A method for deploying edge network resources in a dynamic communication environment

Abstract

The present invention discloses an edge network resource deployment method suitable for dynamic propagation environments. The method takes into account the dynamic changes of content popularity during the propagation process, establishes a content propagation model, analyzes the mutual influence of network resource deployment and content popularity, constructs a dynamic relationship between edge network resource deployment and content popularity, designs a base station link resource allocation scheme, obtains an optimized edge network resource deployment strategy, and improves the service performance and resource utilization efficiency of the edge network.

Description

Technical Field

[0001] The present invention relates to the technical field of network resource management, and in particular to an edge network resource deployment method suitable for a dynamic propagation environment. Background Art

[0002] With the increasing number of mobile devices and the rapid growth of internet data traffic, the load on core network access networks and backhaul links is increasing. The dramatic increase in demand for mobile data services poses a significant challenge to the efficient conversion of mobile communication network resources into service capabilities. Edge caching technology aims to reduce content retrieval latency and alleviate the burden on backhaul links by deploying resources at the edge of the network. By deploying content at edge nodes closer to users, edge networks effectively offload traffic from backhaul links, improving user experience and performance.

[0003] A classic edge caching strategy is Most Popular Caching (MPC), which assumes the popularity of content in edge networks is a static Zipf distribution. Edge base stations cache the most popular content within their coverage area until capacity is reached. Another approach is Least Frequently Used Caching (LFU), in which base stations also cache the most popular content within their coverage area until capacity is reached, but remove the least frequently used content with each update. Mean Probability Caching (MC) involves base stations caching all content with equal probability.

[0004] In the popularity-based cache deployment scheme, content popularity is assumed to be a static Zipf distribution, which does not take into account the dynamic changes in popularity. It is impossible to obtain a high cache hit rate, which affects the utilization efficiency of edge network resources. The cache deployment scheme based on usage frequency counts the request frequency of content in the period before each update and deletes the least frequently used content during the update. This strategy mainly deploys cache based on the content request frequency and content popularity before each update. It does not predict and estimate the popularity of the content after the update, and it also cannot achieve a high cache hit rate. The average probability cache strategy assumes that the popularity of all content is equal, which reduces the accuracy of edge caching. In addition, none of the above schemes take into account the mutual influence between network resource deployment and content popularity, which will affect the accuracy of edge network resource deployment. Summary of the Invention

[0005] Aiming at the problem that the existing technology cannot obtain a high cache hit rate, which affects the utilization efficiency and accuracy of edge network resources, the present invention proposes an edge network resource deployment method suitable for a dynamic propagation environment.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] The present invention provides an edge network resource deployment method applicable to a dynamic propagation environment, comprising the following steps:

[0008] S1. Users within the base station coverage area are divided into four states: susceptible, exposed, infected, and recovered. An edge network content propagation model based on SEIR propagation dynamics is constructed.

[0009] S2. The base station collects request status information for each content. Request status information consists of three parts: the number of users who have not requested the content, i.e., the number of susceptible users; the number of users currently making requests, i.e., the number of exposed users; and the number of users who have made requests and received the corresponding content, i.e., the sum of infected and recovered users. By analyzing the mutual influence of network resource deployment and content popularity, a dynamic relationship between edge network resource deployment and content popularity is constructed.

[0010] S3. Based on the constructed propagation dynamics model, in a dynamic propagation environment, with maximizing the user demand satisfaction rate as the optimization goal, a base station link resource allocation scheme is designed to obtain an optimized edge network resource deployment strategy; the user demand satisfaction rate is defined as the proportion of users who send requests within time T who receive the requested content.

[0011] Furthermore, in step S1, the edge network content propagation model based on SEIR propagation dynamics is constructed as follows:

[0012]

[0013] Among them, S f (P,A,t) represents the number of susceptible users at time t, E f (P,A,t) represents the number of users in contact state at time t, I f (P,A,t) represents the number of infected users at time t, R f (P,A,t) represents the number of users in the recovery state at time t, S f (P,A,t)+E f (P,A,t)+I f (P,A,t)+R f (P, A, t) = N, where N is the number of users at the edge network base station, the number of contents in the network is F, and the size of each content is c f bits, the amount of content that the base station can cache is C; the cache deployment strategy of the base station is P, P = {p1,p2,...,p F}, p f =1 means that the content f is cached by the base station, p f=0 means that the base station has no cache content f; the link allocation strategy of the base station is A, A={a1,a2,...,a F}, a f represents the ratio of link resources allocated by the base station to content f; β f μ represents the probability that a susceptible user will be attracted and interested in content f when contacting an infected user of type I. f The probability that a user loses interest in the content after acquiring it and switches from the infected state to the recovered state, D f Indicates the number of users who received the requested content per unit time.

[0014] Furthermore, the request status information in step S2 is represented as The state information i is expressed as

[0015] Furthermore, the probability β f and μ f By collecting historical data and using the least squares method, the specific process is as follows:

[0016] Establish the following system of equations:

[0017] Qe=B

[0018] Using the least squares method to obtain β f and μ f Approximate value of :

[0019] e=(Q T Q) -1 Q T B

[0020] in,

[0021] e=(β f ,μ f )

[0022]

[0023]

[0024]

[0025]

[0026]

[0027] Furthermore, the number of users D receiving the requested content per unit time f Expressed as:

[0028]

[0029] in, represents the number of users of D2D communication service per unit time, Indicates the number of users served by the base station per unit time, expressed as:

[0030]

[0031] Among them, R d It represents the total downlink rate of the base station, and υ represents the backhaul link rate of the base station.

[0032] Furthermore, the dynamic relationship between edge network resource deployment and content popularity in step S2 is constructed as follows:

[0033] The initial number of users in each state is recorded as [S f (0),E f (0),I f (0),R f (0)], the state transition equation of content f is solved by the constant variation method, and the result is as follows:

[0034]

[0035] In the actual content dissemination process, the dissemination dynamics model will evolve. f (P, A, t) = 0 and When , content propagation will evolve into the SIR model, and the state transition equation is as follows:

[0036]

[0037] E f (P, A, t) = 0 and This moment is recorded as T0 f ;

[0038] Applying the homotopy analysis method to solve the explicit solution of the SIR model, the results are as follows:

[0039]

[0040] Where M is the order, θ = μ f -β f S f (P,A,∞),S f (P, A, ∞) is calculated using the following expression:

[0041]

[0042] Among them, δ m,k , γ m,k is the coefficient, which is obtained by the following expression:

[0043]

[0044]

[0045] The calculation of relevant parameters is as follows:

[0046]

[0047]

[0048]

[0049]

[0050]

[0051] a m,j =χ 3m-j+1 (μ f -β f S f (P,A,∞)-jθ)δ m-1,j -β f χ j ε m,j ,1≤j≤3m+1

[0052] b m,j =χ 3m-j+1 (β f S f (P,A,∞)δ m-1,j -jθγ m-1,j )+β f χ j ε m,j ,1≤j≤3m+1

[0053]

[0054]

[0055] h is the convergence control parameter, When I w When the curve of h reaches a horizontal value, S is obtained by taking the value of h. f (P,A,t),I f (P,A,t) converges to an explicit solution.

[0056] Furthermore, the optimization objective of step S3 is expressed as:

[0057]

[0058] in, represents the change in the number of users in the infected and recovered states within time T, that is, the number of users who received the requested content, represents the sum of the change in the number of susceptible users and the initial number of contact users within time T, that is, the number of users who issued requests within time T, where

[0059] Furthermore, the optimization objective in step S3 is decoupled into two sub-problems and solved separately. The specific steps are as follows:

[0060] S31: Deploy base station cache resources P based on content requests. At the cache update time, sort the number of requests for each content at the current moment, that is, the number of contact users. The sorting result is recorded as The amount of content that the base station can cache is C, and the base station cache The content corresponding to the first C elements in The Cth element in is denoted as E f (0) th , all requests in the content set with a number greater than E f (0) th Cache the content;

[0061] S32: After the base station cache deployment strategy is solved, the optimization problem is transformed into:

[0062]

[0063] where 0≤a f ≤1, The optimization problem is transformed into:

[0064] argmaxη(A)·g(A)

[0065] in

[0066]

[0067]

[0068] Based on the message passing process of belief propagation, the link allocation ratio is discretized with a step size of Δa, and the link resource allocation design scheme is obtained through iterative calculation;

[0069] S33: Allocate the remaining resources of the base station based on a greedy algorithm.

[0070] Furthermore, in the iterative process of step S32, the nodes whose values ​​reach the upper and lower limits will no longer participate in the next iteration. In each round of iteration, the set of nodes participating in the iteration is C a , the set of nodes that do not participate in the iteration is U aAfter each round of iteration, the two sets are updated; the lower limit of the node is taken as Δa, recorded as When T=T0 f When a f The value of is recorded as the upper limit of the node use Indicates that the factor node l is passed to the variable node μ during the nth iteration i The message, in turn, uses Represents the variable node μ during the t-th iteration i The message delivered to factor node l, the specific message delivery process of the nth iteration is:

[0071]

[0072]

[0073]

[0074]

[0075] in Represents α i→A n Ranked Q from largest to smallest m Elements, Q m For set C a One-half the number of elements;

[0076] Node μ i The confidence level is:

[0077]

[0078] The update rule for node values ​​is:

[0079]

[0080] After the node value is updated, the set C a and U a Update and stop iteration when the node values ​​converge to obtain the calculation results.

[0081] Furthermore, the specific process of allocating the remaining resources of the base station in S33 is as follows:

[0082] First calculate the remaining resources a of the base station re :

[0083]

[0084] Execute a reThe greedy strategy calculates each content link allocation ratio each time it is executed. The increase in the number of users N served can be increased by Δa. f :

[0085]

[0086] In all N f Take the maximum value, the corresponding a f Updated to:

[0087] a f =a f +Δa

[0088] This process is repeated until the remaining resources of the base station are allocated.

[0089] Compared with the prior art, the present invention has the following beneficial effects:

[0090] The edge network resource deployment method proposed in the present invention is suitable for dynamic propagation environments. It takes into account the dynamic changes of content popularity during the propagation process, establishes a content propagation model, analyzes the mutual influence of network resource deployment and content popularity, constructs a dynamic relationship between edge network resource deployment and content popularity, designs a base station link resource allocation scheme, obtains an optimized edge network resource deployment strategy, and improves the service performance and resource utilization efficiency of the edge network. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0092] Figure 1 A diagram of an edge network content propagation model based on SEIR propagation dynamics provided by an embodiment of the present invention.

[0093] Figure 2 Flowchart of the parameter acquisition process provided by an embodiment of the present invention.

[0094] Figure 3 Factor graph model provided by the embodiment of the present invention

[0095] Figure 4-Figure 7 The following is a comparison of the demand satisfaction rate changes over time provided by the embodiments of the present invention.

[0096] Figures 8-11 This is a comparison of the cache hit rate changes over time provided by the embodiments of the present invention. DETAILED DESCRIPTION

[0097] Most current caching strategies assume that content popularity follows a static distribution, ignoring the dynamic characteristics of content popularity. In fact, content popularity will change dynamically with the propagation process, and network resource deployment will affect the content propagation process, so edge network resource deployment will have an impact on content popularity. At the same time, the popularity of content will directly affect the network resource deployment plan. Therefore, dynamically analyzing the relationship between edge network deployment plans and user needs (content popularity) can improve the accuracy of edge network resource deployment and improve the efficiency of edge network resource utilization. In response to the above problems, the present invention first proposes an edge network content propagation model based on propagation dynamics, aiming to establish a dynamic relationship between content popularity and network resource deployment, describe the dynamic changes in content popularity, and on this basis proposes an applicable edge network resource deployment strategy. This strategy designs a cache resource deployment plan and a link allocation plan for edge base stations based on dynamic content requests.

[0098] In order to better understand the present technical solution, the method of the present invention is described in detail below with reference to the accompanying drawings.

[0099] The edge network resource deployment method applicable to a dynamic propagation environment provided by the present invention comprises the following steps:

[0100] S1. Users within the base station coverage are divided into four states: susceptible, contact, infected, and recovered. An edge network content propagation model based on SEIR propagation dynamics is constructed.

[0101] Specifically, the present invention considers an edge network base station covering N users, the number of contents in the network is F, and the size of each content is c f bits. The amount of content that the base station can cache is C. The cache deployment strategy of the base station is P, P = {p1, p2, ..., p F}, p f =1 means that the content f is cached by the base station, p f =0 means that the base station does not cache content f. The link allocation strategy of the base station is A, A={a1,a2,...,a F}, a f Represents the ratio of link resources allocated by the base station to content f.

[0102] The present invention uses the SEIR propagation model to describe the edge network content propagation process. Figure 1 As shown, analogous to the epidemic model, users within the base station coverage area can be divided into the following four states:

[0103] Susceptible state S (Susceptible): The user has not been exposed to information about the content f and is a user who may make a request. f(P,A,t) represents the number of susceptible users at time t.

[0104] Contact state E (Exposed): The user has been influenced by the user in the infected state and has become interested in the content f. At the same time, it sends a request and waits to obtain the corresponding content. f (P,A,t) represents the number of contact users at time t.

[0105] Influenced state I: The user has obtained content f and is still interested in it. Users in this state have the ability to influence other users through social network recommendations and provide content f to surrounding users through D2D. f (P,A,t) represents the number of infected users at time t.

[0106] Recovered state R (Recovered): After obtaining content f, the user loses interest in the content. Users in this state cannot influence other users, nor will they be influenced by other users. f (P,A,t) represents the number of users in the recovery state at time t.

[0107] By analogy between this transfer process and the infectious process of an epidemic, we construct a SEIR content propagation model and reasonably assume that the number of people in the entire network remains stable, that is,

[0108] S f (P,A,t)+E f (P,A,t)+I f (P,A,t)+R f (P, A, t) = N. The SEIR state transition equation of content f can be constructed as:

[0109]

[0110] Among them, β f μ represents the probability that a susceptible user will be attracted and interested in content f when contacting an infected user of type I. f The probability that a user loses interest in the content after acquiring it and switches from the infected state to the recovered state, D f Indicates the number of users who received the requested content per unit time.

[0111] β f and μ f It is related to the nature of the content itself. The base station can collect historical data and use the least square method to obtain it. It is reasonable to assume that the edge base station can collect the user's request status information, such as Figure 2 As shown, before each resource deployment update, the request status information of each content is collected and recorded as The request status information consists of three parts: the number of users who have not requested the content, that is, the number of susceptible users; the number of users who are making requests, that is, the number of exposed users; and the number of users who have made requests and received the corresponding content, that is, the sum of the number of infected users and recovered users. Therefore, the state information i is expressed as

[0112] Establish the following system of equations:

[0113] Qe=B

[0114] Using the least squares method to obtain β f and μ f Approximate value of :

[0115] e=(Q T Q) -1 Q T B

[0116] in,

[0117] e=(β f ,μ f )

[0118]

[0119]

[0120]

[0121]

[0122]

[0123] The number of users who receive the requested content per unit time D f Expressed as:

[0124]

[0125] in, represents the number of users of D2D communication service per unit time, Indicates the number of users served by the base station per unit time, expressed as:

[0126]

[0127] Among them, R d It represents the total downlink rate of the base station, and υ represents the backhaul link rate of the base station.

[0128] S2. The base station collects request status information for each content and builds a dynamic relationship between edge network resource deployment and content popularity by analyzing the mutual influence of network resource deployment and content popularity.

[0129] Specifically, according to the constructed propagation dynamics model, the initial number of users in each state is recorded as [S f (0),E f (0),I f (0),R f (0)], the state transition equation of content f is solved by the constant variation method, and the result is as follows:

[0130]

[0131] In the actual content dissemination process, the dissemination dynamics model will evolve. f (P, A, t) = 0 and When , content propagation will evolve into the SIR model, and the state transition equation is as follows:

[0132]

[0133] E f (P, A, t) = 0 and This moment is recorded as T0 f ;

[0134] Applying the homotopy analysis method to solve the explicit solution of the SIR model, the results are as follows:

[0135]

[0136] Where M is the order, θ = μ f -β f S f (P,A,∞),S f (P, A, ∞) is calculated using the following expression:

[0137]

[0138] Among them, δ m,k , γ m,k is the coefficient, which is obtained by the following expression:

[0139]

[0140]

[0141] The calculation of relevant parameters is as follows:

[0142]

[0143]

[0144]

[0145]

[0146]

[0147] a m,j =χ 3m-j+1 (μ f -β f S f (P,A,∞)-jθ)δ m-1,j -β f χ j ε m,j ,1≤j≤3m+1

[0148] b m,j =χ 3m-j+1 (β f S f (P,A,∞)δ m-1,j -jθγ m-1,j )+β f χ j ε m,j ,1≤j≤3m+1

[0149]

[0150]

[0151] h is the convergence control parameter, When I w When the curve of h reaches a horizontal value, S is obtained by taking the value of h. f (P,A,t),I f (P,A,t) converges to an explicit solution.

[0152] S3. Based on the constructed propagation dynamics model, in a dynamic propagation environment, with maximizing the user demand satisfaction rate as the optimization goal, a base station link resource allocation scheme is designed to obtain an optimized edge network resource deployment strategy; the user demand satisfaction rate is defined as the proportion of users who send requests within time T who receive the requested content.

[0153] Specifically, the optimization objective is expressed as:

[0154]

[0155] in, represents the change in the number of users in the infected and recovered states within time T, that is, the number of users who received the requested content, represents the sum of the change in the number of susceptible users and the initial number of contact users within time T, that is, the number of users who issued requests within time T, where

[0156] In order to obtain the edge network resource deployment solution in a dynamic propagation environment with low complexity, the optimization objective is decoupled into two sub-problems and solved separately. The specific steps are as follows:

[0157] S31: Deploy base station cache resources P based on content requests. At the cache update time, sort the number of requests for each content at the current moment, that is, the number of contact users. The sorting result is recorded as The amount of content that the base station can cache is C, and the base station cache The content corresponding to the first C elements in The Cth element in is denoted as E f (0) th , all requests in the content set with a number greater than E f (0) th Cache the content;

[0158] S32: After the base station cache deployment strategy is solved, the optimization problem is transformed into:

[0159]

[0160] where 0≤a f ≤1, The optimization problem is transformed into:

[0161] argmaxη(A)·g(A)

[0162] in

[0163]

[0164]

[0165] Can be constructed as Figure 3 The factor graph model shown in the figure is used to describe the above optimization problem. The present invention designs a message passing process based on belief propagation, discretizes the link allocation ratio with a step size of Δa, and obtains the link resource allocation design scheme through iterative calculation. The allocation ratio of each content must be controlled within a certain range. During the iteration process, nodes whose values ​​reach the upper and lower limits will no longer participate in the next iteration. In each round of iteration, the set of nodes participating in the iteration is C a , the set of nodes that do not participate in the iteration is U a After each round of iteration, the two sets are updated; the lower limit of the node is taken as Δa, recorded as When T=T0 f When a f The value of is recorded as the upper limit of the node

[0166] The present invention uses Indicates that the factor node l is passed to the variable node μ during the nth iteration i The message, in turn, uses Represents the variable node μ during the t-th iteration i The message passed to factor node l. Taking the nth iteration as an example, the specific message passing process of each iteration is as follows:

[0167]

[0168]

[0169]

[0170]

[0171] in Represents α i→A n Ranked Q from largest to smallest m Elements, Q m For set C a One-half the number of elements;

[0172] Node μ i The confidence level is:

[0173]

[0174] The update rule for node values ​​is:

[0175]

[0176] After the node value is updated, the set C a and U a Update and stop iteration when the node values ​​converge to obtain the calculation results.

[0177] S33: After completing the above two steps, there may be remaining base station link resources. The present invention allocates the remaining base station resources based on a greedy algorithm. The specific process is as follows:

[0178] First calculate the remaining resources a of the base station re :

[0179]

[0180] Execute a re The greedy strategy calculates each content link allocation ratio each time it is executed. The increase in the number of users N served can be increased by Δa. f :

[0181]

[0182] In all N f Take the maximum value, the corresponding a f Updated to:

[0183] a f =a f +Δa

[0184] This process is repeated until the remaining resources of the base station are allocated.

[0185] Simulation Verification

[0186] 1. Introduction to simulation scenarios

[0187] In the simulation environment, the number of users within the coverage area of ​​the edge network base station is 1000. The size of each content is 100MB. The proposed solution is simulated under four conditions with different base station cache space and content quantity, which are:

[0188] C=10,F=50;C=20,F=50;C=10,F=100;C=20,F=100. The cache update period of the base station is 30 minutes, and the total downlink rate of the base station is R d The proposed Propagation Dynamics Based Caching Resource Strategy (PBCR) is compared with the popular caching strategy (MPC), the least frequently used caching strategy (LFU), and the mean probability caching strategy (MC).

[0189] 2 Simulation results analysis

[0190] The first is the analysis of user demand satisfaction rate. Figure 4-Figure 7 They respectively represent the changes over time in the user demand satisfaction rate of the PBCR scheme proposed in the present invention and the three comparative schemes in the prior art (i.e., MPC, LFU, and MC) under the four conditions of different base station cache spaces and content quantities mentioned in the simulation scenario introduction. Figure 4-Figure 7 These correspond to four scenarios: C=10, F=50, C=20, F=50, C=10, F=100, and C=20, F=1. Figure 4-Figure 7As shown, the PBCR strategy of the present invention has a higher demand satisfaction rate than other algorithms. The smaller the content quantity F and the larger the base station cache capacity C, the more significant the improvement in demand satisfaction. This is because more network resources and stronger base station service capabilities improve user demand satisfaction. Because the PBCR strategy considers the dynamic nature of content dissemination and allocates resources based on the predicted number of content requests, the performance improvement is even more significant.

[0191] The second is the analysis of cache hit rate. Figures 8-11 The figures respectively represent the changes in cache hit rate over time of the PBCR scheme proposed in the present invention and the three comparison schemes (i.e., MPC, LFU, and MC) under the four conditions of different base station cache spaces and content amounts mentioned in the simulation scenario introduction. Figures 8-11 These correspond to four scenarios: C=10, F=50, C=20, F=50, C=10, F=100, and C=20, F=1. Figures 8-11 As shown, the cache hit rate of the PBCR strategy of the present invention is also higher than that of other algorithms. While popularity-based and frequency-based cache strategies primarily rely on requests during the previous cache update cycle, the PBCR strategy considers the dynamic nature of content dissemination and primarily deploys caches based on the number of requests in the current state. Simulation results demonstrate that the PBCR strategy of the present invention can effectively improve cache hit rates.

[0192] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for deploying edge network resources in a dynamic communication environment, characterized in that: The following steps are involved: S1. Users within the base station coverage area are divided into four states: susceptible, contact, infected, and recovered. An edge network content propagation model based on SEIR propagation dynamics is constructed. The edge network content propagation model based on SEIR propagation dynamics in step S1 is constructed as follows: Among them, S f (P,A,t) represents the number of susceptible users at time t, E f (P,A,t) represents the number of users in contact state at time t, I f (P,A,t) represents the number of infected users at time t, R f (P,A,t) represents the number of users in the recovery state at time t, S f (P,A,t)+E f (P,A,t)+I f (P,A,t)+R f (P, A, t) = N, where N is the number of users at the edge network base station, the number of contents in the network is F, and the size of each content is c f bits, the amount of content that the base station can cache is C; the cache deployment strategy of the base station is P, P = {p1,p2,...,p F }, p f =1 means that the content f is cached by the base station, p f = 0 means that the base station has no cache content f; the link allocation strategy of the base station is A, A = {a1, a2, ..., a F }, a f represents the ratio of link resources allocated by the base station to content f; β f μ represents the probability that a susceptible user will be attracted and interested in content f when contacting an infected user of type I. f The probability that a user loses interest in the content after acquiring it and switches from the infected state to the recovered state, D f Indicates the number of users who receive the requested content per unit time; D f Expressed as: in, represents the number of users of D2D communication service per unit time, Indicates the number of users served by the base station per unit time, expressed as: Among them, R d represents the total downlink rate of the base station, and υ represents the backhaul link rate of the base station; S2. The base station collects request status information for each content. Request status information consists of three parts: the number of users who have not requested the content, i.e., the number of susceptible users; the number of users currently making requests, i.e., the number of exposed users; and the number of users who have made requests and received the corresponding content, i.e., the sum of infected and recovered users. By analyzing the mutual influence of network resource deployment and content popularity, a dynamic relationship between edge network resource deployment and content popularity is constructed. The request status information in step S2 is represented as The state information i is expressed as Probability β f and μ f By collecting historical data and using the least squares method, the specific process is as follows: Establish the following system of equations: Qe=B Using the least squares method to obtain β f and μ f Approximate value of : e=(Q T Q) -1 Q T B in, e=(β f ,m f ) The process of building the dynamic relationship between edge network resource deployment and content popularity in step S2 is as follows: The initial number of users in each state is recorded as [S f (0),E f (0),I f (0),R f (0)], the state transition equation of content f is solved by the constant variation method, and the result is as follows: In the actual content dissemination process, the dissemination dynamics model will evolve. f (P, A, t) = 0 and When , content propagation will evolve into the SIR model, and the state transition equation is as follows: E f (P, A, t) = 0 and This moment is recorded as T0 f ; Applying the homotopy analysis method to solve the explicit solution of the SIR model, the results are as follows: Where M is the order, θ = μ f -β f S f (P,A,∞),S f (P, A, ∞) is calculated using the following expression: Among them, δ m,k , γ m,k is the coefficient, which is obtained using the following expression: The calculation of relevant parameters is as follows: a m,j =x 3m-j+1 (m f -b f S f (P,A,∞)-jθ)δ m-1,j -b f x j e m,j ,1≤j≤3m+1 b m,j =x 3m-j+1 (b f S f (P,A,∞)δ m-1,j -jθγ m-1,j )+b f x j e m,j ,1≤j≤3m+1 h is the convergence control parameter, When I w When the curve of h reaches a horizontal value, S is obtained by taking the value of h. f (P,A,t),I f (P,A,t) converges to an explicit solution; S3. Based on the constructed propagation dynamics model, in a dynamic propagation environment, we design a base station link resource allocation scheme with the optimization goal of maximizing the user demand satisfaction rate, and obtain an optimized edge network resource deployment strategy. The user demand satisfaction rate is defined as the proportion of users who send requests within time T who receive the requested content. The optimization objective of step S3 is expressed as: in, represents the change in the number of users in the infected and recovered states within time T, that is, the number of users who received the requested content, represents the sum of the change in the number of susceptible users and the initial number of contact users within time T, that is, the number of users who issued requests within time T, where 2. The edge network resource deployment method applicable to a dynamic communication environment according to claim 1, characterized in that: Decouple the optimization objective in step S3 into two sub-problems and solve them separately. The specific steps are as follows: S31: Deploy base station cache resources P based on content requests. At the cache update time, sort the number of requests for each content at the current moment, that is, the number of contact users. The sorting result is recorded as The amount of content that the base station can cache is C, and the base station cache The content corresponding to the first C elements in The Cth element in is denoted as E f (0) th , all requests in the content set with a number greater than E f (0) th Cache the content; S32: After the base station cache deployment strategy is solved, the optimization problem is transformed into: where 0≤a f ≤1, The optimization problem is transformed into: argmaxη(A)·g(A) in Based on the message passing process of belief propagation, the link allocation ratio is discretized with a step size of Δa, and the link resource allocation design scheme is obtained through iterative calculation; S33: Allocate the remaining resources of the base station based on a greedy algorithm.

3. The edge network resource deployment method applicable to a dynamic communication environment according to claim 2, characterized in that: During the iteration of step S32, the nodes whose values ​​reach the upper and lower limits will no longer participate in the next iteration. In each round of iteration, the set of nodes participating in the iteration is C a , the set of nodes that do not participate in the iteration is U a After each round of iteration, the two sets are updated; the lower limit of the node is taken as Δa, recorded as When T=T0 f When a f The value of is recorded as the upper limit of the node use Indicates that the factor node l is passed to the variable node μ during the nth iteration i The message, in turn, uses Represents the variable node μ during the t-th iteration i The message delivered to factor node l, the specific message delivery process of the nth iteration is: in Represents α i→A n Ranked Q from largest to smallest m Elements, Q m For set C a One-half the number of elements; Node μ i The confidence level is: The update rule for node values ​​is: After the node value is updated, the set C a and U a Update and stop iteration when the node values ​​converge to obtain the calculation results.

4. The edge network resource deployment method applicable to a dynamic communication environment according to claim 3, characterized in that: The specific process of S33 allocating the remaining resources of the base station is as follows: First calculate the remaining resources a of the base station re : Execute a re The greedy strategy calculates each content link allocation ratio each time it is executed. The increase in the number of users N served can be increased by Δa. f : In all N f Take the maximum value, the corresponding a f Updated to: in f = yes f +Δa This process is repeated until the remaining resources of the base station are allocated.

Images