 
1,2, q, wherein (t)0,t1)、(t1,t2) And (t)n-1,tn) First, second and nth statistical time periods respectively,
for the short-term interest of users under the z-th base station in the area in the g-th content cluster,
and
respectively the number of times of requests of the user to the g-th content cluster in the first statistical time period, the second statistical time period and the n-th statistical time period under the z-th base station,
and
respectively being the users under the z-th base station at the first and the secondThe number of requests for all content clusters in the second and nth statistical time periods.
Optionally, the dividing the cache subspace of each content cluster in the cache space according to the interestingness of the content cluster specifically includes:
using formulas
Determining a cache subspace of a single content cluster; wherein, CgThe cache subspace for the g-th content cluster,
the interest degree of users in the area to the g-th content cluster, C is the total cache space, d is the number of the content clusters, and g is the g-th content cluster.
Optionally, the determining the popularity of the content in each content cluster specifically includes:
using formulas
Determining the popularity of the content in each of the content clusters; wherein s isiIs the name of the content, k is the content popularity ranking of the content in the whole network,
for each content cluster content s iThe popularity of (a) is a parameter of Zipf distribution, N is the total number of contents in the whole network, kjIndicating the ranking of the popularity of content with sequence number j in the web-wide content.
Optionally, the determining the number of the selected contents in each content cluster according to the popularity and the cache subspace specifically includes:
using a formula
Calculating the number of the selected contents in the g-th content cluster; wherein w is the number of contents selected in the content cluster,
to cache whether the ith content is present in the subspace,
is the fractional space of the ith content, CgIs the cache subspace of the g-th content cluster.
Optionally, the sequentially placing the content in the ordered set of placed contents into the LEO satellite node according to the state information of the LEO satellite node according to the scheme that the placing time is the shortest includes:
initializing i, and enabling i to be 1;
calculating the average response time delay of each LEO satellite node in the LEO satellite node set according to the ith content in the placed content ordered set;
sequencing all LEO satellite nodes according to the ascending order of the average response time delay, initializing h, and enabling h to be 1;
judging whether the space of the ith content in the ordered set of the placed contents is larger than the remaining cache space of the selected h-th LEO satellite node or not to obtain a first judgment result;
If the first judgment result shows that the content is the h-th content, increasing the value of the h by 1, and returning to the step of judging whether the space for placing the ith content in the ordered set of contents is larger than the residual cache space of the selected h-th LEO satellite node to obtain a first judgment result;
and if the first judgment result shows that the content is not the h-th LEO satellite node, sending the ith content to the h-th LEO satellite node, increasing the value of i by 1, and returning to the step of calculating the average response time delay of each LEO satellite node in the LEO satellite node set according to the ith content in the placed content ordered set until all the content in the placed content ordered set is traversed.
Optionally, the calculating an average response delay of each LEO satellite node in the LEO satellite node set according to the ith content in the placed content ordered set specifically includes:
making N decision schemes according to the required cache space of each content in the placed content ordered set and the size of the residual cache space of each LEO satellite node;
using a formula
Calculating the average response time delay under each decision scheme; where T is the average response delay under the decision scheme, 
To place the ith content into the response delay of the placeable LEO satellite node for the ith content marked in the decision-making scheme,
required buffer space for ith content, vxThe bandwidth of the x-th communication link between the placeable LEO satellite node of the ith content and the content request point, n is the number of the communication links between the placeable LEO satellite node of the ith content and the content request point, tdIs the propagation delay of the inter-satellite link of the LEO satellite node,
for the number of times the ith content is accessed,
requesting the response time of the ith content from the LEO satellite node for a user request point, wherein Y is the number of the selected contents in all content clusters in the area;
and selecting the decision scheme with the minimum average response time delay for content placement.
Optionally, the making N decision schemes according to the size of the cache space required by each content in the ordered set of placed contents and the remaining cache space of each LEO satellite node specifically includes:
initializing i, and enabling i to be 1;
determining LEO satellite nodes with the content capable of placing LEO satellite nodes and the residual cache space larger than the cache space required by the ith content in the ordered set of placed contents to form a set of LEO satellite nodes with the content capable of placing LEO satellite nodes;
Judging whether the LEO satellite node set capable of being placed in the content is empty or not to obtain a second judgment result;
if the second judgment result shows that the content of the LEO satellite node set can be placed in the ordered set of the placement contents, selecting any LEO satellite node in the LEO satellite node set with the content capable of being placed, marking the LEO satellite node with the ith content capable of being placed as the LEO satellite node with the ith content, updating the residual cache space of the content capable of being placed in the LEO satellite node with the ith content, increasing the value of i by 1, and returning to the step of determining the LEO satellite node with the residual cache space of the content capable of being placed in the LEO satellite node larger than the cache space required by the ith content in the ordered set of the placement contents to form the LEO satellite node set with the content capable of being placed in the ordered set;
and if the second judgment result shows that the content is in the ordered set, giving up the ith content, increasing the value of i by 1, returning to the step of determining the LEO satellite nodes with the residual cache space for placing the LEO satellite nodes larger than the cache space required by the ith content in the ordered set of the placed content to form a set of the LEO satellite nodes with the placeable content, and generating a decision scheme until the content in the ordered set of the placed content is traversed to be finished.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
The invention provides a satellite network cache placement method based on regional user interest perception, which comprises the following steps: extracting request logs of all base stations in an area and establishing a whole network content library set in the area; acquiring state information of an LEO satellite node, wherein the state information comprises: inter-satellite link bandwidth information of the LEO satellite and space capacity information of the LEO satellite; establishing an LEO satellite node set as a cache space according to the state information of the LEO satellite nodes; calculating the label similarity of the contents in the content library set; clustering the content in the content library set into a plurality of content clusters according to the content label similarity; calculating the interest degree of the users in the area to each content cluster; dividing a cache subspace of each content cluster in the cache space according to the interestingness of the content cluster; determining the popularity of the content in each of the content clusters; determining the number of the selected contents in each content cluster according to the popularity and the cache subspace; placing selected ones of the plurality of content clusters into a content collection; ranking all the placing contents according to the popularity of each placing content in the placing content set to obtain a placing content ordered set; and sequentially placing the contents in the placed content ordered set into the LEO satellite node according to the state information of the LEO satellite node according to the scheme with the shortest placing time. The method provided by the invention is superior to a comparison method in the aspects of response delay and cache hit rate of the content, not only reduces the response delay of the content, but also considers the distribution balance of the content.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without creative efforts.
Fig. 1 is a flowchart of a satellite network cache placement method based on regional user interest awareness according to the present invention;
fig. 2 is a scene description diagram of a satellite network cache placement method based on regional user interest awareness according to the present invention;
FIG. 3 is a comparison graph of average response time delay of a satellite network cache placement method based on regional user interest awareness according to the present invention;
FIG. 4 is a comparison graph of average cache hit rates of a satellite network cache placement method based on regional user interest awareness according to the present invention;
FIG. 5 is a comparison graph showing how the cache hit rate of the method for placing a satellite network cache based on regional user interest awareness according to the present invention is affected by the number of base stations;
fig. 6 is a graph of the variation of the average response delay of the satellite network cache placement method based on regional user interest perception according to the Zipf parameter.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
The invention aims to provide a satellite network cache placement method based on regional user interest perception, which can reduce the response time delay of contents and improve the hit rate of the contents.
In order to make the aforementioned objects, features and advantages of the present invention more comprehensible, the present invention is described in detail with reference to the accompanying drawings and the detailed description thereof.
The invention relates to a double-layer satellite network architecture based on SDN (software Defined network), wherein an LEO (Low Earth orbit) layer satellite is a forwarding layer in the architecture. A constellation composed of 3 geo (geostationary orbit) satellites is an SDN control layer and is responsible for state collection of LEO satellites and calculation of cache placement positions.
The application scene of the invention comprises two parts, namely a ground network and a satellite communication network. SDN based satellite networks consist of a GEO satellite constellation and a LEO satellite constellation. The GEO satellite is responsible for monitoring the state of the LEO satellite, is provided with an SDN controller and uniformly controls the network state and the resource configuration of the LEO through an OpenFlow protocol. A forwarding plane is formed by networking a plurality of LEO satellites, and a stable inter-satellite link is formed through a fixed laser link under the management of SDN control.
In the above scenario, the number of ground communication base stations under the coverage of a single LEO satellite is not fixed. Most users access the network through a ground base station, and the base station can count and collect request data of a large number of users. When the communication between the base station and the satellite is established, the request data of a local user can be uploaded to the satellite, and the regional content is selected for caching under the unified decision of the SDN controller. And due to the fact that the SDN controller monitors the state of the whole network satellite, including the state of the LEO satellite and the state of an inter-satellite link, the optimal content placement position can be decided under the global view. Fig. 2 is a network scenario diagram.
As shown in fig. 1, a satellite network cache placement method based on regional user interest perception includes: extracting request logs of all base stations in an area and establishing a whole network content library set in the area; acquiring state information of an LEO satellite node, wherein the state information comprises: inter-satellite link bandwidth information of the LEO satellite and space capacity information of the LEO satellite; establishing an LEO satellite node set as a cache space according to the state information of the LEO satellite nodes; calculating the label similarity of the contents in the content library set; clustering the content in the content library set into a plurality of content clusters according to the content label similarity; calculating the interest degree of the users in the area to each content cluster; dividing a cache subspace of each content cluster in the cache space according to the interestingness of the content cluster; determining the popularity of the content in each of the content clusters; determining the number of the selected contents in each content cluster according to the popularity and the cache subspace; placing selected ones of the plurality of content clusters into a content collection; ranking all the placing contents according to the popularity of each placing content in the placing content set to obtain a placing content ordered set; and sequentially placing the contents in the placed content ordered set into the LEO satellite node according to the state information of the LEO satellite node according to the scheme with the shortest placing time.
The calculating of the tag similarity of the content in the content library set specifically includes: using a formula
Calculating the label similarity of the contents in the content library set; wherein, Sim(s)a,sb) Is the label similarity of content a and content b, saIs content a, sbIs content b, ljFor the jth tag, the number of tags,
the weight of the jth label in content a,
is the weight of the jth label in the content b, and m is the total number of labels.
The content name is divided into a number of matchable fields according to the "/" separator in the packet naming. Dividing fields in the naming format and a full gateway key word set K ═ K1,k2,···,kuThe elements in the are matched.
The theme tag set is L ═ L1,l2,···,lmThat a keyword may belong to multiple topic types, a single content siIs characterized by:
presentation subject label ljIn the content siThe weight in (1).
After the matching of the content matching field and the keyword, one label l is calculated by adopting a statistical methodjIn the content siThe weight in (1). While a keyword may be subordinate to multiple tags, e.g. tag k1：la,···,lbLabel ljIn the content siThe weight calculation formula in (1) is:
wherein the content of the first and second substances,
represents the content siAfter matching the keyword, the keyword depends from the tag l jThe total number of (c) is,
represents the content siThe sums of the numbers of the separated matchable fields. After the VSM vector of content features is established,the cosine similarity of any two content features is calculated as:
clustering the content in the content library set into a plurality of content clusters according to the content label similarity as follows:
randomly draw d samples from the content set as an initial mean vector u1,u2,···,ud}
Initialization
Fori＝1:n
Calculating a sample siAnd each mean vector uj(1. ltoreq. j. ltoreq. d) distance Sim(s)i,uj)；
According to the sum ofiDetermining cluster marks according to the nearest mean vector;
content siDividing into a cluster of responses;
End
For j＝1:d
calculating a new mean vector within each cluster
Summing each sample in the cluster/number of samples in the cluster;
If
the current mean vector ujIs updated to
Else
Keeping the current mean vector unchanged;
End
End
until the mean vector of each cluster is not updated;
output G ═ G1,g2,···,gd}。
The calculating the interest degree of the user in the region in each content cluster specifically includes: using formulas
Calculating the interest degree of the users in the region to each content cluster; wherein the content of the first and second substances,
the interest degree of users in the area to the g-th content cluster, g is the g-th content cluster, q is the number of base stations in the area, u1、u2And uqThe number of users under the first base station, the second base station and the q-th base station respectively, omega is the total number of users in the area, 
And
respectively short-term interest degrees of users under a first base station, a second base station and a q-th base station in the area in the g-th content cluster,
1,2, q, wherein (t)0,t1)、(t1,t2) And (t)n-1,tn) First, second and nth statistical time periods respectively,
for the short-term interest of users under the z-th base station in the area in the g-th content cluster,
and
respectively is the z thThe number of times of requests of the user to the g-th content cluster in the first, second and nth statistical time periods under the base station,
and
the number of requests for all content clusters in the first, second and nth statistical time periods for the user under the z-th base station respectively.
Under the system model, a single LEO satellite covers a plurality of ground base stations. In selecting the most appropriate cache contents to maximize the cache gain of the LEO satellite, the collection of the request data of the base station under the satellite coverage is mainly considered. The group users are relatively fixed under a single base station, and the movement of small-scale users has little influence on the group request data. Since popular content has a life cycle that is not typically too long, the short-term historical request behavior can reflect the user's local preference for content at the base station. Meanwhile, the user's interest in the content may be represented by the number of requests for the content. Defining a statistical time t 0-t1The short-term interest of group users under the base station on a single content cluster g is
Then the
Wherein, the first and the second end of the pipe are connected with each other,
representing the statistical time t of a user under a single base station0-t1The cumulative number of requests for the content cluster g over time,
indicating the statistical time t of the user under the base station0-t1Total number of requests in time. And areAnd in order to eliminate accidental variation caused by over-short selected statistical time, the current value is more accurately represented by a method of increasing the sliding average value of the preamble statistical period, so that the short-term interest P with the preamble periodi shortThe calculation of (t) is:
wherein the content of the first and second substances,
representing the current statistical time period t0-t1The degree of interest in the content of the content,
representing the interestingness within n adjacent preamble periods of the statistical period.
Due to the fact that a plurality of base stations exist in the coverage range of a single satellite, the different activity degrees of users under different base stations are considered, and the users under different base stations also have different interests. Therefore, the interest degree of the user in the single base station for the content cannot represent the interest degree of the whole area, and after the interest degrees of the user in the single base station are respectively calculated, the weighted value of the interest degree of each base station in the whole area is used as the interest preference of the area content, so that the caching income is maximized. The number of base stations in a defined area is q, and the number of fixed users under the base stations is mu respectively 1,μ2···μqThe total number of regional users is omega, and the interest degree of a single base station to a single content cluster can be respectively calculated through a formula 3.5, so that the interest degree of a single content cluster g in a region
Comprises the following steps:
dividing each content in the cache space according to the interestingness of the content clusterThe cluster cache subspace specifically includes: using a formula
Determining a cache space of a single content cluster; wherein, CgIs the buffer space of the g-th content cluster,
the interest degree of the users in the area to the g-th content cluster, C is the total cache space, and d is the number of the content clusters.
The determining the popularity of the content in each content cluster specifically includes: using formulas
Determining the popularity of the content in each of the content clusters; wherein s isiIs the name of the content, k is the content popularity ranking of the content in the whole network,
for each content cluster content siAlpha is a parameter of Zipf distribution, N is the total number of contents in the whole network, kjIndicating the ranking of the popularity of the content with sequence number j in the web-wide content.
The determining the number of the selected contents in each content cluster according to the popularity and the cache subspace specifically includes: using formulas
Calculating the number of the selected contents in the g-th content cluster; wherein w is the number of contents selected in the content cluster, 
For caching whether content s exists in a subspacei，
As a content siC space of proportion ofgThe cache subspace for the g-th content cluster.
The requests for media content by users throughout the network are in accordance with the Zipf distribution. That is, the popularity of the whole network with rank k and content name s is expressed as:
where α is a parameter of the Zipf distribution, and N represents the number of media contents of the whole network.
Due to the limited satellite cache space, the reasonable placement of selected cache content can improve user experience. The process by which a user request is responded to is first analyzed. The single policy calculation is considered herein as a time slice, and the choice of time slice is small. And the concept of virtual nodes and virtual topology is introduced in the satellite network, i.e. user requests within a single time slice are aggregated onto the satellite acting as a virtual node, while the dynamics of the satellite topology of the shorter time slice are not taken into account. The user request is sent to the virtual node, and if the content exists in the node cache, the content is returned. If the content does not exist in the node, the interest packet is sent to the surrounding nodes to search for the required content. If the search exceeds the response time, the content is searched directly from an Internet Service Provider (ISP). Under the ideal condition, the content required by the user can be satisfied in the satellite network.
After the cache space of the whole network is divided according to the interestingness of the user on the content clusters, the content with high popularity is placed in the space distributed by a single content cluster in a descending order according to the popularity of the content. If the size of the selected content to be cached exceeds the residual size of the current content cluster, the content which can be cached in the content popularity list is selected in a forward-delay manner until the cache space allocated to the content cluster is completely allocated. Defining a selected one of the content clusters as g ═ s1,s2···swThe number of the selected contents in a single cluster is w, and the sizes of the contents are respectively w
Then the
The sequentially placing the contents in the placing content ordered set into the LEO satellite node according to the state information of the LEO satellite node according to the scheme that the placing time is the shortest specifically includes: initializing i, and enabling i to be 1; calculating the average response time delay of each LEO satellite node in the LEO satellite node set according to the ith content in the placed content ordered set; sequencing all LEO satellite nodes according to the ascending order of the average response time delay, initializing h, and enabling h to be 1; judging whether the space of the ith content in the ordered set of the placed contents is larger than the remaining cache space of the selected h-th LEO satellite node or not to obtain a first judgment result; if the first judgment result shows that the content is the h-th content, increasing the value of the h by 1, and returning to the step of judging whether the space for placing the ith content in the ordered set of contents is larger than the residual cache space of the selected h-th LEO satellite node to obtain a first judgment result; and if the first judgment result shows that the content is not the h-th LEO satellite node, sending the ith content to the h-th LEO satellite node, increasing the value of i by 1, and returning to the step of calculating the average response time delay of each LEO satellite node in the LEO satellite node set according to the ith content in the placed content ordered set until all the content in the placed content ordered set is traversed.
The sequentially placing the contents in the ordered set of placed contents into the cache space according to the scheme that the placing time is the shortest specifically includes: calculating the average response time delay of each content placement position in the content placeable satellite node set, and selecting the content placeable satellite node with the minimum average response time delay as the placement position of the content; selecting the content with the highest popularity in the ordered set of placed content; judging the size of the placement space of the content with the highest popularity in the ordered set of the placement contents and the size of the remaining cache space of the selected content capable of placing the satellite node, and obtaining a first judgment result; if the first judgment result shows that the placement space of the content with the highest popularity in the placed content ordered set is smaller than the remaining cache space of the selected content where the satellite node can be placed, recording the position of the selected content where the satellite node can be placed, and selecting the next placement position and the content with the highest popularity in the placed content ordered set; if the first judgment result shows that the placement space of the content with the highest popularity in the ordered set of the placement contents is larger than the remaining cache space of the selected content placeable satellite node, selecting the next content with the minimum average response time delay to place the satellite node; and traversing all the contents in the ordered set of the placed contents until the placement is finished.
Making N decision schemes according to the size of the required cache space of each content in the placed content ordered set and the size of the residual cache space of each LEO satellite node; using a formula
Calculating the average response time delay under each decision scheme; where T is the average response delay under the decision scheme,
to place the ith content into the response delay of the placeable LEO satellite node for the ith content marked in the decision-making scheme,
required buffer space for ith content, vxThe bandwidth of the x-th communication link between the placeable LEO satellite node of the ith content and the content request point, n is the number of the communication links between the placeable LEO satellite node of the ith content and the content request point, tdIs the propagation delay of the inter-satellite link of the LEO satellite node,
for the number of times the ith content is accessed,
requesting the response time of the ith content from the LEO satellite node for a user request point, wherein Y is the number of the selected contents in all content clusters in the area; selecting the decision party with the minimum average response time delayAnd placing the contents.
The calculating an average response delay of each LEO satellite node in the LEO satellite node set according to the ith content in the placed content ordered set specifically includes: the making of N decision schemes according to the size of the required cache space of each content in the ordered set of placed contents and the remaining cache space of each LEO satellite node specifically includes: initializing i, and enabling i to be 1; determining LEO satellite nodes with the content capable of placing LEO satellite nodes and the residual cache space larger than the cache space required by the ith content in the ordered set of placed contents to form a set of LEO satellite nodes with the content capable of placing LEO satellite nodes; judging whether the LEO satellite node set capable of being placed in the content is empty or not to obtain a second judgment result; if the second judgment result shows that the content of the LEO satellite node set can be placed in the ordered set of the placement contents, selecting any LEO satellite node in the LEO satellite node set with the content capable of being placed, marking the LEO satellite node with the ith content capable of being placed as the LEO satellite node with the ith content, updating the residual cache space of the content capable of being placed in the LEO satellite node with the ith content, increasing the value of i by 1, and returning to the step of determining the LEO satellite node with the residual cache space of the content capable of being placed in the LEO satellite node larger than the cache space required by the ith content in the ordered set of the placement contents to form the LEO satellite node set with the content capable of being placed in the ordered set; and if the second judgment result shows that the content is in the ordered set, giving up the ith content, increasing the value of i by 1, returning to the step of determining the LEO satellite nodes with the residual cache space for placing the LEO satellite nodes larger than the cache space required by the ith content in the ordered set of the placed content to form a set of the LEO satellite nodes with the placeable content, and generating a decision scheme until the content in the ordered set of the placed content is traversed to be finished.
When the second judgment result shows that the LEO satellite nodes in the LEO satellite node set can be placed in the selected content before the change, and another LEO satellite node is selected as the marked LEO satellite node which can be placed, so that various decision schemes can be generated. And selecting the decision scheme with the minimum average response time delay from the multiple decision schemes for content placement.
The selected contents in each content cluster are collected to form the content clusterSet of content S to be placed in the front areaselect＝{s1,s2,···,sYThe number of the selected contents in the whole area is whole Y, and the sizes of the contents are respectively
The bandwidth of a communication link between a deployment point and a device access node is v1,v2,···,vnThen a single content siThe communication delay of the backhaul is:
wherein, tdRepresenting the propagation delay of the inter-satellite link.
Defining content s at a device access nodeiThe number of times of access is
Device access node requesting content s from point-to-be-deployediHas a response time of
If the content to be deployed exists in the current satellite network and the response time delay from the known equipment access node to the current cached content node is TcurThen the average response time from the device access node to the computing deployment point is:
where Y represents the number of contents in the list currently to be selected for execution. The response time of requesting the content i from the deployment point should be less than the response time delay of the existing nodes in the current network, otherwise the deployment action is meaningless. And the content request delay should not exceed the request tolerance time T tolAnd if the condition is not met, the content is not deployed at the node.
Indicating whether or not there is content s at the current deployment pointiIf the content s already existsiThen, then
Otherwise
Represents the content siSize of (C)remainRepresenting the amount of cache space remaining available in the deployment node, which is known since LEO nodes periodically upload node state information to the SDN controller. The objective function is that the average response time is required to be minimum, and the constraint conditions respectively represent the response time limit from the deployment point to the equipment access node and the cache space limit of the corresponding content category.
The problem of cache content placement is converted into an optimization problem of solving the minimum average response delay of all the contents from a request point to a deployment point for obtaining the contents. The problem can be regarded as a knapsack problem, and the number of the knapsack in each cluster is uncertain, and a placing constraint condition exists, so that the problem is complex. The knapsack problem is solved by a greedy algorithm within the limits of satellite computing capacity and decision time, and local optimization is searched for through multiple iterations, so that global optimization is solved. After the content is selected, the selected contents in each content cluster are collected to form a content set S to be placed in the current area select＝{s1,s2,···,sYRanking the elements in the set again according to popularity to obtain an ordered set
Then taking the average time delay of the placement as an optimization target, and selecting the ordered set each time
And (4) calculating the average response time delay of the content placed in the available nodes under the limitation of the cache space, and selecting the placement method with the lowest average response time delay. And continuously iterates to obtain a local optimal solution,the placement of one video is completed. Thereafter, from the ordered collection
And selecting the next content with the highest popularity and repeating the placing action, wherein when the residual cache space is not enough for storing the new content, the cache placing action is terminated.
The method can delay the cache in the information center network to be placed in the low orbit satellite, and realizes the quick response of the user content request. Compared with the traditional method for acquiring data from the source data center, the method has the advantages that the response time of requesting the content can be shortened, and meanwhile, the hit rate of the content is improved.
Under an SDN-based satellite network architecture, an SDN controller is placed on a GEO satellite. LEO satellite nodes can perceive local cache contents and upload cache updates to SDN nodes in time. And meanwhile, the link state of the whole LEO satellite constellation can be acquired through the SDN controller.
Fig. 3 is a graph showing the average response delay of fig. 4 for requesting the full network content from the virtual node under four different caching policies. With the Zipf parameter set to 0.8. It can be seen that the average response delay of the regional user interest perception cache placement algorithm (RUIPM) provided by the invention is lower than that of the other three cache placement strategies. And when the cache space is small, the response time delay of the four strategies is higher than that of the cache space. The reason is that when the cache space cannot meet the requirement, the cache content needs to be placed in a node farther away from the virtual node, and the response time of the content is increased.
Fig. 4 is a comparison graph of cache hit rates of four caches in different cache spaces. The RUIPM algorithm provided by the invention has great advantages in optimizing the cache hit rate of the whole network, and the reason is that the algorithm divides the cache space according to the interest degree on the basis of sensing the interest of users in the region, and gives consideration to the cache of cold content on the basis of ensuring the cache of hot popular content, thereby ensuring the high cache hit rate. And the cache hit rate is continuously improved along with the increase of the cache capacity, and the hit rate gradually becomes stable after the cache capacity exceeds 300 MB. This means that after the cache capacity is larger than 300MB, most of the content requested by the user is cached in the network.
Fig. 5 is a comparison diagram of the cache hit rate influenced by the number of base stations in an area, and since a plurality of ground communication base stations exist in the coverage area of a single satellite and the area mentioned in the present invention is a concept defined by human, the number of base stations in the area is not fixed. There is also a large difference in user interests from base station to base station. Therefore, the influence of the number of base stations in the comparison area on the cache hit rate is compared. It can be seen from the comparison graph that the cache hit rate mainly depends on the size of the cache space under the same number of base stations. Under the condition of the same cache space, the cache hit rate is gradually reduced along with the increase of the number of the base stations, because the user interests of the base stations are different from those of the base stations, the content cached in the whole network cannot meet the user needs of all the base stations. And in the case of a cache space of 50MB, the hit rate performance of the cache is the worst, because most of the department content cannot be cached in the network when the cache space cannot meet the system requirements at all. And when the cache space is 300MB, most of the whole network content library is cached in the network, so that the cache hit rate is hardly influenced by the number of the base stations.
Fig. 6 is a graph showing the change of the Zipf parameter versus the change of the average response time delay of the whole network in four caching strategies, with the caching capacity being fixed at 300 MB. It can be seen that the performance of the RUIPM algorithm proposed by the present invention is more stable, and as the Zipf parameter becomes larger, the average response delay of the content gradually decreases, because the increase of the Zipf parameter indicates that the user's request gradually becomes more and less on the content, and when the parameter is between 1-1.5, the types of the content cached in the whole network become less and are the most popular content. This is the same idea as the criminal cache placement strategy, which can bring better response latency. The greedy algorithm is adopted to optimize the response time delay of the placement position, partial unhealthy contents are cached due to the adoption of the theme interest division idea, and the contents of the department occupy the optimal placement position of the unhealthy contents, so that the time delay performance is slightly higher than the most popular caching strategy when the Zipf parameter is between 1 and 1.5. In addition, the random placement strategy decides whether to cache the content with random probability, and the algorithm makes the hot content not to be cached sufficiently, so the time delay performance is poor.
The invention provides a LEO satellite network cache placement strategy (RUIPM) based on regional interest perception. The invention designs a low orbit satellite network cache placement method under the ICN-based communication mode. And taking the label of the content as the distance measurement of the content similarity, and dividing the whole network content into a plurality of content clusters through a clustering algorithm. And evaluating the interest preference of the regional users to each content cluster by using historical request data, and dividing the cache space of the whole network content according to preference weight. And defining the placement problem of the content by taking the average time delay as an optimization target, and solving the placement problem by adopting a greedy algorithm. Simulation results show that the method provided by the invention is superior to a comparison method in the aspects of content response delay and cache hit rate, not only reduces the content response delay, but also considers the distribution balance of the content.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
The principle and the implementation mode of the invention are explained by applying a specific example, and the description of the embodiment is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the foregoing, the description is not to be taken in a limiting sense.
Claims (6)
1. A satellite network cache placement method based on regional user interest perception is characterized by comprising the following steps:
extracting request logs of all base stations in an area and establishing a whole network content library set in the area;
acquiring state information of a LEO satellite node, wherein the state information comprises: inter-satellite link bandwidth information of the LEO satellite and space capacity information of the LEO satellite;
establishing an LEO satellite node set as a cache space according to the state information of the LEO satellite nodes;
calculating the label similarity of the contents in the content library set;
Clustering the content in the content library set into a plurality of content clusters according to the content label similarity;
calculating the interest degree of each content cluster of the users in the area;
dividing a cache subspace of each content cluster in the cache space according to the interestingness of the content cluster;
determining a popularity of content in each of the content clusters;
determining the number of the selected contents in each content cluster according to the popularity and the cache subspace;
placing selected content assemblies in a plurality of the content clusters into a content assembly;
ranking all the placed contents according to the popularity of each placed content in the placed content set to obtain a placed content ordered set;
according to the state information of the LEO satellite node, sequentially placing the contents in the placed content ordered set into the LEO satellite node according to the scheme with the shortest placing time, and specifically comprising the following steps:
initializing i, and enabling i to be 1;
calculating the average response time delay of each LEO satellite node in the LEO satellite node set according to the ith content in the placed content ordered set;
sequencing all LEO satellite nodes according to the ascending order of the average response time delay, initializing h, and enabling h to be 1;
Judging whether the space for placing the ith content in the content ordered set is larger than the residual cache space of the selected h-th LEO satellite node or not to obtain a first judgment result;
if the first judgment result shows that the content is the h-th content, increasing the value of the h by 1, and returning to the step of judging whether the space for placing the ith content in the content ordered set is larger than the residual cache space of the selected h-th LEO satellite node to obtain a first judgment result;
if the first judgment result shows that the set of LEO satellite nodes is not the set of LEO satellite nodes, sending the ith content to the h-th LEO satellite node, increasing the value of i by 1, and returning to the step of calculating the average response time delay of each LEO satellite node in the LEO satellite node set according to the ith content in the placed content ordered set until all the content in the placed content ordered set is traversed;
the calculating an average response delay of each LEO satellite node in the LEO satellite node set according to the ith content in the placed content ordered set specifically includes:
making N decision schemes according to the size of the required cache space of each content in the placed content ordered set and the size of the residual cache space of each LEO satellite node;
Using a formula
Calculating the average response time delay under each decision scheme; where T is the average response delay under the decision scheme,
to place the ith content into the response delay of the placeable LEO satellite node for the ith content marked in the decision-making scheme,
required buffer space for ith content, vxThe bandwidth of the x-th communication link between the placeable LEO satellite node of the ith content and the content request point, n is the number of the communication links between the placeable LEO satellite node of the ith content and the content request point, tdIs the propagation delay of the inter-satellite link of the LEO satellite node,
for the number of times the ith content is accessed,
requesting the response time of the ith content from the LEO satellite node for a user request point, wherein Y is the number of the selected contents in all content clusters in the area;
selecting the decision scheme with the minimum average response time delay for content placement;
the making of N decision schemes according to the size of the required cache space of each content in the ordered set of placed contents and the remaining cache space of each LEO satellite node specifically includes:
initializing i, and enabling i to be 1;
determining LEO satellite nodes with the content capable of placing LEO satellite nodes and the residual cache space larger than the cache space required by the ith content in the ordered set of placed contents to form a set of LEO satellite nodes with the content capable of placing LEO satellite nodes;
Judging whether the LEO satellite node set capable of being placed in the content is empty or not to obtain a second judgment result;
if the second judgment result shows that the content can be placed in the LEO satellite node set, selecting any LEO satellite node in the LEO satellite node set, marking the LEO satellite node which can be placed in the ith content, updating the residual cache space of the LEO satellite node which can be placed in the ith content, increasing the value of i by 1, and returning to the step of determining the LEO satellite node which can be placed in the residual cache space of the LEO satellite node in the content is larger than the cache space required by the ith content in the placed content ordered set to form the LEO satellite node set which can be placed in the content;
and if the second judgment result shows that the content is in the ordered set, giving up the ith content, increasing the value of i by 1, returning to the step of determining the LEO satellite nodes with the residual cache space for placing the LEO satellite nodes larger than the cache space required by the ith content in the ordered set of the placing content to form a set of LEO satellite nodes with the placeable content until the content in the ordered set of the placing content is traversed to be finished, and generating a decision scheme.
2. The satellite network cache placement method based on regional user interest awareness according to claim 1, wherein the calculating of the tag similarity of the content in the content library set specifically includes:
Using a formula
Calculating the label similarity of the contents in the content library set; wherein, Sim(s)a,sb) Is the label similarity of content a and content b, saIs content a, sbIs content b, ljFor the jth tag, the number of tags,
the weight of the jth label in content a,
is the weight of the jth label in the content b, and m is the total number of labels.
3. The method for placing the satellite network cache based on the regional user interest perception according to claim 1, wherein the calculating the interest degree of the users in the region in each content cluster specifically includes:
using formulas
Calculating the interest degree of the users in the region to each content cluster; wherein the content of the first and second substances,
the interest degree of the users in the area to the g-th content cluster, q is the number of base stations in the area, u1、u2And uqRespectively under the first base station, the second base station and the q base stationThe number of users, ω is the total number of users in the area,
and
respectively short-term interest degrees of users under a first base station, a second base station and a q-th base station in the area in the g-th content cluster,
1,2, q, wherein (t)0,t1)、(t1,t2) And (t)n-1,tn) First, second and nth statistical time periods respectively,
for the short-term interest of users under the z-th base station in the area in the g-th content cluster, 
And
respectively the number of times of the user's request for the g-th content cluster in the first, second and nth statistical time periods under the z-th base station,
and
the number of requests for all content clusters by the user under the z-th base station in the first, second and nth statistical time periods, respectively.
4. The satellite network cache placement method based on regional user interest perception according to claim 1, wherein the dividing of the cache subspace of each content cluster in the cache space according to the interest degree of the content cluster specifically comprises:
using a formula
Determining a cache subspace of a single content cluster; wherein, CgThe cache subspace for the g-th content cluster,
the interest degree of the users in the area to the g-th content cluster, C is the total cache space, and d is the number of the content clusters.
5. The method for placing the satellite network cache based on the regional user interest perception according to claim 1, wherein the determining the popularity of the content in each content cluster specifically includes:
using formulas
Determining the popularity of the content in each of the content clusters; wherein s isiIs the name of the content, k is the content popularity ranking of the content in the whole network, 
For each content s in the content clusteriThe popularity of (a) is a parameter of Zipf distribution, N is the total number of contents in the whole network, kjIndicating the ranking of the popularity of content with sequence number j in the web-wide content.
6. The satellite network cache placement method based on regional user interest awareness according to claim 1, wherein the determining the number of the selected contents in each content cluster according to the popularity and the cache subspace specifically comprises:
using a formula
Calculating the number of the selected contents in the g content cluster; wherein w is the number of contents selected in the content cluster,
for caching whether content s exists in a subspacei，
Is a content siC is the ratio space ofgIs the cache subspace of the g-th content cluster.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110727166.3A CN113472420B (en) | 2021-06-29 | 2021-06-29 | Satellite network cache placement method based on regional user interest perception |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110727166.3A CN113472420B (en) | 2021-06-29 | 2021-06-29 | Satellite network cache placement method based on regional user interest perception |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113472420A CN113472420A (en) | 2021-10-01 |
| CN113472420B true CN113472420B (en) | 2022-05-24 |
Family
ID=77873624
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110727166.3A Active CN113472420B (en) | 2021-06-29 | 2021-06-29 | Satellite network cache placement method based on regional user interest perception |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113472420B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114827179A (en) * | 2022-05-05 | 2022-07-29 | 鹏城实验室 | Satellite grid-oriented caching method and related device |
| CN115909083B (en) * | 2022-10-31 | 2023-08-08 | 中国科学院软件研究所 | Satellite earth observation discrete interest point clustering planning method and device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107404530A (en) * | 2017-07-27 | 2017-11-28 | 北京邮电大学 | Social networks cooperation caching method and device based on user interest similarity |
| CN109951317A (en) * | 2019-02-18 | 2019-06-28 | 大连大学 | A kind of buffer replacing method of the popularity sensor model based on user's driving |
| CN111641450A (en) * | 2020-06-02 | 2020-09-08 | 西安电子科技大学 | Satellite-ground integrated network communication and cache resource joint scheduling method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200305042A1 (en) * | 2019-06-28 | 2020-09-24 | Intel Corporation | Interest packet routing in information centric networks |
-
2021
- 2021-06-29 CN CN202110727166.3A patent/CN113472420B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107404530A (en) * | 2017-07-27 | 2017-11-28 | 北京邮电大学 | Social networks cooperation caching method and device based on user interest similarity |
| CN109951317A (en) * | 2019-02-18 | 2019-06-28 | 大连大学 | A kind of buffer replacing method of the popularity sensor model based on user's driving |
| CN111641450A (en) * | 2020-06-02 | 2020-09-08 | 西安电子科技大学 | Satellite-ground integrated network communication and cache resource joint scheduling method |
Non-Patent Citations (7)
| Title |
|---|
| Jian Li 等.An ICN/SDN-Based Network Architecture and Efficient Content Retrieval for Future Satellite-Terrestrial Integrated Networks.《IEEE Network》.2019, * |
| Laura Galluccio 等.Caching in information-centric satellite networks.《2012 IEEE International Conference on Communications (ICC)》.2012, * |
| Zhiguo Liu 等.User-Driven Cache Replacement Strategy for Satellite-Terrestrial Networks Based on SDN.《2020 IEEE 6th International Conference on Computer and Communications (ICCC)》.2021, * |
| 徐冀 等.多层卫星网络数据缓存技术研究.《载人航天》.2019,第25卷(第4期), * |
| 朱杰.基于SDN与ICN的卫星网络路由与缓存策略研究.《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》.2019, * |
| 王海月 等.RIDSI:一种融合SDN和兴趣域划分的ICN路由机制.《小型微型计算机系统》.2020,第41卷(第01期),第132-138页. * |
| 胡跃峰 等.Web高速缓存及其在卫星Internet中的应用.《中国有线电视》.2005,(第06期),第566-568页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113472420A (en) | 2021-10-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhong et al. | A deep reinforcement learning-based framework for content caching | |
| Ioannidis et al. | Adaptive caching networks with optimality guarantees | |
| Zhang et al. | Smart proactive caching: Empower the video delivery for autonomous vehicles in ICN-based networks | |
| CN112020103B (en) | Content cache deployment method in mobile edge cloud | |
| Zhang et al. | Proactive video push for optimizing bandwidth consumption in hybrid CDN-P2P VoD systems | |
| CN113472420B (en) | Satellite network cache placement method based on regional user interest perception | |
| Dai et al. | Edge intelligence for adaptive multimedia streaming in heterogeneous internet of vehicles | |
| Khan et al. | On the application of agglomerative hierarchical clustering for cache-assisted D2D networks | |
| CN108063802B (en) | User position dynamic modeling optimization method based on edge calculation | |
| Li et al. | Learning-based hierarchical edge caching for cloud-aided heterogeneous networks | |
| Ghandeharizade et al. | PAVAN: a policy framework for content availabilty in vehicular ad-hoc networks | |
| CN112911614B (en) | Cooperative coding caching method based on dynamic request D2D network | |
| CN112887943B (en) | Cache resource allocation method and system based on centrality | |
| EP3193490B1 (en) | Method and system for distributed optimal caching of content over a network | |
| Garg et al. | In-network caching for hybrid satellite-terrestrial networks using deep reinforcement learning | |
| Liu et al. | Real-time search-driven caching for sensing data in vehicular networks | |
| Taheri et al. | Achieving Performability and Reliability of Data Storage in the Internet of Things | |
| CN114245422A (en) | Edge active caching method based on intelligent sharing in cluster | |
| Jia et al. | Social-aware edge caching strategy of video resources in 5G ultra-dense network | |
| Hoiles et al. | Risk-averse caching policies for YouTube content in femtocell networks using density forecasting | |
| Liu et al. | A regional interest-aware caching placement scheme for reducing latency in the LEO satellite networks | |
| Yu et al. | Analysis of distributed database access path prediction based on recurrent neural network in internet of things | |
| CN113159539A (en) | Joint green energy scheduling and dynamic task allocation method in multilayer edge computing system | |
| Yang et al. | An intelligent caching and replacement strategy based on cache profit model for space-ground integrated network | |
| Xing et al. | Deep reinforcement learning based congestion control mechanism for sdn and ndn in satellite networks |
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 |