Distributed collaborative caching method oriented to unreliable edge environment in novel power system

Abstract

The invention belongs to the technical field of power wireless communication, and particularly relates to a distributed collaborative caching method oriented to an unreliable edge environment in a novel power system. The method comprises the following steps: constructing a distributed collaborative edge cache system model; constructing a maximized cache hit rate problem according to the distributed collaborative edge cache system model; solving the problem of maximizing the cache hit rate to obtain the number of file blocks of the file cache on different types of edge servers; mapping according to the number of cache file blocks of the edge server to obtain an optimal cache scheme of each file; according to the invention, the utilization rate of limited edge resources can be improved, and edge caching service with high cost performance can be provided.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric power wireless communication, and in particular relates to a distributed collaborative caching method for an unreliable edge environment in a new type of electric power system. Background Art

[0002] With the continuous advancement of the construction of new power systems, the number of digital and intelligent devices in power systems has increased rapidly, posing severe challenges to the access capacity, flexibility and reliability of power communication networks. At present, in order to cope with the problem of power communication network congestion caused by massive heterogeneous terminals accessing the power system, caching popular files at the edge of the network has become one of the common examples to solve network congestion, reduce the number of requests to content providers during peak hours, and reduce service delays to support real-time applications. A large amount of structured data, semi-structured data, and unstructured smart distribution data generated by the operation of the power system are received by the power grid data management center and stored in the central cloud. Users access distribution data to the central cloud through various smart terminal devices and edge clouds. When the user access volume is too high, it will bring huge access pressure to the central cloud. Therefore, the central cloud can store various smart distribution data in the edge server in advance to facilitate user access. However, this reliable edge resource is highly limited and expensive.

[0003] Therefore, popular files can be cached on unreliable edge resources to improve cost efficiency. Two types of unreliable resources are considered here. The first is idle resources, or resources that are reserved for other applications but are not fully utilized, which is unreliable because the edge server may erase the cached content to serve higher priority applications. Cached files on unused resources of the edge server can be erased when needed to minimize side effects. In addition, caching on resources that are not used by the edge server is much cheaper than caching on reserved cache resources. The second type of unreliable resources are edge servers owned by various organizations, which may fail or leave the system for various reasons, and the cost of using this type of edge servers is also lower. Using unreliable edge servers to cache popular files in the power grid can not only improve the utilization of limited edge resources, but also provide cost-effective edge caching services. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention proposes a distributed collaborative caching method for an unreliable edge environment in a new power system, the method comprising:

[0005] S1: Building a distributed collaborative edge cache system model;

[0006] S2: Construct the problem of maximizing cache hit rate based on the distributed collaborative edge cache system model;

[0007] S3: Solve the problem of maximizing cache hit rate and obtain the number of file blocks cached on different types of edge servers;

[0008] S4: Obtain the optimal caching solution for each file block according to the number of file blocks cached by the edge server.

[0009] Preferably, the distributed collaborative edge cache system model includes: a plurality of N available edge servers of different types, where the capacity and reliability parameters of servers of the same type are the same; users requesting content at peak hours with an average request rate h per time slot i The file i is encoded into no less than K file blocks. The file i can be reconstructed from any K file blocks. The server of type v can cache at most σ v file blocks, the reliability parameter of the server of type v is q v , the number of file blocks cached by file i on servers of type v is y i,v ∈{0,1,2....N v},N v Indicates the number of servers of type v.

[0010] Preferably, the process of constructing the problem of maximizing cache hit rate includes:

[0011] Get the server's cache availability probability f n,t and the set of edge servers φ that cache the file chunks of file i i (x i ),use represents the set consisting of edge server subsets whose number of cached file blocks of file i is not less than K. For each Calculate the probability P that the cache on M is available and the cache on other edge servers is unavailable i,M,t ;

[0012] According to P i,M,t Calculate the probability that file i is reconstructed;

[0013] Calculate the time-accumulated availability of file i according to the probability that file i is reconstructed;

[0014] The problem of maximizing cache hit rate is constructed based on the time-accumulated availability of file i and the average request rate of users in each time slot during peak hours.

[0015] Furthermore, we calculate the probability P i,M,t The formula is:

[0016]

[0017] Among them, f n,trepresents the cache availability probability of edge server n at time slot t, φ i (x i ) represents the set of edge servers that cache the file blocks of file i.

[0018] Furthermore, the formula for calculating the probability of file i being reconstructed is:

[0019]

[0020] Among them, a i,t represents the probability that file i is reconstructed at time slot t.

[0021] Furthermore, the formula for calculating the time-accumulated availability of file i is:

[0022]

[0023] Among them, a i represents the time-accumulated availability of file i, a i,t represents the probability that file i is reconstructed at time slot t, and T represents the total number of time slots.

[0024] Furthermore, the problem of maximizing cache hit rate is expressed as:

[0025]

[0026] Among them, h i represents the average request rate of users requesting file i during peak hours, a i represents the time-accumulated availability of file i, I represents the file set, x i,n Indicates whether file i is cached on server n, s n represents the capacity of server n, and N represents the number of edge servers.

[0027] Preferably, the process of obtaining the optimal cache solution for each file block according to the mapping of the number of file blocks cached by the edge server includes:

[0028] definition:

[0029]

[0030] If S i,v <E i,v , then for Set x i,n = 1, x of other servers of type v i,n =0;

[0031] If S i,v >E i,v , then for Set xi,n = 1, x of other servers of type v i,n =0;

[0032] If S i,v =E i,v , then for all servers of type v, set x i,n =1;

[0033] If S i,v =E i,v And y i,v = 0, then for all servers of type v, set x i,n =0;

[0034] Among them, S i,v The remainder of the number of edge servers of type v that have cached the first i-1 files divided by the total number of edge servers of type v, y j,v N represents the number of file blocks of file j cached on edge servers of type v, v represents the number of edge servers of type v, E i,v represents the remainder of the number of edge servers of type v that have cached the first i files divided by the total number of edge servers of type v. Indicates the sth i,v +1 edge server, Indicates the Eth of type v i,v Edge servers, x i,n Indicates whether file i is cached on edge server n.

[0035] The beneficial effects of the present invention are as follows: the present invention selects unreliable edge server resources for edge caching, thereby improving the overall resource utilization rate. By performing distributed collaborative caching during off-peak hours, the network congestion problem of the smart grid during peak hours can be solved, and the overall service delay can be reduced. The present invention can not only improve the utilization rate of limited edge resources, but also provide cost-effective edge caching services. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of the distributed collaborative caching method in the unreliable edge environment of the new power system in the present invention;

[0037] Figure 2 This is an example diagram of the DEC system cache file in the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] The present invention proposes a distributed collaborative caching method for unreliable edge environments in new power systems. Figure 1 As shown, the method includes:

[0040] S1: Construct a distributed collaborative edge cache system model.

[0041] In a distributed collaborative edge cache system (DEC) with unreliable cache resources, there are multiple N available edge servers of different types, and the capacity and reliability parameters of servers of the same type are the same; popular files are downloaded and encoded into multiple encoded packages, each of which is referred to as a file block in the following, and each file block is cached on the edge server during non-peak hours to reduce the delay of retrieving files, the congestion of the core network, and the load of the remote server during peak hours. Assuming that the peak hours of a day can be divided into T time slots of equal length, i.e., T∈[T], {1,2,3...T}, since the edge servers are located near the users, the cost of downloading data from the edge servers in the system is much lower than the cost of downloading data from the remote content provider through the core network in terms of delay and throughput.

[0042] The set of popular files is represented by I∈[I], {1, 2, 3...I}. Users who request content from the DEC system will request content at an average request rate h per time slot during peak hours. i Request file i for each time, assuming that all files in I are of the same size, such as Q bits. Each file is divided into K data packets of the same length and encoded into more than K blocks using an MDS code (such as an RS code) (some bits of data in the data packet are deleted or replaced, and these replaced and deleted bits are re-encoded into blocks), and the length of each block is Q / K. Using the MDS code, the original file i can be regenerated from any of its K blocks, where K is a pre-specified and known parameter. K is fixed in order to facilitate the adjustment of the bit rate of different files, that is, the number of file blocks. The blocks of the file are stored on different edge servers.

[0043] The set of edge servers is represented by N = {1, 2..., N}. Assume that N>K, and the size of server n is S n ·M / K position, equivalent to S n That is, server n can cache at most S n file blocks.

[0044] Considering that there are many types of edge servers with unreliable resources, we assume that there are V types of edge servers, and the same type of server has the same capacity s n and reliability parameter p n , where V is much smaller than N. Assume N v represents the number of servers of type v∈V. A server of type v can cache at most σ v file blocks, the reliability parameter of the server of type v is q v , the number of file blocks cached by file i on servers of type v is y i,v ∈{0,1,2....N v},N v Indicates the number of servers of type v. Use Y i To represent the set {y i,v} v∈[V] ,and

[0045] S2: Construct the problem of maximizing cache hit rate based on the distributed collaborative edge cache system model.

[0046] If server n wants to contribute its cache capacity to DEC for applications with higher priority, or fails for some reason, it will directly erase the cached blocks or leave the system. If, for various reasons, the user cannot download the data block from the edge server within a time below a given threshold, it is also considered an edge server failure. For server n∈N, get the server's cache availability probability f n,t , that is, for each block cached on server n, the probability of being available in time slot t is f n,t In particular, for any server n∈N, the availability of its cache is modeled as a geometric distribution, where f n,t =(p n ) t ,p n is the reliability parameter of server n, which is equal to the probability that the cache on server n is available at time slot t.

[0047] For example, Figure 2As shown, there are 5 files of the same length, marked with different colors. Each data packet is divided into K=3 data packets, encoded into no less than 3 blocks, and cached on edge servers 1, 2, 3, and 4. A file can be regenerated from any 3 of its encoded blocks. Server 1 fails at time slot t1. The user can regenerate any of the five files by downloading the encoded blocks from other edge servers before the failure of server 1. When time slot t2>t1, the user can regenerate the red, yellow, gold, and orange files by encoding blocks on the available edge servers 2, 3, and 4. The purple file cannot be restored by encoding blocks on the available edge servers, and the purple file can only be obtained through the content source.

[0048] Get the set of edge servers φ that cache the file chunks of file i i (x i ),use represents the set consisting of the subset of edge servers whose number of cached file blocks of file i is not less than K, that is, For each Calculate the probability P that the cache on M is available and the cache on other edge servers is unavailable i,M,t :

[0049]

[0050] Among them, f n,t represents the cache availability probability of edge server n at time slot t, φ i (x i ) represents the set of edge servers that cache the file blocks of file i.

[0051] According to P i,M,t Calculate the probability that file i is reconstructed, that is, the availability of file i in the time slot t∈T i,t :

[0052]

[0053] Calculate the time-accumulated availability a of file i according to the probability that file i is reconstructed i :

[0054]

[0055] The problem of maximizing cache hit rate is constructed based on the time-accumulated availability of file i and the average request rate of users in each time slot during peak hours. Specifically:

[0056] If the cached encoding blocks on the edge server are less than k, the user must download the data through the congested core network, which is very time-consuming. Therefore, the goal of the system is to maximize the cache hit rate, that is, the expected size of the file downloaded from the edge server during peak hours is as follows:

[0057]

[0058] Constraint: The number of file blocks cached on server n is not greater than the size of the server, that is,

[0059] Among them, x i,,n ∈{0,1} is a binary variable indicating whether a file block of file i is placed on edge server n. represents a collection of variables associated with file i, represents the set of all decision variables. x contains the number and location information of blocks of each file. In the present invention, the blocks in each server are independent, that is, some blocks in a server may fail (be erased), while other blocks in the same server may be available.

[0060] Therefore, the problem of maximizing cache hit rate is expressed as:

[0061]

[0062] S3: Solve the problem of maximizing the cache hit rate and obtain the number of file blocks cached on different types of edge servers.

[0063] For a server of type v, there is p n =q v , f n,t =(p n ) t =(q v ) t ; For a given y i , the expected availability of file i in time slot t is expressed as a i,t =A t (y i ), if y i <K,a i,t = 0. If y i >K, then Among them, f v,t represents the cache availability probability of type v server in time slot t, when y i,v <j v When it represents the total number of cache blocks of type v server, j=(j1,j2......j V ), j v Indicates the total number of cached file blocks of servers of type v.

[0064] Therefore, the objective function of maximizing the cache hit rate problem can be written as:

[0065]

[0066] Then the problem of maximizing cache hit rate can be rewritten as:

[0067]

[0068] Solve the above problem:

[0069] When there are only the first j files, and the total capacity of the edge servers of type v∈[V] is z v In the case of file blocks, w(z1,z2...z n ,j) represents the current optimal target value. In addition, r(z1,z2...z n ,j) The total capacity of the edge server with only the first j files and type v is z v For convenience of representation, the present invention uses w(z,j) and r(z,j) as w(z1,z2...z n ,j) and r(z1,z2...z n ,j) is an abbreviation of

[0070] Next, design the initialization and dynamicization of w(z,j) and r(z,j):

[0071] First, when j = 1,

[0072]

[0073] For j∈[2,3...I], z v ={0,1...σ v N v}, v∈V, w(z,j) and r(z,j) are calculated as follows:

[0074]

[0075] First, initialize w(z,1) and r(z,1) through the formula;

[0076] Then, w(z,j) and r(z,j) are iteratively updated using the above dynamic update formula to obtain the local optimal choice;

[0077] Finally, by backtracking the local optimal decision recorded in r(z,j), the optimal placement solution Y is obtained. i =r(z,i), which is the number of file blocks cached on edge servers of different types.

[0078] S4: Obtain the optimal caching solution for each file block according to the number of file blocks cached by the edge server.

[0079] A mapping scheme is designed here to map the optimal solution of the placement scheme to the optimal solution of the original problem, that is, to obtain the optimal cache solution for each file. Specifically:

[0080] For i∈[I],v∈V, define:

[0081]

[0082] If S i,v <E i,v , then for Set x i,n = 1, x of other servers of type v i,n =0;

[0083] If S i,v >E i,v , then for Set x i,n = 1, x of other servers of type v i,n =0;

[0084] If S i,v =E i,v , then for all servers of type v, set x i,n =1;

[0085] If S i,v =E i,v And y i,v = 0, then for all servers of type v, set x i,n =0;

[0086] Among them, S i,v The remainder of the number of edge servers of type v that have cached the first i-1 files divided by the total number of edge servers of type v, y j,v N represents the number of file blocks of file j cached on edge servers of type v, v represents the number of edge servers of type v, E i,v represents the remainder of the number of edge servers of type v that have cached the first i files divided by the total number of edge servers of type v. Indicates the sth i,v +1 edge server, Indicates the Eth of type v i,v Edge servers, x i,n Indicates whether file i is cached on edge server n.

[0087] After obtaining the optimal caching solution for each file, the system executes the optimal caching solution, which can not only improve the utilization rate of limited edge resources, but also provide cost-effective edge caching services.

[0088] The above embodiments further illustrate the purpose, technical solutions and advantages of the present invention in detail. It should be understood that the above embodiments are only preferred implementation modes of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A distributed collaborative caching method for unreliable edge environments in new power systems, characterized in that: include: S1: Building a distributed collaborative edge cache system model; S2: Construct the problem of maximizing cache hit rate based on the distributed collaborative edge cache system model; S3: Solve the problem of maximizing cache hit rate and obtain the number of file blocks cached on different types of edge servers; S4: Obtain the optimal caching solution for each file based on the number of file blocks cached by the edge server.

2. According to claim 1, a distributed collaborative caching method for unreliable edge environments in new power systems is characterized in that: The distributed collaborative edge cache system model includes: multiple N available edge servers of different types, with the same capacity and reliability parameters for servers of the same type; users requesting content at peak times with an average request rate h per time slot. i The file i is encoded into no less than K file blocks. The file i can be reconstructed from any K file blocks. The server of type v can cache at most σ v file blocks, the reliability parameter of the server of type v is q v , the number of file blocks cached by file i on servers of type v is y i,v ∈{0,1,2....N v },N v Indicates the number of servers of type v.

3. According to claim 1, a distributed collaborative caching method for unreliable edge environments in new power systems is characterized in that: The process of constructing the problem of maximizing cache hit rate involves: Get the server's cache availability probability f n,t and the set of edge servers φ that cache the file chunks of file i i (x i ), using ω i (x i ) represents the set consisting of edge server subsets whose number of cached file blocks of file i is not less than K. For each Calculate the probability P that the cache on M is available and the cache on other edge servers is unavailable i,M,t ; According to P i,M,t Calculate the probability that file i is reconstructed; Calculate the time-accumulated availability of file i according to the probability that file i is reconstructed; The problem of maximizing cache hit rate is constructed based on the time-accumulated availability of file i and the average request rate of users in each time slot during peak hours.

4. A distributed collaborative caching method for unreliable edge environments in a new power system according to claim 3, characterized in that: Calculate the probability P i,M,t The formula is: Among them, f n,t represents the cache availability probability of edge server n at time slot t, φ i (x i ) represents the set of edge servers that cache the file blocks of file i.

5. A distributed collaborative caching method for unreliable edge environments in a new power system according to claim 3, characterized in that: The formula for calculating the probability that file i is reconstructed is: Among them, a i,t represents the probability that file i is reconstructed at time slot t.

6. A distributed collaborative caching method for unreliable edge environments in a new power system according to claim 3, characterized in that: The formula for calculating the time-accumulated availability of file i is: Among them, a i represents the time-accumulated availability of file i, a i,t represents the probability that file i is reconstructed at time slot t, and T represents the total number of time slots.

7. A distributed collaborative caching method for unreliable edge environments in a new power system according to claim 3, characterized in that: The problem of maximizing cache hit rate is expressed as: Among them, h i represents the average request rate of users requesting file i during peak hours, a i represents the time-accumulated availability of file i, I represents the file set, x i,n Indicates whether file i is cached on server n, s n represents the capacity of server n, and N represents the number of edge servers.

8. The distributed collaborative caching method for unreliable edge environments in a new power system according to claim 1, characterized in that: The process of mapping the number of file blocks cached by the edge server to obtain the optimal cache solution for each file block includes: definition: If S i,v <E i,v , then for Set x i,n = 1, x of other servers of type v i,n =0; If S i,v >E i,v , then for Set x i,n = 1, x of other servers of type v i,n =0; If S i,v =E i,v , then for all servers of type v, set x i,n =1; If S i,v =E i,v And y i,v = 0, then for all servers of type v, set x i,n =0; Among them, S i,v The remainder of the number of edge servers of type v that have cached the first i-1 files divided by the total number of edge servers of type v, y j,v N represents the number of file blocks of file j cached on edge servers of type v, v Indicates the number of edge servers of type v, E i,v represents the remainder of the number of edge servers of type v that have cached the first i files divided by the total number of edge servers of type v. Indicates the Sth i,v +1 edge server, Indicates the Eth of type v i,v Edge servers, x i,n Indicates whether file i is cached on edge server n.

Images