Multifunctional task offloading method for heterogeneous cellular networks based on VNF instance sharing

Abstract

The present invention claims protection for a heterogeneous cellular network multifunctional task offloading method based on VNF instance sharing, which belongs to the field of communication technology. In view of the problem that the uneven distribution of users in the edge scenario of heterogeneous cellular networks may cause load imbalance of base station servers, thereby leading to a decrease in resource utilization and an increase in task offloading delay, a heterogeneous cellular network multifunctional task offloading method based on VNF instance sharing is proposed. According to the user service distribution and the SFC type required for task offloading, the method adopts the particle swarm optimization idea to establish a heterogeneous cellular network SFC deployment optimization model, and through the exploration and update of the VNF instantiation location and number, it iteratively optimizes the SFC mapping and resource allocation decision, thereby improving the VNF instance sharing degree, effectively improving the task offloading completion rate, and reducing the system unit task delay.

Description

Technical Field

[0001] The present invention belongs to the field of communication technology, and in particular relates to a heterogeneous cellular network multifunctional task offloading method based on VNF instance sharing. Background Art

[0002] With the rapid development of mobile communication networks, a large number of computationally intensive and latency-sensitive task offloading applications have emerged. These applications are typically composed of one or more interdependent functional components, which complicates the management and scheduling of network resources. By introducing Network Function Virtualization (NFV) technology into heterogeneous cellular networks, network functions are abstracted into Virtual Network Functions (VNFs), and VNFs are orchestrated into Service Function Chaining (SFCs), providing effective network function support for task offloading.

[0003] Existing research on task offloading in heterogeneous cellular networks mostly focuses on single-function and single-base station task offloading, but fails to consider the offloading of multifunctional tasks in heterogeneous cellular network scenarios based on SFCs. Existing research on SFC deployment typically employs a fixed resource allocation approach when mapping VNFs, failing to consider the coordinated resource allocation of each VNF in the SFC based on the service load of the mapped node to improve resource utilization and meet service needs. To address the issue of uneven user distribution at the edge of heterogeneous cellular networks, which can cause load imbalances in base station servers, leading to decreased resource utilization and increased task offloading latency, the present invention proposes a method for multifunctional task offloading in heterogeneous cellular networks based on VNF instance sharing. This method maximizes network resource utilization, effectively improving the task offloading completion rate, and reducing the system unit task latency. Summary of the Invention

[0004] The present invention aims to solve the above problems in the prior art. It proposes a method for multifunctional task offloading in heterogeneous cellular networks based on VNF instance sharing. The technical solution of the present invention is as follows:

[0005] A method for offloading multifunctional tasks in heterogeneous cellular networks based on VNF instance sharing comprises the following steps:

[0006] 101. Use particle swarm algorithm to establish heterogeneous cellular network service function chain SFC deployment optimization model, assuming the number of network nodes is N, the number of particles in the particle swarm is J, and each particle is represented by F j ={f j}, where j represents the particle number, f jrepresents the number of virtual network function VNF instances of type f in the jth particle, and initializes the particle swarm, where f j Take a random integer in the interval [0,|N|] and let the optimal solution objective function value of each particle be in, They represent the optimal VNF instance deployment number decision, optimal SFC node mapping decision, optimal SFC link mapping decision, and optimal SFC resource allocation decision of the jth particle, respectively. The global optimal solution objective function value Γ{F,X,Y,Z}=0, where F,X,Y,Z represent the global optimal VNF instance deployment number decision, global optimal SFC node mapping decision, global optimal SFC link mapping decision, and global optimal SFC resource allocation decision, respectively. The number of particle swarm iterations is I. Let the counting variables i=0, j=0;

[0007] 102. Let i = i + 1. If i ≤ I, jump to step 103. Otherwise, jump to step 107.

[0008] 103. Let j = j + 1. If j ≤ J, jump to step 104. Otherwise, let j = 0 and jump to step 102.

[0009] 104. According to F j and user task offloading request, construct node mapping decision X j , link mapping decision Y j and resource allocation decision Z j , and update F j ;

[0010] 105. Calculate the objective function value Γ{F j ,X j ,Y j ,Z j},if make Jump to step 106, otherwise, jump to step 103;

[0011] 106. If make Jump to step 103, otherwise, jump to step 103;

[0012] 107. Output the global optimal solution {F, X, Y, Z} and end.

[0013] Furthermore, in step 104, the node mapping decision X is constructed. j , link mapping decision Y j and resource allocation decision Z j , and update F jThe method comprises the following steps:

[0014] 1) Let the user set be U, and the task tolerance delay T of each user u u , sort the elements in set U in ascending order, set temporary set U′=U, and count variable w=0;

[0015] 2) Let w=w+1, if w≤|F j |, let the temporary set N′=N, the counting variable m=0, remove the set N′ that satisfies Z n / a n ≤z0 condition node, where Z n represents the total computing resources of node n, a n represents the number of virtual network function (VNF) instances of node n, z0 represents the basic computing resource overhead of each VNF instance, jump to step 3), otherwise, jump to step 4);

[0016] 3) For the set F j The wth element f in j , if m<f j , let m=m+1, select from N′ the value that satisfies max(Z n / a n ) node n deploys a VNF instance of type f, a n =a n +1, and remove node n from N′, jump to step 3), otherwise, jump to step 2);

[0017] 4) If the collection Take the first user u from the set U′, and build the service function chain SFC node and link mapping decision for user u in the jth particle based on the VNF instance type and network resource status of each node n in the set N. according to and the available computing resources of the node, construct the SFC resource allocation decision of user u in the jth particle Jump to step 4), otherwise, jump to step 5);

[0018] 5) For each node n in the set N, remove the VNF instances that are not assigned task computing resources and update F j The corresponding f j ;

[0019] 6) Decision based on SFC node mapping Calculate the remaining available computing resources For the set N that satisfies Each node n, according to the order of elements in U, takes turns to add the minimum unit computing resources for each offload task carried by the node and update the resource allocation decision until

[0020] 7) Output SFC deployment decision The algorithm ends.

[0021] Furthermore, in step 4), the SFC node and link mapping decision of user u in the jth particle is constructed. The following steps are involved:

[0022] 11) Initialize the temporary set N′=N, Count variables k = 0, l = 0, determine the access node of user u according to the base station to which the user belongs Make temporary variables

[0023] 12) If k<|V u |, let k = k + 1, where V u Indicates the ordered set of VNFs in the SFC required by user u. Jump to step 13). Otherwise, jump to step 16.

[0024] 13) The set N′ contains the required instances of the kth VNF in SFC and satisfies The nodes join the set N″, where represents the minimum computational delay required for user u’s offload task on node n;

[0025] 14) For each base station n in the set N″, calculate the minimum cost path P from node τ to n τ,n , and remove nodes that do not have minimum cost paths from N″;

[0026] 15) Calculate the mapping value χ of each node n in N″ n , take out the node n with the highest mapping value from the set N″ and add it as the mapping node of the kth VNF to the SFC node mapping decision of user u Pathway P τ,n The SFC link mapping decision Y of user u is added as the lth virtual link j u , let l = l + 1, τ = n, remove node n from the set N′, and jump to step 12);

[0027] 16) Output the SFC node and link mapping decision of user u The algorithm ends.

[0028] Furthermore, the minimum computational delay required for the offloading task of user u on node n in step 13) is The calculation method is shown in formula (1):

[0029]

[0030] Among them, d u represents the amount of task data for user u, g u represents the task complexity of user u;

[0031] The minimum cost path P in step 14) τ,n The calculation method is as follows:

[0032] Calculate the minimum cost path P from node τ to n using the minimum cost path algorithm τ,n , where the link cost δ of the physical link e e The calculation method is shown in formula (2):

[0033]

[0034] in, represents the remaining available bandwidth resources of physical link e, B e represents the bandwidth resource capacity of the physical link e, b u Indicates the transmission bandwidth required for task offloading by user u;

[0035] Furthermore, in step 15), the node n mapping value x n The calculation method is shown in formula (3):

[0036]

[0037]

[0038]

[0039] in, represents the normalized remaining available computing resources of node n, α、H τ,n Represent the remaining available computing resources of node n, weight factor, path P τ,n The distance is calculated as shown in formula (4). Represents path P τ,n The normalized distance is calculated as shown in formula (5).

[0040] Furthermore, in step 4), the SFC resource allocation decision of user u in the jth particle is constructed The following steps are involved:

[0041] 21) Node mapping decision based on user u The mapping nodes of SFC are added to the temporary ordered set N′ in sequence, and the link mapping decision of user u is made. The bandwidth resource b required by the virtual linku Add SFC resource allocation decision And obtain the total transmission delay of the task offloading path in, is the cumulative transmission delay of user u’s task data through each physical link on the offloading path, and the counting variable k is set to 0;

[0042] 22) Calculate the minimum computational delay required for user u to offload the task if Jump to step 23), otherwise, jump to step 29);

[0043] 23) If k<|V u |, let k=k+1, jump to step 24), otherwise, jump to step 25);

[0044] 24) Calculate the task computing resources that the kth node n in N′ needs to allocate to the corresponding VNF in the SFC required by user u Jump to step 23);

[0045] 25) If the task calculates the total delay Jump to step 26), otherwise, jump to step 28);

[0046] 26) All available computing resources The node is removed from the set N′ if according to The value of , sort the elements in the set N ′ in descending order, and add a unit of computing resource z1 to the VNF instance required by user u on the first node n in the set N ′, let Jump to step 27), otherwise, jump to step 29);

[0047] 27) If Jump to step 26), otherwise, jump to step 28);

[0048] 28) Output the resource allocation decision of user u in the jth particle

[0049] 29) The algorithm ends.

[0050] Furthermore, the minimum computational delay required for user u to offload the task in step 22) is The calculation method is shown in formula (6):

[0051]

[0052] Among them, d u represents the amount of task data for user u, g u represents the task complexity of user u;

[0053] In step 24), node n needs to allocate task computing resources to the corresponding VNF in the SFC required by user u. The calculation method is shown in formula (7):

[0054]

[0055] in, represents the load of node n, and the calculation method is shown in formula (8);

[0056]

[0057] In formula (8), Z n Represents the total computing resources of node n.

[0058] The total delay of the task calculation in steps 25) and 27) The calculation method is shown in formula (9):

[0059]

[0060] Furthermore, in step 104, f is updated j The method is as follows:

[0061] For each node n in the set N, if there is a VNF instance of type f on node n that has not been assigned task computing resources, remove the instance from node n and set f j =f j -1; then update f according to formula (10) j ;

[0062]

[0063]

[0064] Where Δf j represents the change in the number of deployed VNF instances of type f. The calculation method is shown in formula (11), 0≤ω≤1, 0≤γ1≤1, 0≤γ2≤1, represents the number of deployed VNF instances of type f in the current optimal solution, f * represents the number of deployed VNF instances of type f in the global optimal solution.

[0065] Furthermore, the calculation method of the objective function in steps 105 and 106 is shown in formula (12):

[0066]

[0067] Among them, 0≤β≤1, Indicates the system task completion rate, The advantages and beneficial effects of the present invention are as follows:

[0068] The present invention discloses a method for offloading multifunctional tasks in heterogeneous cellular networks based on VNF instance sharing. Existing research on the problem of task offloading in heterogeneous cellular networks mostly considers the task offloading of a single function and a single base station, but does not consider the offloading of multifunctional tasks based on SFC in heterogeneous cellular network scenarios. Existing research on the problem of SFC deployment usually adopts a fixed resource allocation method when mapping VNFs, but does not consider the overall planning of resource allocation for each VNF in the SFC according to the business load of the mapping node to improve resource utilization and meet business needs. In view of the problem that uneven user distribution may cause load imbalance of base station servers in the edge scenario of heterogeneous cellular networks, thereby leading to decreased resource utilization and increased task offloading delay, the present invention adopts the particle swarm optimization idea to establish an SFC deployment optimization model for heterogeneous cellular networks, plans the location deployment of VNF instances according to the distribution of base station computing resources, builds SFC mapping decisions according to the SFC type required for user task offloading and the network resource status, and builds SFC resource allocation decisions according to the user task offloading requirements and base station resource load. Through the iterative optimization of VNF instance deployment locations, SFC mapping and resource allocation decisions, the sharing degree of VNF instances is effectively improved, thereby maximizing the task offloading completion rate and reducing the system unit task delay. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 The present invention provides a flow chart of a method for unloading tasks in a heterogeneous cellular network based on load balancing in a preferred embodiment. DETAILED DESCRIPTION

[0070] The following will describe the technical solutions in the embodiments of the present invention in detail with reference to the accompanying drawings. The described embodiments are only a part of the embodiments of the present invention.

[0071] The technical solution of the present invention to solve the above technical problems is:

[0072] The concepts and models involved in the content of this invention are as follows.

[0073] 1. Network Model

[0074] Assume that a heterogeneous cellular network consists of several macro and micro base stations. Each macro and micro base station is equipped with a server with sufficient computing power, allowing the deployment of any type of VNF instances. Macro base stations can communicate directly via wired links or through relay forwarding by the macro base station. Each macro base station manages multiple micro base stations. A macro base station communicates directly with its micro base stations via wired links, while micro base stations belonging to the same macro base station forward communications via their respective macro base stations. Users access the heterogeneous cellular network by establishing wireless links with access base stations. The required SFCs are orchestrated and constructed based on user offload requirements. Different VNFs within the same SFC must be mapped to different base stations, and VNF instances of the same type can be shared across different SFCs.

[0075] 2. Other symbols involved in the present invention are explained as follows.

[0076] u: terminal user u

[0077] N: base station set

[0078] n: base station n

[0079] E: Physical link set

[0080] e: physical link e

[0081] d u : The amount of uninstall task data for user u

[0082] g u : The number of CPU cycles required per unit data volume for user u's offload task (computational complexity)

[0083] T u : Maximum tolerable delay of user u's offload task

[0084] V u : An ordered set of VNFs required for user u's offload task

[0085] : Available computing resources of base station n

[0086] : Available bandwidth resources of physical link e

[0087] z0: Basic computing resources required to instantiate VNF

[0088] :Node n is the task computing resource allocated to user u for offloading tasks

[0089] b u : The transmission bandwidth required by user u to offload the task

[0090] The technical solutions of the present invention are described as follows.

[0091] 1. The minimum computational delay required for user u's offload task on node n The calculation method is shown in formula (1):

[0092]

[0093] Among them, d u represents the amount of task data for user u, g u represents the task complexity of user u;

[0094] 2. Link cost δ of physical link e e , the calculation method is shown in formula (2):

[0095]

[0096] in, represents the remaining available bandwidth resources of physical link e, B e represents the bandwidth resource capacity of physical link e, b u Indicates the transmission bandwidth required for task offloading by user u;

[0097] 3. Mapping value χ of node n n , the calculation method is shown in formula (3):

[0098]

[0099] 4. Normalized remaining available computing resources of node n The calculation method is shown in formula (4):

[0100]

[0101] 5. Pathway P τ,n The normalized distance The calculation method is shown in formula (5):

[0102]

[0103] 6. Minimum computational delay required for user u to offload a task The calculation method is shown in formula (6):

[0104]

[0105] 7. Node n needs to allocate task computing resources to the corresponding VNF in the SFC required by user u The calculation method is shown in formula (7):

[0106]

[0107] in, represents the load of node n;

[0108] 8. Load of node n The calculation method is shown in formula (8):

[0109]

[0110] Among them, Z n represents the total computing resources of node n;

[0111] 9. Total delay of task calculation The calculation method is shown in formula (9):

[0112]

[0113] 10. The number of deployed VNF instances of type f in the jth particle f j , the calculation method is shown in formula (10):

[0114]

[0115] Where Δf j represents the change in the number of deployed VNF instances of type f in the jth particle; 11. The change in the number of deployed VNF instances of type f in the jth particle Δf j , the calculation method is shown in formula (11):

[0116]

[0117] Among them, 0≤ω≤1, 0≤γ1≤1, 0≤γ2≤1, represents the number of deployed VNF instances of type f in the current optimal solution, f * represents the number of deployed VNF instances of type f in the global optimal solution; 12. The objective function Γ(F,X,Y,Z) is calculated as shown in formula (12):

[0118]

[0119] Among them, 0≤β≤1, Indicates the system task completion rate, Represents the normalized system unit task delay; 13. Sub-algorithm 1: Construct SFC mapping decision and SFC resource allocation decision

[0120] 1) Let the user set be U, and the task tolerance delay T of each user u u , sort the elements in set U in ascending order, set temporary set U′=U, and count variable w=0;

[0121] 2) Let w=w+1, if w≤|F j|, let the temporary set N′=N, the counting variable m=0, remove the set N′ that satisfies Z n / a n ≤z0 condition node, where Z n represents the total computing resources of node n, a n represents the number of VNF instances of node n, z0 represents the basic computing resource overhead of each VNF instance, jump to step 3), otherwise, jump to step 4);

[0122] 3) For the set F j The wth element f in j , if m<f j , let m=m+1, select from N′ the value that satisfies max(Z n / a n ) node n deploys a VNF instance of type f, a n =a n +1, and remove node n from N′, jump to step 3), otherwise, jump to step 2);

[0123] 4) If the collection Take the first user u from the set U′, and build the SFC node and link mapping decision of user u in the jth particle based on the VNF instance type and network resource status of each node n in the set N. according to and the available computing resources of the node, construct the SFC resource allocation decision of user u in the jth particle Jump to step 4), otherwise, jump to step 5);

[0124] 5) For each node n in the set N, remove the VNF instances that are not assigned task computing resources and update F j The corresponding f j ;

[0125] 6) Decision based on SFC node mapping Calculate the remaining available computing resources For the set N that satisfies Each node n, according to the order of elements in U, takes turns to add the minimum unit computing resources for each offload task carried by the node and update the resource allocation decision until

[0126] 7) Output SFC deployment decision The algorithm ends. 14. Sub-algorithm 1: Construct SFC mapping decision

[0127] 11) Initialize the temporary set N′=N, Count variables k = 0, l = 0, determine the access node of user u according to the base station to which the user belongs Make temporary variables

[0128] 12) If k<|V u |, let k = k + 1, where V u Indicates the ordered set of VNFs in the SFC required by user u. Jump to step 13). Otherwise, jump to step 16.

[0129] 13) The set N′ contains the required instances of the kth VNF in SFC and satisfies The nodes join the set N″, where represents the minimum computational delay required for user u’s offload task on node n;

[0130] 14) For each base station n in the set N″, calculate the minimum cost path P from node τ to n τ,n , and remove nodes that do not have minimum cost paths from N″;

[0131] 15) Calculate the mapping value χ of each node n in N″ n , take out the node n with the highest mapping value from the set N″ and add it as the mapping node of the kth VNF to the SFC node mapping decision of user u Pathway P τ,n The SFC link mapping decision for user u added as the lth virtual link Let l = l + 1, τ = n, remove node n from the set N′, and jump to step 12);

[0132] 16) Output the SFC node and link mapping decision of user u The algorithm ends.

[0133] 15. Sub-algorithm 2: Constructing SFC resource allocation decision

[0134] 21) Node mapping decision based on user u The mapping nodes of SFC are added to the temporary ordered set N′ in sequence, and the link mapping decision of user u is made. The bandwidth resource b required by the virtual link u Add SFC resource allocation decision And obtain the total transmission delay of the task offloading path in, is the cumulative transmission delay of user u’s task data through each physical link on the offloading path, and the counting variable k is set to 0;

[0135] 22) Calculate the minimum computational delay required for user u to offload the task if Jump to step 23), otherwise, jump to step 29);

[0136] 23) If k<|V u |, let k=k+1, jump to step 24), otherwise, jump to step 25);

[0137] 24) Calculate the task computing resources that the kth node n in N′ needs to allocate to the corresponding VNF in the SFC required by user u make Jump to step 23);

[0138] 25) If the task calculates the total delay Jump to step 26), otherwise, jump to step 28);

[0139] 26) All available computing resources The node is removed from the set N′ if according to The value of , sort the elements in the set N ′ in descending order, and add a unit of computing resource z1 to the VNF instance required by user u on the first node n in the set N ′, let Jump to step 27), otherwise, jump to step 29);

[0140] 27) If Jump to step 26), otherwise, jump to step 28);

[0141] 28) Output the resource allocation decision of user u in the jth particle

[0142] 29) The algorithm ends.

[0143] A method for offloading multifunctional tasks in heterogeneous cellular networks based on VNF instance sharing, wherein the specific implementation method includes the following steps:

[0144] Step 1: Use particle swarm optimization to establish a heterogeneous cellular network service function chain (SFC) deployment optimization model. Assume that the number of network nodes is N, the number of particles in the particle swarm is J, and each particle is represented by F. j ={f j}, where j represents the particle number, f j represents the number of virtual network function VNF instances of type f in the jth particle, and initializes the particle swarm, where f j Take a random integer in the interval [0,|N|] and let the optimal solution objective function value of each particle be in, They represent the optimal VNF instance deployment number decision, optimal SFC node mapping decision, optimal SFC link mapping decision, and optimal SFC resource allocation decision of the jth particle, respectively. The global optimal solution objective function value Γ{F,X,Y,Z}=0, where F,X,Y,Z represent the global optimal VNF instance deployment number decision, global optimal SFC node mapping decision, global optimal SFC link mapping decision, and global optimal SFC resource allocation decision, respectively. The number of particle swarm iterations is I. Let the counting variables i=0, j=0;

[0145] Step 2: Let i = i + 1. If i ≤ I, jump to step 103; otherwise, jump to step 107.

[0146] Step 3: Set j = j + 1. If j ≤ J, jump to step 104. Otherwise, set j = 0 and jump to step 102.

[0147] Step 4: According to F j and user task offloading request, construct node mapping decision X j , link mapping decision Y j and resource allocation decision Z j , and update F j ;

[0148] Step 5: Calculate the objective function value Γ{F j ,X j ,Y j ,Z j},if make Jump to step 106, otherwise, jump to step 103;

[0149] Step 6: If make Jump to step 103, otherwise, jump to step 103;

[0150] Step 7: Output the global optimal solution {F, X, Y, Z} and end.

[0151] The systems, devices, modules, or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0152] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0153] The above embodiments should be understood as merely illustrating the present invention and not as limiting the scope of protection of the present invention. After reading the contents of the present invention, technicians may make various changes or modifications to the present invention, and these equivalent changes and modifications also fall within the scope defined by the claims of the present invention.

Claims

1. A method for offloading multifunctional tasks in heterogeneous cellular networks based on VNF instance sharing, characterized in that: The following steps are involved:

101. Use particle swarm algorithm to establish heterogeneous cellular network service function chain SFC deployment optimization model. Assume that the number of network nodes is N, the number of particles in the particle swarm is J, and each particle is represented by F j ={f j }, where j represents the particle number, f j represents the number of virtual network function VNF instances of type f in the jth particle, and initializes the particle swarm, where f j Take a random integer in the interval [0,|N|] and let the optimal solution objective function value of each particle be in, They represent the optimal VNF instance deployment number decision, optimal SFC node mapping decision, optimal SFC link mapping decision, and optimal SFC resource allocation decision of the jth particle, respectively. The global optimal solution objective function value Γ{F,X,Y,Z}=0, where F,X,Y,Z represent the global optimal VNF instance deployment number decision, global optimal SFC node mapping decision, global optimal SFC link mapping decision, and global optimal SFC resource allocation decision, respectively. The number of particle swarm iterations is I. Let the counting variables i=0, j=0; 102. Let i = i + 1. If i ≤ I, jump to step 103. Otherwise, jump to step 107.

103. Let j = j + 1. If j ≤ J, jump to step 104. Otherwise, let j = 0 and jump to step 102.

104. According to F j and user task offloading request, construct node mapping decision X j , link mapping decision Y j and resource allocation decision Z j , and update F j ; 105. Calculate the objective function value Γ{F j ,X j ,Y j ,Z j },if make Jump to step 106, otherwise, jump to step 103; 106. If make Jump to step 103, otherwise, jump to step 103; 107. Output the global optimal solution {F, X, Y, Z} and end; In step 104, the node mapping decision X is constructed. j , link mapping decision Y j and resource allocation decision Z j , and update F j The method comprises the following steps: 1) Let the user set be U, and the task tolerance delay T of each user u u , sort the elements in set U in ascending order, set temporary set U′=U, and count variable w=0; 2) Let w=w+1, if w≤|F j |, let the temporary set N′=N, the counting variable m=0, remove the set N′ that satisfies Z n / a n ≤z0 condition node, where Z n represents the total computing resources of node n, a n represents the number of virtual network function (VNF) instances of node n, z0 represents the basic computing resource overhead of each VNF instance, jump to step 3), otherwise, jump to step 4); 3) For the set F j The wth element f in j , if m<f j , let m=m+1, select from N′ the value that satisfies max(Z n / a n ) node n deploys a VNF instance of type f, a n =a n +1, and remove node n from N′, jump to step 3), otherwise, jump to step 2); 4) If the collection Take the first user u from the set U′, and build the service function chain SFC node and link mapping decision for user u in the jth particle based on the VNF instance type and network resource status of each node n in the set N. according to and the available computing resources of the node, construct the SFC resource allocation decision of user u in the jth particle Jump to step 4), otherwise, jump to step 5); 5) For each node n in the set N, remove the VNF instances that are not assigned task computing resources and update F j The corresponding f j ; 6) Decision based on SFC node mapping Calculate the remaining available computing resources For the set N that satisfies Each node n, according to the order of elements in U, takes turns to add the minimum unit computing resources for each offload task carried by the node and update the resource allocation decision until 7) Output SFC deployment decision The algorithm ends; In step 4), the SFC resource allocation decision of user u in the jth particle is constructed. The following steps are involved: 21) Node mapping decision based on user u The mapping nodes of SFC are added to the temporary ordered set N′ in sequence, and the link mapping decision of user u is made. The bandwidth resource b required by the virtual link u Add SFC resource allocation decision And obtain the total transmission delay of the task offloading path in, is the cumulative transmission delay of user u’s task data through each physical link on the offloading path, and the counting variable k is set to 0; 22) Calculate the minimum computational delay required for user u to offload the task if Jump to step 23), otherwise, jump to step 29); 23) If k<|V u |, let k=k+1, jump to step 24), otherwise, jump to step 25); 24) Calculate the task computing resources that the kth node n in N′ needs to allocate to the corresponding VNF in the SFC required by user u make Jump to step 23); 25) If the task calculates the total delay Jump to step 26), otherwise, jump to step 28); 26) All available computing resources The node is removed from the set N′ if according to The value of , sort the elements in the set N ′ in descending order, and add a unit of computing resource z1 to the VNF instance required by user u on the first node n in the set N ′, let Jump to step 27), otherwise, jump to step 29); 27) If Jump to step 26), otherwise, jump to step 28); 28) Output the resource allocation decision of user u in the jth particle 29) The algorithm ends; The minimum computational delay required for user u to offload the task in step 22) The calculation method is shown in formula (6): Among them, d u represents the amount of task data for user u, g u represents the task complexity of user u; In step 24), node n needs to allocate task computing resources to the corresponding VNF in the SFC required by user u. The calculation method is shown in formula (7): in, represents the load of node n, and the calculation method is shown in formula (8); In formula (8), Z n represents the total computing resources of node n; The calculation method of the objective function in steps 105 and 106 is shown in formula (12): Among them, 0≤β≤1, Indicates the system task completion rate, Indicates the normalized system unit task latency.

2. A heterogeneous cellular network multifunctional task offloading method based on VNF instance sharing according to claim 1, characterized in that: In step 4), the SFC node and link mapping decision of user u in the jth particle is constructed. The following steps are involved: 11) Initialize the temporary set N′=N, Count variables k = 0, l = 0, determine the access node of user u according to the base station to which the user belongs Make temporary variables 12) If k<|V u |, let k = k + 1, where V u Indicates the ordered set of VNFs in the SFC required by user u. Jump to step 13). Otherwise, jump to step 16. 13) The set N′ contains the required instances of the kth VNF in SFC and satisfies The nodes join the set N″, where represents the minimum computational delay required for user u’s offload task on node n; 14) For each base station n in the set N″, calculate the minimum cost path P from node τ to n τ,n , and remove nodes that do not have minimum cost paths from N″; 15) Calculate the mapping value χ of each node n in N″ n , take out the node n with the highest mapping value from the set N″ and add it as the mapping node of the kth VNF to the SFC node mapping decision of user u Pathway P τ,n The SFC link mapping decision for user u added as the lth virtual link Let l = l + 1, τ = n, remove node n from the set N′, and jump to step 12); 16) Output the SFC node and link mapping decision of user u The algorithm ends.

3. A heterogeneous cellular network multifunctional task offloading method based on VNF instance sharing according to claim 2, characterized in that: The minimum computational delay required for the offloading task of user u on node n in step 13) The calculation method is shown in formula (1): Among them, d u represents the amount of task data for user u, g u represents the task complexity of user u; The minimum cost path P in step 14) τ,n The calculation method is as follows: Use the minimum cost path algorithm to calculate the minimum cost path P from node τ to n τ,n , where the link cost δ of the physical link e e The calculation method is shown in formula (2): in, represents the remaining available bandwidth resources of physical link e, B e represents the bandwidth resource capacity of the physical link e, b u Indicates the transmission bandwidth required for task offloading by user u.

4. A heterogeneous cellular network multifunctional task offloading method based on VNF instance sharing according to claim 3, characterized in that: In step 15), the node n maps the value x n The calculation method is shown in formula (3): in, represents the normalized remaining available computing resources of node n, α、H τ,n Represent the remaining available computing resources of node n, weight factor, path P τ,n The distance is calculated as shown in formula (4). Represents path P τ,n The normalized distance is calculated as shown in formula (5).

5. The method for offloading multifunctional tasks in heterogeneous cellular networks based on VNF instance sharing according to claim 1, characterized in that: The total delay of the task calculation in steps 25) and 27) The calculation method is shown in formula (9):

6. A heterogeneous cellular network multifunctional task offloading method based on VNF instance sharing according to claim 1, characterized in that: In step 104, f is updated j The method is as follows: For each node n in the set N, if there is a VNF instance of type f on node n that has not been assigned task computing resources, remove the instance from node n and set f j =f j -1; then update f according to formula (10) j ; Where Δf j represents the change in the number of deployed VNF instances of type f. The calculation method is shown in formula (11), 0≤ω≤1, 0≤γ1≤1, 0≤γ2≤1, represents the number of deployed VNF instances of type f in the current optimal solution, f * represents the number of deployed VNF instances of type f in the global optimal solution.

Images