CN114173379A - Multi-user computing unloading method based on 5G private network shunt - Google Patents

Abstract

The invention belongs to the field of wireless communication, and particularly relates to a multi-user computing unloading method based on a 5G private network shunt; the method comprises the following steps: constructing a multi-user computing unloading framework based on the 5G private network shunt; constructing an objective function of the combined unloading total time delay and the resource allocation balance degree according to the multi-user computing unloading framework; solving an objective function combining the total unloading time delay and the resource allocation balance degree to obtain an optimal solution of the objective function; determining an unloading decision by a user according to the optimal solution of the objective function; the user unloads the task according to the determined unloading decision; the invention has good expansibility and safety, effectively utilizes the bandwidth and the throughput of each UE device, strengthens the network data processing capability and improves the flexibility and the usability of the network; the method can effectively reduce the average unloading time delay and the average transmission time delay of multiple users, simultaneously balance the workload of each mobile edge terminal device or edge server, and has good economic benefit.

Description

Multi-user computing unloading method based on 5G private network shunt

Technical Field

The invention belongs to the field of wireless communication, and particularly relates to a multi-user computing unloading method based on a 5G private network shunt.

Background

In recent years, mobile user equipment (mobile phones, tablets, etc.) are becoming an important tool for learning, entertainment, social networking and business development, with the rapid growth of mobile data traffic, and this growth is expected to continue in the future. The amount of computing and data exchange between mobile user equipment and the cloud is also increasing dramatically due to the rapid growth of mobile data traffic. To address this situation, Mobile Edge Computing (MEC) technology has been introduced to supplement and enrich cloud Computing. Because a Mobile Edge Computing (MECS) Server is directly deployed on a base station in the Mobile Edge Computing technology, a Mobile user equipment needing Computing and unloading does not need to select a remote cloud Server, but can adopt a near strategy, namely, the Computing load is unloaded to the Edge Server nearby or directly unloaded to a terminal node with other idle resource capacity in the local, so that the Mobile user can receive the near cloud Computing service in the coverage of a wireless access network, and enjoy high-quality network experience, namely, the Mobile Edge Computing expands the Computing resources and storage resources from a core wide area network to the Edge network.

The problems currently encountered by mobile user equipment in offloading local computing tasks to the MECS are: the data transmission of the offloading process causes non-negligible delay between the mobile device and the MECS, and how to cooperatively handle different resource requirements of multiple users in the case that the multiple mobile users need to compute offloading remains difficult. In the prior art, most methods predict the processing time of each task on each candidate MECS using linear regression, while offloading the mobile user's computing tasks to the appropriate MECS based on the previously observed states of the MECS. However, when the number of mobile users needing to compute offload is large and there are multiple MECS to accept offload services, how to solve the problem of minimizing the total delay of multi-user offload and balancing the load of multiple MECS at the same time still lacks a practical solution.

Disclosure of Invention

In view of this, the present invention provides a multi-user computing offloading method based on a 5G private network splitter, including: .

S1: constructing a multi-user computing unloading framework based on the 5G private network shunt; the 5G private network splitter comprises user plane function equipment UPF and a small mobile edge computing platform miniMEP;

s2: constructing an objective function of the combined unloading total time delay and the resource allocation balance degree according to the multi-user computing unloading framework;

s3: solving an objective function combining the total unloading time delay and the resource allocation balance degree to obtain an optimal solution of the objective function; determining an unloading decision by a user according to the optimal solution of the objective function;

s4: and the user unloads the task according to the determined unloading decision.

Preferably, the building of the multi-user computation uninstalling framework based on the 5G private network splitter comprises building of a computation uninstalling total delay framework and building of a computation resource distribution balance framework.

Further, the constructing of the computation offload total latency framework includes:

taking miniMEP as a mobile edge computing cooperative server, and recording the resource available state of each idle service node; using federated resource pairs (p)_j,c_j) Represents the availability status of the computing and storage resources of service node j, where p_jRepresenting computing resources of the service node, c_jRepresenting a storage resource of the service node; defining the intensive computing task of a mobile user k as T_kUsing a federated resource pair (z)_k,e_k) Representing the resource requirement of a mobile user k, wherein the service node is a task T_kThe provided computing resource is z_kService node is task T_kThe storage resource provided is e_k；

Every t time, the mobile user k couples its own joint resource requirements (z)_k,e_k) Sending to miniMEP;

miniMEP according to the residual computing resources, the residual storage resources and the combined resource demand pair (z) of the mobile user of each service node_k,e_k) And calculating the total unloading time delay of each mobile user.

Further, the formula for calculating the total unloading time delay is as follows:

wherein f is₁(x, y) represents an objective function of the total time delay of the offloading, K represents the total number of users, M represents the total number of service nodes, d_kjRepresenting the computational resource margin of the mobile edge compute collaboration server, z_iIndicating the computing resources requested by the user, e_iRepresenting the storage resources applied by the user, J representing the service node set, x_kjIndicating an unload decision parameter.

Further, the computing resource allocation balance framework comprises:

defining a concentration degree parameter u representing that each unloading decision occupies each service node resource, and defining a vector A ═ a (a) representing the accumulated load of each five-fortune node resource₁,a₂,a₃...a_M) (ii) a Wherein, a_jRepresents the normalized cumulative workload of service node j;

and calculating the resource allocation balance degree according to the concentration degree parameter and the vector of the resource accumulated load of each service node.

Further, each service node resource accumulation load a_jIs defined as:

wherein, a_jRepresents the normalized cumulative workload, k, of the service node j₁Indicating the degree of importance to the use of computing resources, k₂Expressing the degree of importance to the use of the storage resource, K expressing the total number of users, x_kjIndicating unload decision parameterAnd (4) counting.

Further, the formula for calculating the resource allocation balance degree is as follows:

wherein f is₂(x, y) represents an objective function of resource allocation balance, M represents the total number of service nodes, y represents the total number of service nodes_jIndicates a busy decision parameter, a_jElements in the vector representing the cumulative load of each service node resource,

and v represents the mean value of the accumulated load of each service node resource, and the mean square error of the accumulated load of each service node resource.

Preferably, the objective function of the joint offload total latency and the resource allocation balance degree is as follows:

min[f(x,y)]＝min[w₁f₁(x,y)+w₂f₂(x,y)]

C1：0≤w₁≤1，0≤w₂≤1；

C2：w₁+w₂＝1

C3：

C4：

wherein f is₁(x, y) an objective function representing the total delay of the unloading, f₂(x, y) denotes resource allocation balancingTarget function of scale, w₁Weight value, w, assigned to the sum of the unloaded delays₂Representing a weight value for resource allocation with a degree of balance, K representing a total number of users, M representing a total number of service nodes, z_kIndicating a service node as a task T_kProvided computing resource, x_kjDenotes an unload decision parameter, p_jRepresenting the computational resources of the service node, y_jIndicates a busy decision parameter, e_kIndicating a service node as a task T_kStorage resources provided, c_jRepresenting the storage resources of the serving node.

Preferably, the process of solving the objective function of the combined total time delay for offloading and the resource allocation balance degree includes: solving the objective function by adopting a weighted resource optimization algorithm, wherein the solving step comprises the following steps:

s1: determining an objective function f₁Weight w of (x, y)₁Function f₂Weight w of (x, y)₂；

S2: according to the weight w₁And a weight w₂Preprocessing an objective function and an inequality constraint condition to obtain a coefficient matrix x of each variable in the objective function;

s3: traversing elements in the coefficient matrix x, and sequentially taking the elements in the coefficient matrix as the value of intcon;

s4: initializing a coefficient matrix A of inequality constraint conditions and a constraint vector b of the inequality constraint conditions; initializing an equality constraint coefficient matrix Aeq and a constraint vector beq of equality constraints; the optimization interval lb of the variable x is an all-zero vector, and ub is an all-1 vector;

s5: and calling an introping () function to implement linear programming on the target function f (x, y) according to the coefficient matrix A of the initialized inequality constraint condition, the constraint vector b of the initialized inequality constraint condition, the coefficient matrix Aeq of the initialized equality constraint condition, the constraint vector beq of the initialized equality constraint condition and the value of intcon, wherein the finally obtained x is the optimal solution xopt.

Further, the formula for solving the objective function by using the weighted resource optimization algorithm is as follows:

[x,f(x,y)]＝intlinprog(f,intcon,A,b,Aeq,beq,lb,ub)

A×x≤b

Aeq×x＝beq

lb≤x≤ub

wherein x represents a coefficient matrix of a variable in an objective function, f (x, y) is the objective function, intling () is a function for performing linear programming on the objective function, intcon represents the position of an integer decision variable in x, a represents a coefficient matrix of an inequality constraint, b represents a constraint vector of the inequality constraint, Aeq represents a coefficient matrix of an equality constraint, beq represents a constraint vector of the equality constraint, lb represents a lower constraint interval limit of the variable x, and ub represents an upper constraint interval limit of the variable x.

The invention has the beneficial effects that: the invention provides a multi-user computing unloading method based on UPF equipment, mini MEP and other 5G private network shunts, which has good expansibility and safety, ensures high availability of industrial Internet equipment through a cluster consisting of a plurality of edge terminal equipment with idle resources, effectively utilizes the bandwidth and throughput of each UE equipment through load balancing, strengthens the network data processing capacity, and improves the flexibility and availability of the network; the method can effectively reduce the average unloading time delay and the average transmission time delay of multiple users, simultaneously balance the workload of each mobile edge terminal device or edge server, and has good economic benefit.

Drawings

FIG. 1 is a block diagram of a system framework of the present invention;

FIG. 2 is a flow chart of a method for multi-user computational offloading incorporating UPF and miniMEP in accordance with the present invention;

FIG. 3 is a flow chart of the present invention for constructing a linear optimization objective function;

FIG. 4 is a flowchart of a weighted resource optimization algorithm of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The invention has proposed a multi-user calculation unloading method based on 5G private network shunt, the structure of the system frame of the invention is shown in figure 1, mainly include the integral framework and user Plane functional equipment UPF (user Plane function) that needs under the applicable 5G scene, small-scale Mobile Edge computing platform miniMEP (mini Mobile Edge platform) and local terminal equipment appear in the position; the base station bound with the mini UPF is used for acquiring data information of the industrial internet equipment transmitted by the CPE and task information which is required to be unloaded by the local equipment, and transmitting the data information to the mini UPF for data screening processing; the miniUPF equipment is used for receiving and acquiring data information transmitted by the base station, screening and judging the data information, receiving data packets which accord with rules into factory edge cloud, collecting the load condition of each CPE by the mini MEP, carrying out corresponding processing, and forwarding the data packets which do not accord with the rules into the core network, so that the time delay that the data packets which visit the local network must pass through the core network is reduced. The mini MEP is used as an edge cooperative server for cooperating the mobile edge terminal and the server, collects the unloading tasks and related information sent from the current user terminal equipment, judges which edge terminal equipment and servers in a local network have sufficient calculation and storage resources, and judges which edge terminal equipment and service to unload to after calculation of an objective function is optimal, thereby solving the problem that the current terminal equipment does not have sufficient calculation task capacity.

The overall process of implementing user offloading according to UPF and miniMEP in the present invention is as shown in fig. 2, specifically, as shown in fig. 3, a method for implementing multi-user computing offloading based on a 5G private network splitter in the present invention includes:

s1: constructing a multi-user computing unloading framework based on the 5G private network shunt; the 5G private network flow divider comprises UPF equipment and a small mobile edge computing platform miniMEP;

s2: constructing an objective function of the combined unloading total time delay and the resource allocation balance degree according to the multi-user computing unloading framework; the unloading total time delay comprises unloading time delay and transmission time delay;

s3: solving an objective function combining the total unloading time delay and the resource allocation balance degree to obtain an optimal solution of the objective function; determining an unloading decision by a user according to the optimal solution of the objective function;

s4: and the user unloads the task according to the determined unloading decision.

Constructing a multi-user calculation unloading framework based on the 5G private network shunt comprises constructing a calculation unloading total time delay framework and constructing a calculation resource distribution balance degree framework; the computation unloading framework based on the 5G private network shunt can ensure that user data directly enters a local enterprise network through UPF shunt without passing through a core network, thereby reducing transmission delay and improving the safety of the whole unloading process; the method for constructing the total computation and unloading delay framework comprises the following steps: taking miniMEP as a mobile edge computing cooperative server, and recording the resource available state of each idle service node; using federated resource pairs (p)_j,c_j) Represents the availability status of the computing and storage resources of service node j, where p_jRepresenting computing resources of the service node, c_jRepresenting a storage resource of the service node; defining the intensive computing task of a mobile user k as T_kUsing a federated resource pair (z)_k,e_k) Representing the resource requirement of a mobile user k, wherein the service node is a task T_kThe provided computing resource is z_kService node is task T_kThe storage resource provided is e_k(ii) a The service node comprises a mobile edge terminal and an edge server; every t time, at the beginning of each t time period, the mobile user k couples the own joint resource requirement (z)_k,e_k) Sending to miniMEP; miniMEP according to the residual computing resources, the residual storage resources and the combined resource demand pair (z) of the mobile user of each service node_k,e_k) The total offload delay is calculated for each mobile user.

Computing resource margin d of mobile edge computing collaboration server_kjComprises the following steps:

wherein z is_iIndicating the computing resources requested by the user, e_iThe storage resources applied by the user are represented, and J represents a service node set.

In order to represent the service and served relationship between mobile users and mobile edge terminals and edge servers, a matrix X is introduced_i1×MRepresenting the overall unloading decision, k rows and j columns of the matrix, and using the element x of the matrix_kjDefined as an offload decision parameter, indicating whether the mobile user K offloads the computing task to the service node j, where K belongs to K and j belongs to M. K is the total number of mobile users, M is the total number of service nodes, and the unloading judgment parameter x_kjThe assignments are as follows:

it is assumed that a mobile terminal can only offload computation tasks to at most one edge terminal node or edge server, and therefore at most one of the elements in each row of the matrix X has a value of 1, and the remaining elements in this row are all 0. Further, the lead-in vector Y is Y_i1×MTo identify whether service node j offers offload services for mobile users, its element y_i(j belongs to M) is defined as a busy judgment parameter which indicates whether the service node j provides the unloading service for the mobile user, and the busy judgment parameter is assigned as follows:

the method for constructing the frame of the computing resource allocation balance degree comprises the following steps:

to evaluate the load balancing of the offloading decision, two parameters are defined: one is a parameter u characterizing the concentration degree of the unloading scheme occupying the service node, and the other is a vector A ═ of the accumulated load of the edge terminal node or the edge server resource (a)₁,a₂,a₃...a_M). Element a of A_j(j. epsilon. M) and the parameter u are normalized parameters with values of [0,1 [ ]]Within the scope, the concentration level parameter u is defined as:

the vector A of the resource accumulation load of each service node is equal to (a)₁,a₂,a₃...a_M) Middle element a_jIs defined as:

wherein, a_j(j ∈ M) represents the normalized cumulative workload of service node j (comprehensively considering the computing resource and storage resource usage loads), and the normalized resource cumulative loads a of all M service nodes₁,a₂,a₃...a_MComponent resource cumulative load vector a ═ a₁,a₂,a₃...a_M)；k₁Indicating the degree of importance to the use of computing resources, k₂Representing the degree of importance of the use of the storage resource, the weight k₁And k₂K is more than or equal to 0₁,k₂Is less than or equal to 1, and k₁+k₂1 is ═ 1; preferably, the same importance is attached to the usage of the computing resources and storage resources of the edge terminal node or edge server, i.e. k₁＝k₂＝0.5。

The method for constructing the objective function of the total time delay of the joint unloading and the resource allocation balance degree comprises the following steps:

firstly, minimizing the total time delay of the calculation unloading of the mobile user, improving the experience of the mobile user, secondly, balancing the resource use load on the service node, and avoiding that the unbalanced calculation unloading scheme prematurely exhausts the resource on the service node as much as possible, therefore, the optimization of the calculation unloading decision is the optimization of two targets of the total time delay of the unloading and the resource distribution balance degree, and the two targets are converted into a linear optimization problem based on an integer, so that an objective function combining the total time delay of the unloading and the resource distribution balance degree is constructed:

min[f(x,y)]＝min[w₁f₁(x,y)+w₂f₂(x,y)]

C1：0≤w₁≤1，0≤w₂≤1；

C2：w₁+w₂＝1

C3：

C4：

wherein f is₁(x, y) an objective function representing the total delay of the unloading, f₂(x, y) an objective function representing the degree of balance of resource allocation, w₁Weight value, w, assigned to the sum of the unloaded delays₂A weight value indicating an assignment of a resource utilization balance,

and v represents the mean value of the accumulated load of each service node resource, and the mean square error of the accumulated load of each service node resource.

The method for solving the objective function of the combined unloading total time delay and the resource allocation balance degree adopts a weighted resource optimization algorithm to solve the objective function, and comprises the following steps:

for a fully distributed computing offload scheme, when all edge terminals or edge servers have completionWhen the resource use loads of the same proportion are all the same, f₂(x, y) takes a minimum possible value of 0; as the difference of the resource utilization load ratios of different edge terminals or edge servers becomes larger, and the distribution of the edge terminals or edge servers occupied by the unloading calculation tends to be concentrated, f₂The value of (x, y) also increases, asymptotically reaching 1, and the imbalance degree of resource usage is the highest. The constraint on the resource load of each edge terminal or edge server is simple, that is, the sum of the resources allocated to each mobile user must not exceed the total resource upper limit. Using linear weighting to weight the function f₁(x,y)、f₂(x, y) are aggregated together to obtain an objective function f (x, y) representing the whole optimization problem, and when f (x, y) obtains the minimum value, the total unloading time delay of the calculation unloading scheme and the resource allocation balance degree of the service node are generally optimal; as shown in fig. 4, the step of solving includes:

s1: determining an objective function f₁Weight w of (x, y)₁Function f₂Weight w of (x, y)₂；

S2: according to the weight w₁And a weight w₂Preprocessing an objective function and an inequality constraint condition to obtain a coefficient matrix x of each variable in the objective function;

s3: traversing elements in the coefficient matrix x, and sequentially taking the elements in the coefficient matrix as the value of intcon;

s4: initializing a coefficient matrix A of inequality constraint conditions and a constraint vector b of the inequality constraint conditions; initializing an equality constraint coefficient matrix Aeq and a constraint vector beq of equality constraints; the optimization interval lb of the variable x is an all-zero vector, and ub is an all-1 vector;

s5: and calling an introping () function to implement linear programming on the target function f (x, y) according to the coefficient matrix A of the initialized inequality constraint condition, the constraint vector b of the initialized inequality constraint condition, the coefficient matrix Aeq of the initialized equality constraint condition, the constraint vector beq of the initialized equality constraint condition and the value of intcon, wherein the finally obtained x is the optimal solution xopt.

The formula for solving the objective function by adopting the weighted resource optimization algorithm is as follows:

[x,f(x,y)]＝intlinprog(f,intcon,A,b,Aeq,beq,lb,ub)

A×x≤b

Aeq×x＝beq

lb≤x≤ub

wherein x represents a coefficient matrix of a variable in an objective function, f (x, y) is the objective function, intling () is a function for performing linear programming on the objective function, intcon represents the position of an integer decision variable in x, a represents a coefficient matrix of an inequality constraint, b represents a constraint vector of the inequality constraint, Aeq represents a coefficient matrix of an equality constraint, beq represents a constraint vector of the equality constraint, lb represents a lower constraint interval limit of the variable x, and ub represents an upper constraint interval limit of the variable x.

The optimal solution xopt is the optimal solution of an objective function of the joint unloading total time delay and the resource allocation balance degree, and then the system guides each mobile terminal to carry out unloading according to an unloading scheme corresponding to xopt.

The invention provides a multi-user computing unloading method based on UPF equipment, mini MEP and other 5G private network shunts, which has good expansibility and safety, ensures high availability of industrial Internet equipment through a cluster consisting of a plurality of edge terminal equipment with idle resources, effectively utilizes the bandwidth and throughput of each UE equipment through load balancing, strengthens the network data processing capacity, and improves the flexibility and availability of the network; the method can effectively reduce the average unloading time delay and the average transmission time delay of multiple users, simultaneously balance the workload of each mobile edge terminal device or edge server, and has good economic benefit.

The above-mentioned embodiments, which further illustrate the objects, technical solutions and advantages of the present invention, should be understood that the above-mentioned embodiments are only preferred embodiments of the present invention, and should not be construed as limiting the present invention, and any modifications, equivalents, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. A multi-user computing unloading method based on a 5G private network splitter is characterized by comprising the following steps:

s1: constructing a multi-user computing unloading framework based on the 5G private network shunt; the 5G private network splitter comprises user plane function equipment UPF and a small mobile edge computing platform miniMEP;

s2: constructing an objective function of the combined unloading total time delay and the resource allocation balance degree according to the multi-user computing unloading framework;

s3: solving an objective function combining the total unloading time delay and the resource allocation balance degree to obtain an optimal solution of the objective function; determining an unloading decision by a user according to the optimal solution of the objective function;

s4: and the user unloads the task according to the determined unloading decision.

2. The method according to claim 1, wherein the constructing of the multi-user computation offload framework based on the 5G private network splitter comprises constructing a computation offload total delay framework and constructing a computation resource allocation balance framework.

3. The method of claim 2, wherein the constructing of the total delay frame for computation offload comprises:

taking miniMEP as a mobile edge computing cooperative server, and recording the resource available state of each idle service node; using federated resource pairs (p)_j,c_j) Represents the availability status of the computing and storage resources of service node j, where p_jRepresenting computing resources of the service node, c_jRepresenting a storage resource of the service node; defining the intensive computing task of a mobile user k as T_kUsing a federated resource pair (z)_k,e_k) Representing the resource requirement of a mobile user k, wherein the service node is a task T_kThe provided computing resource is z_kService node is task T_kThe storage resource provided is e_k；

Every t times, mobile user k will be selfThe already federated resource requirement pairs (z)_k,e_k) Sending to miniMEP;

miniMEP according to the residual computing resources, the residual storage resources and the combined resource demand pair (z) of the mobile user of each service node_k,e_k) And calculating the total unloading time delay of each mobile user.

4. The multi-user calculation unloading method based on the 5G private network splitter according to claim 3, wherein the total calculation unloading delay formula is as follows:

wherein f is₁(x, y) represents an objective function of the total time delay of the offloading, K represents the total number of users, M represents the total number of service nodes, d_kjRepresenting the computational resource margin of the mobile edge compute collaboration server, z_iIndicating the computing resources requested by the user, e_iRepresenting the storage resources applied by the user, J representing the service node set, x_kjIndicating an unload decision parameter.

5. The multi-user computing offloading method based on the 5G private network splitter according to claim 2, wherein constructing the computing resource allocation balance framework comprises:

defining a concentration degree parameter u representing that unloading decisions occupy resources of each service node, and defining a vector A which represents the accumulated load of the resources of each service node as (a)₁,a₂,a₃...a_M) (ii) a Wherein, a_jRepresents the normalized cumulative workload of service node j;

and calculating the resource allocation balance degree according to the concentration degree parameter and the vector of the resource accumulated load of each service node.

6. The method of claim 5, wherein the cumulative load a of each service node resource is used as a load for the multi-user computation offload based on the 5G private network splitter_jThe expression of (a) is:

wherein, a_jRepresents the normalized cumulative workload, k, of the service node j₁Indicating the degree of importance to the use of computing resources, k₂Expressing the degree of importance to the use of the storage resource, K expressing the total number of users, x_kjIndicating an unload decision parameter.

7. The multi-user computing offloading method based on the 5G private network splitter according to claim 5, wherein the computing resource allocation balance formula is:

wherein f is₂(x, y) represents an objective function of resource allocation balance, M represents the total number of service nodes, y represents the total number of service nodes_jIndicates a busy decision parameter, a_jIndicating that each service node resource is cumulatively negativeThe elements in the vector of the charges are,

and v represents the mean value of the accumulated load of each service node resource, and the mean square error of the accumulated load of each service node resource.

8. The multi-user computing offloading method based on the 5G private network splitter according to claim 1, wherein an objective function combining total offloading delay and resource allocation balance is:

min[f(x,y)]＝min[w₁f₁(x,y)+w₂f₂(x,y)]

C1：0≤w₁≤1，0≤w₂≤1；

C2：w₁+w₂＝1

C3：

C4：

wherein f is₁(x, y) an objective function representing the total delay of the unloading, f₂(x, y) an objective function representing the degree of balance of resource allocation, w₁Weight value, w, assigned to the sum of the unloaded delays₂Representing a weight value for resource allocation with a degree of balance, K representing a total number of users, M representing a total number of service nodes, z_kIndicating a service node as a task T_kProvided computing resource, x_kjDenotes an unload decision parameter, p_jRepresenting the computational resources of the service node, y_jIndicates a busy decision parameter, e_kIndicating a service node as a task T_kStorage resources provided, c_jRepresenting the storage resources of the serving node.

9. The multi-user computing offloading method based on the 5G private network splitter according to claim 1, wherein the process of solving the objective function combining the total offloading delay and the resource allocation balance degree comprises: solving the objective function by adopting a weighted resource optimization algorithm, wherein the solving step comprises the following steps:

s1: determining an objective function f₁Weight w of (x, y)₁Function f₂Weight w of (x, y)₂；

S2: according to the weight w₁And a weight w₂Preprocessing an objective function and an inequality constraint condition to obtain a coefficient matrix x of each variable in the objective function;

s3: traversing elements in the coefficient matrix x, and sequentially taking the elements in the coefficient matrix as the value of intcon;

s4: initializing a coefficient matrix A of inequality constraint conditions and a constraint vector b of the inequality constraint conditions; initializing an equality constraint coefficient matrix Aeq and a constraint vector beq of equality constraints; the optimization interval lb of the variable x is an all-zero vector, and ub is an all-1 vector;

s5: and calling an introping () function to implement linear programming on the target function f (x, y) according to the coefficient matrix A of the initialized inequality constraint condition, the constraint vector b of the initialized inequality constraint condition, the coefficient matrix Aeq of the initialized equality constraint condition, the constraint vector beq of the initialized equality constraint condition and the value of intcon, wherein the finally obtained x is the optimal solution xopt.

10. The multi-user computing offloading method based on the 5G private network splitter according to claim 9, wherein a formula for solving the objective function by using a weighted resource optimization algorithm is as follows:

[x,f(x,y)]＝intlinprog(f,intcon,A,b,Aeq,beq,lb,ub)

A×x≤b

Aeq×x＝beq

lb≤x≤ub

wherein x represents a coefficient matrix of a variable in an objective function, f (x, y) is the objective function, intling () is a function for performing linear programming on the objective function, intcon represents the position of an integer decision variable in x, a represents a coefficient matrix of an inequality constraint, b represents a constraint vector of the inequality constraint, Aeq represents a coefficient matrix of an equality constraint, beq represents a constraint vector of the equality constraint, lb represents a lower constraint interval limit of the variable x, and ub represents an upper constraint interval limit of the variable x.

Images