CN106792995A - The user access method of content low time delay transmission is ensured in a kind of following 5G networks - Google Patents

Abstract

The invention discloses a user access method for ensuring low-latency transmission of content in a future 5G network, belonging to the field of mobile communication. Specifically, first build a small base station mobile network simulation scene with cache, divide users into several larger clusters according to the required content, and count the sum of the multicast group upper limit and the user acceptance upper limit of all base stations. And delete redundant clusters according to this limit, continue to judge if the user cluster is less than the sum of the upper limit of the multicast group, then filter and classify each cluster, and divide it into sub-clusters, so that the user cluster is equal to the sum of the upper limit of the multicast group ; Then, execute the KM algorithm on all sub-clusters and multicast frequency bands of the base station, and finally, the base station sends the user clustering results and access base station results to the users to complete the communication. The invention enables limited channel resources to serve more users, improves the service efficiency of the base station, reduces unnecessary wireless link transmission, reduces the total delay of the content distribution process, and minimizes the total delay of users in the system.

Description

A user access method that guarantees low-latency transmission of content in future 5G networks

technical field

The invention belongs to the field of mobile communication, and describes a user access method for ensuring low-latency transmission of content in a future 5G network.

Background technique

According to Cisco's white paper, mobile data traffic will account for about 75% of global data traffic in 2019. Due to the popularity of smart devices, mobile data traffic will show a growth spurt in the future. Today's macro cellular network capacity can no longer be further expanded to meet the growing traffic demand.

In order to solve this problem, a small cell (Small Cell) network is introduced. Under the coverage of the macro cellular network, a large number of FemtoCell, PicoCell, MicroCell, WiFi AP, etc. are deployed to form a heterogeneous network. In the heterogeneous small cell network, the dense deployment of small cell base stations greatly improves the system capacity and meets the network traffic requirements.

However, the rate at which the small cell base station acquires content from the content provider is limited by the rate of the wireless backhaul link, which further affects the user's service quality requirements (such as the user's service delay, etc.). Because the user's service delay includes the sum of the wireless backhaul link delay from the content provider to the serving base station and the forward link transmission delay from the serving base station to the user.

In order to solve this problem, a caching device is added to the base station and the content obtained from the content provider is stored in the caching device of the base station in advance, so that the user can directly obtain the content from the serving base station; The load of the wireless backhaul link to the base station can reduce the delay and delay jitter from the content provider to the base station to ensure the user's delay requirement.

In the existing research on the small base station with caching, it is mostly assumed that users in the system have similar demand probabilities for content.

For example, Document 1: Khreishah A, Chakareski J, Gharaibeh A. Joint caching, routing and channel allocation scheme in cooperative small cell network [J].arXiv preprint arXiv:1605.09307,2016;

Document 2: Li H, Wang Z, Hu D. A Load Balancing Scheme for Combined Wireless Link and Backhaul in Buffer-Based Small Cell Networks [C]//Personal, Indoor and Mobile Radio Communications (PIMRC), 2015 IEEE No. 26 Annual International Symposium. IEEE, 2015:1889-1894.

Document 3: Khreishah A, Chakareski J. Cooperative caching in multi-small cell cooperative system [C]

//2015 IEEE Symposium on Computer Communication (INFOCOM WK-SHPS).IEEE,2015:257-262.

Document 4: Liu R, Yin H, Cai X, et al. Collaborative caching scheme in content-oriented network [J]. IEEE Communications, 2013, 17(4): 781-784.

The above documents [1-4] did not use the user clustering method according to the similarity of the user request content in the small cell base station loaded with the cache device and the selection scheme of the user access network. Afterwards, unified access can improve the utilization rate of the cache.

Literature [5] Dehghan M, Seetharam A, Jiang B, etc. Complexity of Optimal Routing and Content Caching Schemes in Heterogeneous Networks [C]//2015 IEEE Computer Communication Conference (INFOCOM).IEEE,2015:936- 944. The problem of optimizing joint request routing addressing and content caching in heterogeneous networks is studied, and the goal is to optimize the average content access delay. However, this literature considers the needs of each user independently, without considering the similarity of content requests between users.

Literature [6] ElBamby M S, Bennis M, Saad W, etc. Content-based User Clustering and Caching in Wireless Small Cell Networks [C]//The 11th International Symposium on Wireless Communication Systems (ISWCS) in 2014.IEEE ,2014:945-949. A joint user clustering and caching scheme is proposed in a wireless small cell network. The proposed scheme allows users to be divided into different clusters by using social similarity, and then each cluster will be connected to Suitable small cell base stations. In this way, the small cell base station can effectively buffer the most popular content, reducing service delay.

Simulation results show that the proposed algorithm outperforms random caching schemes and non-clustering learning algorithm schemes in terms of service delay and offload gain by deploying popular content close to small cell users.

Although there have been some studies on user access network selection based on content caching in heterogeneous small cell networks, most of them ignore the similarity of user request content. If users are clustered and accessed according to content requirements, the utilization rate of cache resources of the small cell base station can be further improved. In addition, existing studies mostly assume that small cell base stations provide services in a unicast manner, and most studies ignore the benefits brought by multicast. If the small cell base station provides the content distribution service to the clustered users in a multicast manner, the channel utilization rate of the small cell base station can be further improved.

Contents of the invention

The present invention proposes a user access method that guarantees low-latency transmission of content in the future 5G network, which minimizes the overall delay of the system while meeting user needs as much as possible and improving spectrum efficiency.

Specific steps are as follows:

Step 1. For a certain macro base station, establish a communication scenario of requesting content between a small cell base station within coverage and a user;

The communication scenario includes: K small cell base stations with cache, I user who initiates the request and J request contents;

The set of K small cell base stations is SBS _k represents the kth small cell base station;

The set of I users is UE _i represents the i-th user;

The set of J request contents is CT _j indicates the jth content.

Step 2, respectively defining the association relationship between the small cell base station, the user and the request content in the communication scenario;

It is set that all small cell base stations use multicast to provide services to users; and each user can only connect to one small cell base station at any time; each user can only request one content at a time, and each content are the same size.

First, the association relationship between the user and the requested content is represented by an association matrix D;

element is an indicator variable representing user i’s demand for content j, if user i has demand for content j, then otherwise,

Then, the association relationship between the content and the small cell base station is represented by an association matrix N;

element Indicates the demand indicator variable of the content j on the kth small cell base station, if the content j has a demand on the kth small cell base station, then otherwise,

Finally, the coverage list from the small cell base station to the user is represented by an association matrix L;

L(i,k)= _{li ik}

The element _{li ik} represents an indicator variable of whether user i is covered by the kth small cell base station, if user i is covered by the kth small cell base station, _{li ik} =1, otherwise, _lik =0.

The element whose value is 1 in the i-th row of the correlation matrix L and the corresponding column label k form a set which is It is used to represent the set of base stations that user i can choose to access.

The association relationship between the user and the small cell base station is represented by an association matrix A;

A(i,k)=a _ik

The element a _ik is an indicator variable for user i to access the kth small cell base station, if user i accesses the kth small cell base station, then _aik =1, otherwise, _aik =0.

Step 3. Define user clusters and content collections according to the relationship between users and requested content;

For content j, all users who need content j are divided into the same cluster U _j , expressed as follows;

For each user cluster that is not empty, record the content requested by the user in all clusters to form a set of the number of requested content

Step 4. According to the association relationship between the user and the small cell base station, define the number of user access for each small cell base station, which is the same as the total number of users in the respective cluster groups that can be accommodated;

The formula is as follows:

is the total number of users who have accessed the kth small cell base station;

Represents the clusters of all users who have demand for content j for the cached content j on the kth small cell base station;

m(k) is the tag value of the cluster, which is related to the kth small cell base station;

Step 5. Perform initial clustering of all users according to the content of the request according to the definition of the cluster, and initialize the access users of each small cell base station;

Initially, all users who have needs for the same content are divided into the same cluster; the number of user clusters formed by initialization is the same as the number of content requested by users in the system same;

Initially, count the number of users who have accessed each small cell base station; if the kth small cell base station has no new user access, then

At the same time, the delay matrix is initially defined as Indicates the transmission delay from the user to the small cell base station, where the element t _ik is the transmission delay when user i accesses the kth base station;

Step 6. Each small cell base station updates its actual available multicast group upper limit and the user upper limit that can be accessed;

Updating the upper limit of the multicast group of the kth small cell base station is:

N _k represents the original multicast group upper limit of the kth small cell base station;

Updating the upper limit of access users of the kth small cell base station is:

F _k is the original upper limit number of access users of the kth small cell base station;

Step 7. Determine the number of user clusters formed by initialization Whether it is greater than the sum of the updated multicast group upper limit of K small cell base stations i.e. whether there is If so, discard the clusters one by one from the cluster with the least number of users until Go to step eight; otherwise, go to step eight;

Step 8. Update all user serial numbers in the remaining clusters to form a set And determine the total number of users after the update Whether it is greater than the sum of the updated access user upper limit of K small cell base stations That is, is it satisfied If so, discard the clusters one by one from the cluster with the least number of users until Go to step nine; otherwise, go to step nine.

Step 9. Determine whether the updated number of user clusters satisfies If yes, go to step 10 to check the consistency of user coverage in each cluster; otherwise, when satisfying Go to step eleven;

Step ten, split each user cluster into several sub-clusters with more consistent coverage;

Specific steps are as follows:

Step 1001, when the cluster is discarded, update the user set from the remaining cluster Sort all users by number from small to large;

Step 1002, select the user whose current number is i' as a reference, and select a cluster where user i' is located according to the requested content j' in the cluster where user i' is currently located;

initial

Step 1003, sequentially select the remaining users i" in the cluster, and calculate the access set of user i" The access set of each user in the cluster is counted separately.

Step 1004, sequentially determine whether there are at most two different values in the access set between user i" and reference user i', that is, whether there are If yes, divide user i" into the sub-cluster where reference user i' is located, and enter step 1005, otherwise directly enter step 1005;

The initial element of the subcluster where the reference user i' is located is only the reference user i';

Step 1005, select the next user in the cluster in turn, return to step 1004, until all the users in the cluster are compared, and get two clusters: the sub-cluster where the reference user i' is located and the cluster formed by the remaining users;

Step 1006, judging whether the number of clusters after splitting satisfies If yes, the loop ends; otherwise, go to step 1007.

Step 1007, for the clusters formed by the remaining users, select the user with the smallest number as the reference user, and return to step 1003 until the cluster can no longer be divided;

The sub-cluster label of this cluster is represented by m'; Denotes the m'th sub-cluster of content j';m'={1,2,3...}.

Step 1008, update the user set from the remaining clusters Select the next benchmark user in , and return to step 1002 until the number of split clusters satisfies Then the loop ends.

Step 11: For the generated user sub-clusters and the upper limit of the actual available multicast group, execute the best matching KM algorithm according to the delay weight.

The delay weight refers to the sum of the delays corresponding to all users in each sub-cluster to the base station;

According to the user subcluster obtained in step 10 or the update cluster obtained in step 7, a bipartite graph is constructed in combination with the actual multicast group upper limit, where the weight of an edge is the sum of the time delays from all users in the corresponding user cluster to the small cell base station. The matching problem between user clusters and each channel of all small cell base stations is mapped to a weighted bipartite graph to find the best matching problem.

Step 12, according to the result obtained by the KM algorithm, the label of each user sub-cluster is corresponding to the base station allocated by it, which is about to becomes Update base station admission set And send the serial number of the base station that it should access to each user, and the communication ends.

The advantage of the present invention is: a user access method that guarantees low-latency transmission of content in the future 5G network, clusters users based on the similarity of user content requests, and users access the serving base station in the form of clusters, and at the same time The base station serves users in the form of multicast. This scheme uses the combination of clustering algorithm and KM algorithm to solve discrete problems, and realizes a user access scheme that guarantees low-latency transmission of content with low complexity.

Description of drawings

FIG. 1 is a schematic diagram of a user cluster communication scenario based on content caching between a small cell base station and a user according to the present invention;

2 is a flowchart of a user access method for ensuring low-latency transmission of content in a future 5G network according to the present invention;

Fig. 3 is the flow chart of the method that each user cluster is split into several sub-clusters in the present invention;

Fig. 4 is the relationship figure of user number and system total time delay under three kinds of algorithms of the present invention;

Fig. 5 is the relationship diagram of content number and system total time delay under three kinds of algorithms of the present invention;

Fig. 6 is a relationship diagram between the number of base stations and the total system delay under the three algorithms of the present invention.

specific embodiment

The specific implementation method of the present invention will be described in detail below in conjunction with the accompanying drawings.

When selecting an access network for a user, the present invention comprehensively considers the specific content requirements of the user in the current system, the load of the base station, the restriction on the number of multicast group accesses, and the restriction on the number of users; The degree of user clustering, the user accesses the serving base station in the form of a cluster, and the base station provides content services to the user in the form of multicast; put the users with the same content requirements in the same cluster, and then cover according to the base station The difference of the situation is divided into different sub-clusters, and the base station selects the multicast group based on these sub-clusters, so that limited channel resources can serve more users, and improve the service efficiency of the base station, thereby reducing unnecessary wireless link transmission , to reduce the total delay of the content distribution process and minimize the total delay of users in the system.

The present invention first builds a small base station mobile network simulation scene with cache, divides users into several larger clusters according to the required content, and counts the sum of the multicast group upper limit of all base stations and the user acceptance upper limit of all base stations sum, and delete redundant clusters according to this limit, continue to judge if the user cluster is less than the sum of the upper limit of the multicast group, then filter and classify each cluster, and divide it into sub-clusters, so that the user cluster is equal to the multicast group The sum of the upper limit; then, execute the KM algorithm for all sub-clusters and multicast frequency bands of the base station, and finally, the base station sends the user clustering results and access base station results to the users, and notifies each base station to prepare to receive user requests and complete the communication.

The specific implementation steps are shown in Figure 2, as follows:

Step 1. For a certain macro base station, establish a communication scenario of requesting content between a small cell base station within coverage and a user;

As shown in FIG. 1 , consider the downlink of a cellular network of K small cell base stations with buffers under the coverage of a macro base station, and the communication system includes I users and J contents. The set of K small cell base stations is SBS _k represents the kth small cell base station;

The set of I users is UE _i represents the i-th user;

The set of J request contents is CT _j indicates the jth content.

Step 2, respectively defining the association relationship between the small cell base station, the user and the request content in the communication scenario;

Assume that all small cell base stations use multicast to provide services to users; each user can only request one content at any time, and each user can only connect to one small cell base station at any time; The contents are the same size.

Define the association matrix D to represent the association relationship between the user and the requested content, where the element is an indicator variable of user i’s demand for content j, and its element It is defined as follows:

Define the association matrix N, which represents the association relationship between the content and the small cell base station, where the element is the demand indicator variable of the content j on the kth small cell base station, and its element It is defined as follows:

Define an association matrix L to represent the coverage list from the small cell base station to the user, where the element _lik is the indicator variable whether user i is covered by the small cell base station k, and its element L(i, k)= _lik is defined as follows:

The element whose value is 1 in the i-th row of the correlation matrix L and the corresponding column label k form a set which is It is used to represent the set of base stations that user i can choose to access.

Define an association matrix A to represent the association relationship between users and small cell base stations, where the element a _ik is an indicator variable that uses i to access the kth base station:

Step 3. Define user clusters and content collections according to the relationship between users and requested content;

User clustering is determined by content requirements:

For content j, all users who need content j are divided into the same cluster U _j , expressed as follows;

For each user cluster that is not empty, record the content requested by the user in all clusters to form a set of the number of requested content At the same time as the label set of the user's initial cluster:

Step 4. According to the association relationship between the user and the small cell base station, define the number of user access for each small cell base station, which is the same as the total number of users in the respective cluster groups that can be accommodated;

The groups accessed by small cell base stations are determined by user access conditions: Indicates the users connected under the small cell base station k;

The relationship between user clustering and small cell base station access grouping: Indicates that the number of users accessed under the small cell base station k is consistent with the sum of users in the number of accessed user clusters;

is a subset of U _j whose tag value is m(k), corresponding to the kth small cell base station;

Step 5. Perform initial clustering of all users according to the content of the request according to the definition of the cluster, and initialize the access users of each small cell base station;

Initially, all users who have needs for the same content are divided into the same cluster; the number of user clusters formed by initialization is the same as the number of content requested by users in the system same;

Initially, count the number of users who have accessed each small cell base station; if the kth small cell base station has no new user access, the small cell base station access set

At the same time, the delay matrix is initially defined as Indicates the transmission delay from the user to the small cell base station, where the element t _ik is the transmission delay when user i accesses the kth base station:;

Step 6. Each small cell base station updates its actual available multicast group upper limit and the user upper limit that can be accessed;

Updating the upper limit of the multicast group of the kth small cell base station is:

N _k represents the original multicast group upper limit of the kth small cell base station;

Updating the upper limit of access users of the kth small cell base station is:

F _k is the original upper limit number of access users of the kth small cell base station;

Step 7. Determine the number of user clusters formed by initialization Whether it is greater than the sum of the updated multicast group upper limit of K small cell base stations i.e. whether there is If so, discard the clusters one by one from the cluster with the least number of users until Go to step eight; otherwise, go to step eight;

Step 8. Update all user serial numbers in the remaining clusters to form a set And determine the total number of users after the update Whether it is greater than the sum of the updated access user upper limit of K small cell base stations That is, is it satisfied If so, discard the clusters one by one from the cluster with the least number of users until Go to step nine; otherwise, go to step nine.

Step 9. Determine whether the updated number of user clusters satisfies If yes, go to step 10 to check the consistency of user coverage in each cluster; otherwise, when satisfying Go to step eleven;

Step ten, split each user cluster into several sub-clusters with more consistent coverage;

As shown in Figure 3, the specific steps are as follows:

Step 1001, when the cluster is discarded, update the user set from the remaining cluster Sort all users by number from small to large;

Step 1002, select the user whose current number is i' as a reference, and select a cluster where user i' is located according to the requested content j' in the cluster where user i' is currently located;

initial

Step 1003, sequentially select the remaining users i" in the cluster, and calculate the access set of user i" The access set of each user in the cluster is counted separately.

Step 1004, sequentially judge whether there are at most two different values in the access set between user i" and reference user i', that is, whether there are If yes, divide user i" into the sub-cluster where reference user i' is located, and enter step 1005, otherwise directly enter step 1005;

The initial element of the subcluster where the reference user i' is located is only the reference user i';

Step 1005, select the next user in the cluster in turn, return to step 1004, until all the users in the cluster are compared, and get two clusters: the sub-cluster where the reference user i' is located and the cluster formed by the remaining users;

Step 1006, judging whether the number of clusters after splitting satisfies If yes, the loop ends; otherwise, go to step 1007.

Step 1007, for the clusters formed by the remaining users, select the user with the smallest number as the reference user, and return to step 1003 until the cluster can no longer be divided;

The sub-cluster label of this cluster is represented by m'; Denotes the m'th sub-cluster of content j';m'={1,2,3...}.

Step 1008, update the user set from the remaining clusters Select the next benchmark user in , and return to step 1002 until the number of split clusters satisfies Then the loop ends.

Step 11: For the generated user sub-clusters and the upper limit of the actual available multicast group, execute the best matching KM algorithm according to the delay weight.

The delay weight refers to the sum of the delays corresponding to all users in each sub-cluster to the base station;

The present invention carries out mathematical modeling with the goal of minimizing the total time delay of the system. The user's service delay is the sum of the forward link transmission delay from the serving base station to the user and the wireless backhaul link delay from the content provider to the serving base station. In the model of the present invention, it is assumed that the content requested by the user must first be cached on the base station before being forwarded to the user. The backhaul link delay from the content provider to the base station is equivalent to the forward link delay from the base station to the user. Therefore, in the current access allocation period, the request of the user whose requested content is not in the cache of the base station will be postponed to the next access allocation period. Moreover, since the content distribution process from the content provider to the serving base station should be performed in the previous period, the optimized objective function here does not consider the time delay from the content provider to the serving base station.

The goal of system optimization in the present invention is to minimize the total system delay, that is, the objective function should be expressed as:

t _ik is the time delay for user i to access the kth small cell base station SBS _k ;

The total system delay is subject to the following constraints:

N _k represents the upper limit of the multicast group of the k-th small cell base station; c _k is the buffer list vector of the k-th base station, which contains the information whether contents 1 to J are cached in base station k.

Among the constraints, the constraint C ₁ means that the i-th user can and can only access one small cell base station at the same time;

Constraint C ₂ means that the number of access users of the kth small cell base station must not exceed the upper limit F _k of users that the small cell base station can access;

Constraint _C3 means that user groups with different content requirements accessed by the kth small cell base station cannot exceed the broadcast group upper limit N _k ;

Constraint C ₄ means that the i-th user can only request at most one content at the same time;

Constraint C ₅ means that for the kth small cell base station, the total number of user accesses should be equal to the sum of the number of users who demand all the content of the small cell base station. That is, the total number of users connected to the base station of the kth small cell The total number of all users in the user cluster set requesting the cached content on the small cell base station is the same.

The model finally needs to solve the user's access selection scheme, that is, the matrix A, to obtain the best match between the user and the small cell base station. According to the user subcluster obtained in step 10 or the update cluster obtained in step 7, a bipartite graph is constructed in combination with the actual multicast group upper limit, where the weight of an edge is the sum of the delays from all users in the corresponding user cluster to the small cell base station. The matching problem between user clusters and each channel of all small cell base stations is mapped to a weighted bipartite graph to find the best matching problem.

In order to analyze this problem, the definition of bipartite graph in graph theory and the assumptions used in the present invention are firstly introduced.

The weighted bipartite graph can be expressed as G=(X, Y, E). Among them, G represents a complete bipartite graph; X represents a user cluster; Y represents all candidate small cell base station channels; E is a set of edges in the bipartite graph, representing the link between the user cluster and the small cell base station channel; edge xy The weight of is the sum of the delays from all users in the subcluster to the base station, expressed as w(xy). When a user group determines to access a channel of a small cell base station, that is, an IR in X is determined to be associated with an IR in Y, it is called that the IR in X (or in Y) is allocated. A subgraph containing all vertices and the edges between the assigned vertices is called a matching of graph G. Let S be a subgraph of G, and N _G (S) is called the complement graph of S in G, which is defined as the subgraph obtained by removing the edge set and top set of S from the edge set and top set of G. If M is a matching in graph G, on a track P(u,v) in G, u and v are vertices in X and Y that are not allowed by M respectively, but the edges on P(u,v) alternate If the ground does not appear in M and appears in M, then P(u,v) is called an augmentable orbit of M. In order to apply the KM algorithm to the weighted bipartite graph, it is also necessary to assign a top label l(v) to each top v, when x∈X and y∈Y and l(x)+l(y)≥w(xy), This kind of top mark is called normal top mark. The initial normal top marks referred to in the following steps follow the following definitions:

The specific steps of the KM algorithm are as follows:

Step 1101, with all user sub-clusters (that is, multicast groups) and base station acceptance set is the top, construct graph G; select the initial normal top mark for each top of graph G, construct graph G _l , and determine the initial matching M in G _l ;

Step 1102, if all IMs in X are allocated by M, go to Step 1104; otherwise, take the IM u in _G1 that is not allocated by M, let S={u},

Step 1103, if Go to step 1104, if Then take:

Then order

Step 1104, select An item y in , if y has been betrothed by M, and yz∈M, then S∪{z}→S, T∪{y}→T, go to step 1103; otherwise, S∪{z}→S, take An M in G _l can augment the orbit P(u,y), so that Go to step 1102;

Step 1105, if the number card(A)=I of the elements in the set A or the top of X is all allocated by M, then stop, M is the best match sought, and output M;

Step 12, according to the result obtained by the KM algorithm, the label of each user sub-cluster is corresponding to the base station allocated by it, which is about to becomes Update base station admission set And send the serial number of the base station that it should access to each user, and the communication ends.

Obtain each user sub-cluster (that is, multicast group) from M and base station acceptance set The corresponding access relationship, and send each user the serial number of the base station that it should access, and the communication ends.

Specific examples:

The simulation scenario is set to a common simulation network configuration of a heterogeneous network. The communication system is set as follows (the settings are not used as variables): 5 small cell base stations, the load capacity of each base station is 30 users and a random number of multicast groups, and each base station has pre-cached content. There are 200 users randomly distributed within a radius of 1km around the base station. The total number of content in the system is 20, and each content has the same size.

This application mainly conducts simulation analysis from the aspects of the impact of the increase of users on the total system delay, the impact of the increase of content on the total system delay, and the impact of different numbers of base stations on the total system delay. In order to reflect the performance of the proposed strategy, two algorithms, Random Allocation Algorithm (RA) and Average Allocation Algorithm (EA), are used for comparison.

As shown in Figure 4, it shows the relationship between the number of users and the total system delay when the number of base stations and the number of content are fixed for KM algorithm, EA algorithm and RA algorithm. It can be seen that as the number of users increases, the total system delay also increases. This is because the number of users in the system increases, and the number of users who may access also increases, so the total system delay also increases. In addition, the total system delay under the KM algorithm is lower than that under the other two algorithms, because the KM algorithm obtains the optimal solution, which overcomes the shortcomings of the other two algorithms in general minimization. In addition, it can be seen from the running time of the simulation that the suboptimal solutions obtained by the EA algorithm and the RA algorithm are not very different, but the algorithm complexity is obviously lower than that of the KM algorithm. It can be seen that although the traditional general minimization algorithm has low performance, it is still widely used because it saves time when the computing and processing hardware is backward in the past.

As shown in Figure 5, when the number of users and the number of base stations are fixed, the relationship between the number of content and the total system delay under different numbers of users (150 users and 200 users) is shown. It can be seen that as the number of content increases, the total energy consumption of the system decreases. This is because although the number of content in the system increases, the average number of users who are interested in each content is decreasing (because the number of users remains unchanged), so when the number of clusters increases, the number of users in each cluster decreases. The delay becomes smaller. Also, the number of users is 150 and lower than the curve for the number of users of 200. The reason is also that as the number of users in each content cluster becomes smaller, the overall latency is lower.

As shown in Fig. 6, when the number of users and the number of contents are fixed, the relationship between the number of different base stations and the total system delay is shown. It can be seen that as the number of base stations increases, the total system delay increases. This is because the number of base stations in the system increases, and the number of users that can be accessed also increases, so the total system delay also increases. In addition, it can be clearly observed that the time delay increases significantly when the number of base stations increases from 3 to 5. The reason is that when the number of base stations is 3, users cannot fully access the base stations, and some users are rejected for content requests due to the combination of exceeding the number of base station accesses. When the number of base stations is 5 or more, users can already fully access the base stations, so users who need some unstored content (that is, content with a large transmission delay) can also access the base station, so that The total latency becomes relatively higher. It can be seen that the user access selection scheme proposed by the present invention has a great performance improvement in reducing the total delay of the current mobile network system when the base stations are not yet densely deployed and the small base stations are usually overloaded.

The present invention is generally divided into the following three stages: user clustering stage, base station detection stage and access selection stage;

In the stage of user clustering, according to the specific content requirements of all users in the system, users with the same content requirements are divided into the same large cluster. After clustering, the number of existing clusters should not be greater than the total number of contents.

In this stage, users are naturally divided into several large clusters according to content request information sent by users to the small cell base station. The reason why users are clustered according to different content requests is that the communication scenario assumed by the present invention uses multicast technology, and in order to maximize the advantages of multicast technology, it is necessary to make requests with the same content (that is, through Users of the same multicast group get content) are in the same cluster at the beginning.

In the base station detection phase, the small cell base station needs to count how many multicast groups it has accepted, and also needs to count the number of users. The new multicast group upper limit and user access upper limit are obtained after the users who have already connected are removed from the access upper limit. Then filter and classify each cluster in the first step according to the user coverage table, and divide it into several subclusters that cannot be further divided. The so-called "non-dividable" means that all users in the sub-cluster can choose to access the same range of small cell base stations.

The above classification criteria and removal methods are also in line with the results of intuitive observation: after the user cluster is completed, it can first be calculated by user cluster whether it exceeds the upper limit of small cell base station access. At this time, the access limit is multicast group The number of user clusters, that is, the total number of user clusters cannot exceed the sum of the upper limit of all base station multicast groups; then after the cluster with a small number of users is determined to be discarded, it will no longer be possible to observe the limit of the number of users from the number of small cell base station multicast groups , so it is necessary to specifically calculate the total number of users in the remaining user clusters at this time, and compare it with the sum of the upper limit of the total number of users that each small cell base station can accommodate, and then remove the cluster with a small number of users; finally, if the second step is executed using If the number of user clusters is less than the sum of the upper limit of the multicast group of the small cell base station, the larger cluster can be divided into smaller clusters, so that the number of user clusters is just equal to the sum of the upper limit of the base station multicast group, so as to make full use of the multicast communication. Advantages to bring better service experience to users.

In the access selection stage, for the purpose of optimizing the objective function, the delay limit of the user and the transmit power limit of the small cell base station are calculated. Execute the KM algorithm on all sub-clusters and multicast frequency bands of small cell base stations. The obtained result is the multicast realization scheme that the user accesses according to the content.

It is worth noting that the user cluster access network in this step chooses to use the centralized KM algorithm, and the centralized algorithm has certain requirements for the specific executor of the algorithm. In the present invention, it is assumed that the entity executing this algorithm is all small cell base stations, and each small cell base station adopts a distributed calculation method, and carries out data exchange between small cell base stations along with the calculation; the calculation process is centralized management, The operation process is managed by a base station selected by a specific algorithm. Its data exchange time delay is small, and it is convenient to order each small cell base station to perform its own specific operation steps for this algorithm. The selection algorithm for managing small cell base stations is not involved in the present invention, and when the KM algorithm is executed, it is considered that the management base station has been selected after the end of the base station detection phase and before the start of the access selection phase.

The present invention clusters users based on the similarity of user content requirements, the base station adopts the method of multicasting to send content, and uses the combination of the clustering algorithm and the KM algorithm to realize the scheme of user access selection; allocate the base stations that users should access , to minimize the total delay of the system with acceptable complexity.

Claims (4)

1. the user access method of content low time delay transmission is ensured in a kind of following 5G networks, it is characterised in that specific steps are such as Under：

Step one, for certain macro base station, set up the communication of request content between the small-cell base station in coverage and user Scene；

Communication scenes include：The K small-cell base station with caching, the I user for initiating request and J request content；

K small-cell base station collection is combined intoSBS_kRepresent k-th small-cell base station；

I user's collection is combined intoUE_iRepresent i-th user；

J request content collection is combined intoCT_jRepresent j-th content；

Small-cell base station, the incidence relation between user and request content in step 2, respectively definition communication scenes；

Set all of small-cell base station and service is all provided the user using multicast mode；And each user is most at any time Can only connect on a small-cell base station；Each user can only at most ask a content, the size phase of each content every time Together；

Step 3, the incidence relation according to user and request content, define the cluster and properties collection of user；

For content j, by all users point for having demand to content j in same cluster U_j, it is expressed as follows；

Indicator variables of the user i to content j demands is represented, if user i has demand to content j,Otherwise,

For each user's cluster of non-NULL, the content being requested by a user in all clusters is recorded, form request content number set

Step 4, the incidence relation according to user and small-cell base station, define user's access number of each small-cell base station, with Total number of users in respective receptive cluster group is identical；

Formula is as follows：

K-th total number of users of small-cell base station has been accessed for all；a_ikIt is that user i accesses k-th finger of small-cell base station Show variable, if user i accesses k-th small-cell base station, a_ik=1, otherwise, a_ik=0；

Represent for the cache contents j on k-th small-cell base station, the cluster of all users for having demand to content j；

M (k) is the mark value of the cluster, related to k-th small-cell base station；

Step 5, all users are carried out into initial sub-clustering according to the content of request by the definition of cluster, and to each small-cell base station Accessing user initialized；

When initial, same cluster is divided into the user that certain identical content has demand by all；Initialize user's sub-clustering of formation The content number being requested by a user in number and systemIt is identical；

When initial, accessed user's number of each small-cell base station is counted；If k-th small-cell base station does not have new user to access,

Meanwhile, original definition time delay matrix isRepresent user to the propagation delay time of small-cell base station, wherein element t_ik Propagation delay time when being user i k-th base station of access；

Step 6, each small-cell base station update each actual available multicast class upper limit and the user's upper limit that can be accessed；

The multicast class upper limit for updating k-th small-cell base station is：

N_kRepresent k-th original multicast class upper limit of small-cell base station；Represent demands of the content j on k-th small-cell base station Indicator variable, if content j has demand on k-th small-cell base station,Otherwise,

The upper limit for updating k-th small-cell base station accessing user is：

F_kIt is k-th original accessing user's upper limit number of small-cell base station；

Step 7, the number for judging user's sub-clustering that initialization is formedMulticast after whether being updated more than K small-cell base station Class upper limit sumWhether haveIf it is, since the minimum cluster of number of users, this is given up one by one Cluster, untilInto step 8；Otherwise, into step 8；

Step 8, all user's sequence numbers updated in remaining cluster constitute setAnd judge the user's total number after updatingWhether Accessing user's upper limit sum after being updated more than K small-cell base stationWhether meetIf it is, Then since the minimum cluster of number of users, the cluster is given up one by one, untilInto step 9；Otherwise, into step Nine；

Step 9, judge update after user's sub-clustering number whether meetIf it is, into step 10, checking The uniformity of user's coverage in each cluster；Otherwise, satisfaction is worked asInto step 11；

Step 10, each user's cluster is split it is divided into the more consistent submanifold of several coverages；

Step 11, to user's submanifold for generating and can actually use multicast class upper limit, best match KM calculation is performed by time delay weights Method；

Time delay weights refer to time delay summation of all users of correspondence to base station in each submanifold；

Step 12, the result tried to achieve by KM algorithms, the label of each user's submanifold is corresponded to the base station of its distribution, updates base Receiving of standing is gatheredAnd send its base station sequence number that should be accessed, sign off to each user.

2. the user access method of content low time delay transmission is ensured in a kind of following 5G networks as claimed in claim 1, and it is special Levy and be, described step two is specially：

First, user represents with the incidence relation incidence matrix D of request content；

$D(i,j)=d_i^j$

Then, content is represented with the incidence relation incidence matrix N of small-cell base station；

$N(k,j)=n_k^j$

Finally, small-cell base station is represented to the covering tabulation incidence matrix L of user；

L (i, k)=l_ik

Element l_ikWhether user i is represented by k-th indicator variable of small-cell base station covering, if user i is by k-th cell Base station covers, then l_ik=1, otherwise, l_ik=0；

The element that incidence matrix L the i-th row intermediate values are 1, corresponding row mark k are constituted into setI.e.For table Show that user i may be selected the set of the base station of access；

User represents with the incidence relation incidence matrix A of small-cell base station；

A (i, k)=a_ik。

3. the user access method of content low time delay transmission is ensured in a kind of following 5G networks as claimed in claim 1, and it is special Levy and be, described step ten is comprised the following steps that：

Step 1001, after cluster is given up, gather from the renewal user of remaining clusterIt is middle to arrange all users from small to large by number Sequence；

Step 1002, selection current number are in the user of i ' asks as benchmark, in the cluster according to active user i ' places Hold j', choose certain cluster at user i ' places；

Initially

The access set of step 1003, remaining user i chosen successively in the cluster ", calculating user i "In counting the cluster respectively The access set of each user；

Step 1004, judge user i successively " in access set and reference user i ' between with the presence or absence of at most two differences Whether value, i.e., haveIf it is, by user i " submanifold at reference user i ' places is partitioned into, into step 1005, otherwise it is directly entered step 1005；

The submanifold finite element at reference user i ' places is only reference user i '；

Step 1005, the next user chosen successively in the cluster, return to step 1004, until all of user compares in the cluster Relatively finish, obtain two clusters：The cluster that the submanifold and remaining users at reference user i ' places are formed；

Step 1006, judgement split it is point complete after number of clusters mesh whether meetIf it is, circulation terminates；Otherwise, Into step 1007；

Step 1007, the cluster formed for remaining users, choose the user of lowest number as reference user, return to step 1003, can not be untill division until the cluster；

The submanifold label m' that the cluster is divided altogether is represented；Represent the m' submanifold of content j '；M'={ 1,2,3... }；

Step 1008, from setIt is middle to choose next reference user, return to step 1002, until the number of clusters mesh satisfaction split pointThen circulation terminates.

4. the user access method of content low time delay transmission is ensured in a kind of following 5G networks as claimed in claim 1, and it is special Levy and be, described step ten one is specially：The renewal cluster that the user's submanifold or step 7 obtained according to step 10 are obtained, knot Close actual multicast class upper limit construction bipartite graph, wherein the power on side in correspondence user's cluster all users to small-cell base station when Prolong sum；Matching problem between user's cluster and each channel of all small-cell base stations is mapped as weighting bipartite graph and solves most preferably The problem of matching.