Method for secure mobile edge computing and caching in ultra-dense millimeter wave networks

Abstract

The application discloses a method for secure mobile edge computing and caching in an ultra-dense millimeter wave network, which comprises the following steps: constructing a network system according to network basic information, and constructing an optimization problem under the constraint of the network system; obtaining an initial solution set according to the optimization problem, defining the initial solution set as a grey wolf population, initializing the population by using a Tent chaotic mapping method to obtain a new initial population, then performing the operations of searching for prey, surrounding the prey and attacking the prey on the initial population to obtain an optimal solution, then performing adaptive Gaussian-Kolmogorov mutation on the optimal solution to obtain a final solution, avoiding falling into a local optimum, and performing secure computing and caching unloading optimization configuration according to the global final solution. The application has the advantages of multi-base station caching and unloading resource configuration, and can well achieve the target of minimizing energy consumption and cost under the condition of meeting power, time delay and safety constraints.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wireless communication, in particular to a method for secure mobile edge computing and caching in an ultra-dense millimeter wave network. BACKGROUND

[0002] With the rapid development of 5G communication technology, new computing-intensive applications such as holographic communication, cloud gaming, industrial digital twin, smart city perception, remote surgery, and AI reasoning services have emerged in large numbers, placing higher demands on network latency, computing power, and data security. However, due to the inherent bottleneck in the hardware size and endurance design of terminal devices, both the computing speed and available energy are difficult to support continuous high-intensity operation requirements. Mobile edge computing (MEC) as an effective solution, by deploying computing and caching resources at the network edge, enables terminal devices to offload computing tasks to edge servers for execution, thereby significantly reducing response latency, saving energy consumption, and improving user experience.

[0003] To make computing power closer to users, an ultra-dense network architecture is introduced into the MEC system. Under this architecture, multiple small base stations are densely deployed within the coverage of a macro base station, and users can simultaneously establish connections with multiple base stations to achieve multi-access edge computing. This architecture effectively improves network capacity and resource utilization, but also brings new challenges. For example, close distances between base stations can cause serious signal interference, and frequent user switching can lead to task interruption or cache invalidation. In addition, to ensure the security of the task offloading process, the system often needs to introduce encryption, authentication, and other mechanisms, which further increase the computing cost of edge nodes. Therefore, how to reasonably schedule caching and computing resources while ensuring data security and optimize task offloading strategies has become a key problem that needs to be solved.

[0004] To address the above problems, it is necessary to study how to improve the overall performance of the system, reduce task response time, and achieve efficient use of resources while ensuring data security. However, the traditional grey wolf algorithm has poor search capability and is prone to local search, which has obvious shortcomings in practical applications and needs to be improved. SUMMARY

[0005] To address the current technical shortcomings, the present application proposes a method for secure mobile edge computing and caching in an ultra-dense millimeter wave network, aiming to overcome the shortcomings of previous research that neglects new variables in the offloading process, resulting in degraded algorithm optimization performance.

[0006] To achieve the above purpose, the present application provides the following technical solution: a method for secure mobile edge computing and caching in an ultra-dense millimeter wave network, comprising the following steps:

[0007] Step S1: Obtain network system of user terminal device, base station, channel, task and security service in cache offloading assisted multi-access mobile edge computing network, network basic information construction network system; The network system includes network model, communication model, cache and calculation model and security model; Under the constraint of network system, an optimization problem is constructed;

[0008] Step S2: Introducing Tent chaotic mapping method, nonlinear convergence factor and adaptive Gaussian-Kossai variation strategy in traditional grey wolf algorithm to form improved grey wolf algorithm; The improved grey wolf algorithm is used for searching and processing the optimization problem, and the global optimal solution is output; The search processing includes search prey stage, surround prey stage, attack prey stage and global optimal grey wolf updating operation stage;

[0009] Step S3: When the global optimal solution does not change after 100 times of iteration of the improved grey wolf algorithm, the adaptive Gaussian-Kossai variation strategy of stagnation judgment mechanism is used to disturb the current global optimal solution, and the disturbed global optimal solution is output;

[0010] Step S4: Based on the disturbed global optimal solution, the multi-access mobile edge computing network offloading optimization configuration is executed.

[0011] Further, the optimization problem is constructed, which is represented as:

[0012] ;

[0013] In the formula, is minimized; is the objective function; represents a set composed of all user and base station association decision variables; ; is a universal quantifier; is a decision indicator of whether the user is associated with the base station; n is the base station; is a base station index set; o is the user; is a user index set; represents a set composed of all user millimeter wave subchannel selection decision variables, ; is a decision indicator of user selected millimeter wave subchannel; s is the millimeter wave subchannel; is a millimeter wave subchannel index set; represents a set of decision variables of whether the user's computing task is executed at the base station, ; is an offloading decision indicator of the user's computing task at the base station; is the computing task; is a computing task index set; a set of decision variables representing whether the user task selects a cipher algorithm, , a decision indicator whether the user's computational task selects a cipher algorithm of the user's local device; u is a cipher algorithm of the user's local device; a set of indices for cipher algorithms of the user's local device; a set representing the computational capacity of all users, ; a computational capacity of the user device; a set representing the transmit power of all users, ; a transmit power of the user o; a local computational overhead of the user o; is restricted to; a constraint condition that the user's task processing latency does not exceed the maximum execution time of the computational task; a processing latency of the computational task y of the user o; a maximum allowed execution time of the computational task y of the user o; a constraint condition that for each user o, the sum of the association indices with all base stations n is 1; a constraint condition that for each user o, the sum of the selection variables of all millimeter wave sub-channels s is 1; a constraint condition that for each computational task y of each user o, the sum of the task execution variables on all base stations n does not exceed 1; a constraint condition that for each computational task y of each user o, the sum of the selection variables of all cipher algorithms is 1; a constraint condition that the overall security vulnerability cost of the user o does not exceed the maximum allowed security vulnerability cost; a user security vulnerability cost; a maximum allowed user security vulnerability cost; a constraint condition of the minimum and maximum allowed computational capacity of the user; represents a positive number set to be small enough; a maximum allowed computational capacity of the user; a constraint condition of the lower and upper limits of the transmit power of the user o; an upper limit of the user transmit power; a constraint condition that the association index of the user with the base station is a binary variable taking 0 or 1; a constraint condition that the variable of the user selecting the millimeter wave sub-channel is a binary variable taking 0 or 1; a constraint condition that the variable of the user's computational task executing at the base station is a binary variable taking 0 or 1; The variable for selecting a cipher algorithm for a user is a binary variable, which is subject to a constraint of taking 0 or 1.

[0014] Further, the specific process of outputting the global optimal solution in step S2 is as follows:

[0015] According to the constraint condition of the optimization problem, a set of initial solution is randomly generated within the constraint range of the population initialization; the initial solution set is disturbed and redistributed by using the Tent chaotic mapping method to construct the initial population;

[0016] The maximum number of iterations of the improved grey wolf algorithm , and the current iteration number is set to 1;

[0017] Define a virtual user set ; ; For any specific virtual user in the virtual user set ; The total number of virtual users is

[0018] The single grey wolf of the initial population is , and the total number of grey wolves in the initial population is , ; Replace with ; The user associated with the base station corresponding to the grey wolf is an index set; The base station index when the user is associated with the grey wolf is represented; The set of millimeter wave sub-channels selected by the user corresponding to the grey wolf is The millimeter wave sub-channel index when the user is associated with the grey wolf is represented; The set of base stations serving the virtual user corresponding to the grey wolf is The base station index serving the virtual user when the user is associated with the grey wolf is represented; The set of cipher algorithms selected by the virtual user corresponding to the grey wolf is The encryption algorithm index selected by the virtual user when the user is associated with the grey wolf is represented; The set of local computing capabilities of the user corresponding to the grey wolf is The local computing capability when the user is associated with the grey wolf is represented; The set of transmit powers of the user corresponding to the grey wolf is The transmit power when the user is associated with the grey wolf is represented;

[0019] The fitness function of the grey wolf in the initial population is constructed, which is represented as:

[0020] ;

[0021] In the formula, For users The penalty factor; For users computational tasks The penalty factor; Represents the fitness function;

[0022] Given a random initial value within the interval [0,1] of the initial population, a chaotic sequence is generated by repeatedly applying this initial value. Chaotic sequence Mapping to the solution space, in which the chaotic sequence is represented Converting to the position of the gray wolf in the gray wolf algorithm, the gray wolf position update process is represented as follows:

[0023] ;

[0024] In the formula, For the Tent chaotic mapping method The value of the next iteration; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the user's side, corresponding to the first The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; In the t-th iteration, the v-th gray wolf is in the t-th iteration. For each user, the initial position of the parameter corresponding to the k-th decision variable; This represents the upper limit of the value of the k-th decision variable parameter; This represents the lower bound of the value of the parameter of the k-th decision variable; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the [number] user, corresponding to the [number]th [user] The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the [number] user, corresponding to the [number]th [user] The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the [number] user, corresponding to the [number]th [user] The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the [number] user, corresponding to the [number]th [user] The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter;

[0025] Using the fitness function Calculate all gray wolves in the gray wolf population The fitness value; based on , , , , , The fitness values ​​are determined, and the gray wolf with the highest fitness value is selected. As the best gray wolf in the entire game;

[0026] Determine the current iteration number Is it less than or equal to the maximum number of iterations? Current iteration number Less than or equal to the maximum number of iterations Then, the gray wolf population will undergo a prey search phase, a prey encirclement phase, a prey attack phase, and a global optimal gray wolf update phase.

[0027] Current iteration number Greater than the maximum number of iterations Then, the globally best gray wolf in the target population is output as the global optimal solution.

[0028] Furthermore, the specific process for executing the prey-hunting phase is as follows:

[0029] During the prey-hunting phase, the gray wolf population was classified according to its fitness value, with the gray wolves having the highest fitness being divided into... Wolf, Wolf, Wolf; Wolf, Wolf, All wolves are alpha wolves; during the prey-hunting optimization phase, gray wolves within the pack update their own positions based on the alpha wolf's optimal position using the following formula:

[0030] ;

[0031] In the formula, for The distance from the optimal position of the leader wolf; It is a random factor; For the vth gray wolf in the th In a scenario with one user, the optimal position of the leader wolf in the user-base station correlation index; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with a single user, the optimal position of the leader wolf in the millimeter-wave sub-channel index; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with one user, the optimal position of the leader wolf in the service base station set; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with multiple users, the optimal position of the leader wolf in the cryptographic algorithm index; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with a single user, the optimal position of the leader wolf in terms of local computing power; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with one user, the optimal position of the leader wolf in terms of transmit power; For the first In the iteration, the v-th wolf is in the ... In a scenario with one user, the updated position of the user-base station correlation index; The convergence factor; For the first In the iteration, the v-th wolf is in the ... In a scenario with one user, the updated position of the sub-channel index; For the first In the iteration, the v-th wolf is in the ... In a scenario with one user, the updated location of the service base station set; For the first The updated position of the local computing capability of the vth wolf in the scenario of the euh user at the ith iteration; The updated position of the password algorithm index of the vth wolf in the scenario of the euh user at the ith iteration; The updated position of the local computing capability of the vth wolf in the scenario of the euh user at the ith iteration; The updated position of the local computing capability of the vth wolf in the scenario of the euh user at the ith iteration; The updated position of the local computing capability of the vth wolf in the scenario of the euh user at the ith iteration; The updated position of the local computing capability of the vth wolf in the scenario of the euh user at the ith iteration; The updated position of the local computing capability of the vth wolf in the scenario of the euh user at the ith iteration. Further, the specific process of the surrounding prey stage is as follows:

[0032] In the surrounding prey stage, the step length of the grey wolf is controlled by

[0033] and ; wherein, ; is a nonlinear convergence factor; ; represents generating a first random number between 0 and 1; is a natural constant; when , it indicates that the grey wolf approaches the prey; when , it indicates that the grey wolf moves away from the prey; wherein,

[0034] , represents generating a second random number between 0 and 1; when , it indicates that the influence of the obstacle on the grey wolf is small; when , it indicates that the influence of the obstacle on the grey wolf is large; When the grey wolf approaches the prey and the influence of the obstacle on the grey wolf is small, the attacking prey stage is executed; when the grey wolf moves away from the prey and the influence of the obstacle on the grey wolf is large, the fitness value is recalculated.

[0035] Further, the specific process of the attacking prey stage is as follows:

[0036] Let

[0037] represent , , ; Let represent , , ; Let represent , ; Let represent , ; Let 、 、 ; let denote 、 、 ;

[0038] attack the prey phase, denoted by:

[0039] ;

[0040] wherein is the distance of the alpha wolf from in the dimension; is a first independent random factor; is the position of the alpha wolf at iteration t of the v-th grey wolf in the decision dimension for the o-th user; is the distance of the beta wolf from in the dimension; is a second independent random factor; is the position of the beta wolf at iteration t of the v-th grey wolf in the decision dimension for the o-th user; is the distance of the delta wolf from in the dimension; is a third independent random factor; is the position of the delta wolf at iteration t of the v-th grey wolf in the decision dimension for the o-th user; is the distance of the alpha wolf from in the dimension; is the position of the alpha wolf at iteration t of the v-th grey wolf in the decision dimension for the o-th user; is the distance of the beta wolf from in the dimension; is the position of the wolf at iteration t of the v-th grey wolf in the decision dimension for the o-th user; is the distance of the delta wolf from in the dimension; is the position of the wolf at iteration t of the v-th grey wolf in the decision dimension for the o-th user; is the distance of the alpha wolf from in the dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the alpha wolf; is the distance of the vth grey wolf from the alpha wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the beta wolf; is the distance of the vth grey wolf from the beta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the delta wolf; is the distance of the vth grey wolf from the delta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the alpha wolf; is the distance of the vth grey wolf from the alpha wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the beta wolf; is the distance of the vth grey wolf from the beta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the delta wolf; is the distance of the vth grey wolf from the delta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the alpha wolf; is the distance of the vth grey wolf from the alpha wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the beta wolf; is the distance of the vth grey wolf from the beta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the delta wolf; is the distance of the vth grey wolf from the delta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the alpha wolf; is the distance of the vth grey wolf from the alpha wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the beta wolf; is the distance of the vth grey wolf from the beta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the delta wolf; is the distance of the vth grey wolf from the delta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the alpha wolf; is the distance of the vth grey wolf from the alpha wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the beta wolf; is the distance of the vth grey wolf from the beta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the delta wolf; is the distance of the vth grey wolf from the delta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the alpha wolf; is the distance of the vth grey wolf from the alpha wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the beta wolf; is the distance of the vth grey wolf from the beta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the delta wolf; is the distance of the vth grey wolf from the delta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the alpha wolf; is the distance of the vth grey wolf from the alpha wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the beta wolf; is the distance of the vth grey wolf from the beta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the delta wolf; is the distance of the vth grey wolf from the delta wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the alpha wolf; is the distance of the vth grey wolf from the alpha wolf on the decision dimension; is the position of the vth grey wolf on the decision dimension for the oth user at the tth iteration of the beta wolf; is the distance of the vth grey wolf from the beta wolf on the decision dimension; distance in f-dimension; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the alpha wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the beta wolf; distance in f-dimension; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the beta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; distance in f-dimension; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the alpha wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the alpha wolf; is the step size factor corresponding to the alpha wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the beta wolf; is the step size factor corresponding to the beta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the step size factor corresponding to the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the step size factor corresponding to the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the step size factor corresponding to the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the step size factor corresponding to the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the step size factor corresponding to the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the step size factor corresponding to the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the step size factor corresponding to the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the step size factor corresponding to the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; is the position of the vth grey wolf for the oth user in f decision dimensions at the tth iteration for the delta wolf; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension; For the vth grey wolf at the t+1th iteration for the oth user in the decision dimension. Further, the specific process of the updating operation stage of the global optimal grey wolf is:

[0041] The updating operation stage of the global optimal grey wolf is represented as:

[0042]

[0043] ;

[0044] , , , , , as the global optimal solution.

[0045] Further, the specific process of outputting the perturbed global optimal solution is:

[0046] After obtaining the global optimal solution, a stagnation judgment mechanism is introduced. When the global optimal solution does not change after 100 consecutive iterations, it is determined that the improved grey wolf algorithm is in stagnation, and the adaptive Gaussian Cauchy mutation strategy of the stagnation judgment mechanism is started to perturb the current global optimal solution, represented as: ​​​

[0047] ;

[0048] In the formula, is the global optimal position of the user-base station association index after disturbance of the vth grey wolf in the scene of the oth user; is the weight coefficient of the Cauchy distribution disturbance term; is a random number subject to Cauchy distribution; is the weight coefficient of the Gaussian distribution disturbance term; is a random number subject to Gaussian distribution; is the global optimal position of the millimeter wave subchannel index after disturbance of the vth grey wolf in the scene of the oth user; is the global optimal position of the service base station set after disturbance of the vth grey wolf in the scene of the oth user; is the global optimal position of the cryptographic algorithm index after disturbance of the vth grey wolf in the scene of the oth user; is the global optimal position of the local computing capability after disturbance of the vth grey wolf in the scene of the oth user; is the global optimal position of the transmit power after disturbance of the vth grey wolf in the scene of the oth user;

[0049] the value of the current iteration number t of the improved grey wolf algorithm is added by 1;

[0050] after updating the value of t, the global optimal solution after disturbance of the grey wolf is , , , , , as the global optimal solution after disturbance of the grey wolf; when the current iteration number t reaches the maximum iteration number, the global optimal solution after disturbance is output.

[0051] Compared with the prior art, the present application has the following beneficial effects:

[0052] The present application adopts the improved grey wolf algorithm AGWO: high-quality initial population is generated through Tent chaotic mapping method, nonlinear convergence factor dynamically balances global search and local development capability, adaptive Gaussian Cauchy mutation strategy greatly disturbs the optimal solution when it is stagnant, in the early stage of the algorithm, Cauchy expands the search range of the group, in the later stage of the algorithm, Gaussian makes individuals search in a small range, enhances the local search ability of the group, and makes the improved grey wolf algorithm quickly converge. The improved grey wolf algorithm iteratively outputs the global optimal solution, finally executes the multi-access mobile edge computing network offloading optimization configuration based on the global optimal solution after disturbance, significantly reduces the energy consumption and cost, meets the power, delay and security constraints, effectively improves the system energy efficiency and offloading success rate. BRIEF DESCRIPTION OF DRAWINGS​

[0053] Figure 1 A flow chart of the method of the present application.

[0054] Figure 2 A diagram showing the effect of the number of user iterations on the fitness value according to the present application.

[0055] Figure 3 A diagram showing the effect of the density of user task volume on energy consumption according to the present application.

[0056] Figure 4 A diagram showing the effect of the density of user task volume on total cost according to the present application. DETAILED DESCRIPTION

[0057] As shown in Figure 1 , the present application provides a technical solution: a method for secure mobile edge computing and caching in an ultra-dense millimeter wave network, comprising:

[0058] Step S1: Obtain network base information of user terminal devices, base stations, channels, tasks and security services in a cache offloading assisted multi-access mobile edge computing network to construct a network system; the network system includes a network model, a communication model, a cache and computing model and a security model; an optimization problem is constructed under the constraints of the network system;

[0059] Step S2: Introduce Tent chaotic mapping method, nonlinear convergence factor and adaptive Gaussian-Kolmogorov mutation strategy in the traditional grey wolf algorithm to form an improved grey wolf algorithm; use the improved grey wolf algorithm to search and process the optimization problem, and output the global optimal solution; the search and processing includes search prey stage, surround prey stage, attack prey stage and global optimal grey wolf update operation stage;

[0060] Step S3: When the global optimal solution does not change after 100 iterations of the improved grey wolf algorithm, disturb the current global optimal solution through the adaptive Gaussian-Kolmogorov mutation strategy of the stagnation judgment mechanism, and output the disturbed global optimal solution;

[0061] Step S4: Perform multi-access mobile edge computing network offloading optimization configuration based on the disturbed global optimal solution.

[0062] Among them, the cache offloading assisted multi-access mobile edge computing network is an ultra-dense MEC network.

[0063] Among them, the specific process of constructing the network model is:

[0064] In the ultra-dense MEC network, there are base stations, and the base station index set is , wherein Represents any base station in the MEC network. This represents the number of base stations in an ultra-dense MEC network, each equipped with a cache and compute server; the ultra-dense MEC network also includes users, with the user index set as follows. ,in, Represents any user in an ultra-dense MEC network. This represents the number of users in an ultra-dense MEC network; each user has There are various computational tasks, and the set of computational task indices is as follows: ,in, Indicates user A type of computational task, Indicates user Types of computational tasks;

[0065] Each type of computing task can be handled by the user or offloaded to edge servers in a super-dense MEC network. When a computing task is offloaded to an edge server, security measures are required, using different cryptographic algorithms for encryption and decryption. These cryptographic algorithms include... The set of cryptographic algorithm indices is [number]. ,in, Represents any cryptographic algorithm. Indicates the number of cryptographic algorithms;

[0066] The user's computing tasks are divided into There are 1 subtask, and the subtask index set is 1 ,in, This represents the user's subtask. Indicates the number of subtasks for a user; base station and user are based on Each subtask communicates via millimeter-wave spectrum, and uses the K-means algorithm to divide all users into groups based on the physical location of the base stations. A cluster;

[0067] The cluster where the user belongs is denoted as Different clusters use different millimeter-wave spectra, but base stations within the same cluster use the same millimeter-wave spectrum. The millimeter-wave spectrum is divided into... There are 1 identical millimeter-wave sub-channels, and the set of millimeter-wave sub-channel indices is: ,in, Represents any millimeter-wave sub-channel. This indicates the number of millimeter-wave sub-channels. Millimeter-wave sub-channels are used by users associated with base stations in the same cluster via non-orthogonal multiple access (NOMA), and each user can only be associated with one base station.

[0068] A network model is constructed based on base stations, users, computing tasks, cryptographic algorithms, subtasks, millimeter-wave sub-channels, combined with the cache and computing servers equipped in the base stations, as well as the K-means algorithm and millimeter-wave spectrum.

[0069] The specific process of constructing the communication model is as follows:

[0070] Because secure mobile edge computing and caching methods in ultra-dense millimeter-wave networks involve computational offloading, which typically occurs over large-scale time slots, slow fading channels must be used in the communication model instead of fast fading channels. To better represent path loss, a multi-slope channel model is adopted, which includes line-of-sight (LoS) and non-line-of-sight (NLoS) paths. Specifically;

[0071] Channel gain between base station and user have Path components; where The set of indices for each path component is ; For the nth base station and the user, ... The overall channel gain corresponding to each link; In a multi-slope channel model, the nth base station, the... Channel gain of the first propagation path under a multi-link configuration; In a multi-slope channel model, the nth base station, the... The channel gain of the i-th propagation path under the given links;

[0072] Distance representation between base station and user ; For the nth base station and the user, ... The distance between the links;

[0073] ,express:

[0074] (1);

[0075] In the formula, Let be the channel gain of the i-th path in a line-of-sight scenario; This refers to the path loss term in the line-of-sight scenario, representing the distance between the base station and the user. To meet the conditions; For the first The probability of a path; For line-of-sight propagation; Let be the channel gain of the i-th path in a non-line-of-sight scenario;

[0076] The path loss indices for line-of-sight and non-line-of-sight paths are respectively and ; is the line-of-sight path loss exponent corresponding to the i-th propagation path; is the non-line-of-sight path loss exponent corresponding to the i-th propagation path;

[0077] denotes:

[0078] (2);

[0079] wherein, denotes the distance threshold of the i-th path;

[0080] Since different clusters use different millimeter wave frequency spectrums, but the base stations in the same cluster use the same millimeter wave frequency spectrum, for any user, there is only intra-cluster interference, and there is no inter-cluster interference, so the uplink data rate formula of the user on the millimeter wave subchannel of the base station is:

[0081] (3);

[0082] wherein, is the uplink data rate corresponding to the user o, the millimeter wave subchannel s, and the base station n; is the bandwidth of the millimeter wave subchannel; is the base of the logarithmic operation; is the transmit power of the user o; is the set of base stations n that produce intra-cluster interference to the user o; is the power of the additive white Gaussian noise (noise variance); is the interfering user; is the channel gain between the user o and the interfering user ; is the millimeter wave subchannel index of the user o; is the identification of the millimeter wave subchannel or cluster to which the interfering user belongs; is the base station index of the user o; is the base station index of the interfering user ; is the cluster in which the base station is located;

[0083] Based on the path components between the base station and the user, the distance between the base station and the user, and the multi-slope channel model combining line-of-sight and non-line-of-sight paths, a communication model of the uplink data rate of the user on the millimeter wave subchannel of the base station is constructed.

[0084] wherein, the specific process of constructing the cache and calculation model is:

[0085] The calculation task of any user is represented as ​,in, Let y be the set of computational tasks y for user o; It is the data size of the user's computational task; It is the CPU cycles required for a user to complete a computing task; This is the maximum execution time of the user's computational task; It is the security level of the user's computing tasks;

[0086] According to the formula Calculate the time (latency) required for the user's computing task to be executed locally; Indicates the computing power of the user equipment; This indicates the offloading indicator for the user's computing tasks at the base station. This indicates that the program will be executed locally. This indicates that the process will be offloaded to the base station for execution.

[0087] According to the formula Calculate the energy consumption required for the user's computing tasks to be executed locally; It is the energy coefficient of the chip architecture; This means that a user's task can be executed on a maximum of one base station;

[0088] According to the formula Calculate users The task is calculated by uploading the time to the local or virtual base station through network operation mode (NOMO). The millimeter-wave sub-channel decision indicator selected by the user; This serves as a decision indicator for whether user o is associated with base station n; For base station n, user o, and task y, a cache decision indicator is provided. For virtual base stations; For virtual base stations , The cache decision indicator for user o's computation task y; For virtual base stations The unload decision indicator corresponding to the computation task y of user o;

[0089] in, This indicates that the user has selected the millimeter-wave sub-channel. , This indicates that the user did not select a millimeter-wave sub-channel. ; This indicates that the user is associated with the base station. This indicates that the user is not associated with the base station; This indicates that the user's computing tasks are cached at the base station. This indicates that the user's computing task is not cached at the base station;

[0090] According to the formula Calculate users The energy consumption of computing tasks uploaded to the base station or virtual base station for computing via network operation mode;

[0091] According to the formula Calculate users The task involves the backhaul time between the base station and the virtual base station, where, This indicates the freight rate for return transportation between the base station and the virtual base station;

[0092] According to the formula Calculate users The remote execution time of the task between the base station and the virtual base station; This means that a user's task can only be executed on one base station; This indicates the execution time of the user's task at the base station; The computational task y for user o is performed at the virtual base station. Execution time of edge computing (MEC) at the location;

[0093] According to the formula Computing base station Allocated to users based on their computing power percentage computational tasks computing resources ,in, Indicates base station Maximum computing power; Select cryptographic algorithm q to encrypt the computation task y for user o; For users computational tasks Decision variables can only be encrypted using one cryptographic algorithm; For set All users except o; For set Other computational tasks besides y; For base stations Other users Other computational tasks The corresponding uninstallation indicator; For other users Complete other computing tasks Required CPU cycles; For other users Other computational tasks The decision variable for choosing a particular cryptographic algorithm for encryption; For other users Other computing tasks Data size;

[0094] based on the time (latency) required to perform the computing task locally, the energy consumption required to perform the computing task locally, the time for the base station or virtual base station to perform the computation, the energy consumption for the base station or virtual base station to perform the computation, the backhaul time between the base station and the virtual base station, the remote execution time between the base station and the virtual base station, the computing task a cache and a computing model are constructed based on the computing resources.

[0095] The specific process of constructing the security model is as follows:

[0096] Before the user uploads the computing task to the base station for execution, the computing task needs to be encrypted first, and then the base station needs to decrypt the computing task before execution; the probability of failure when the user uses the encryption algorithm is , wherein is the probability of encryption failure when the user o uses the encryption algorithm u for the computing task y; represents the expected protection level of the task, represents the protection level of the cryptographic algorithm; is a natural constant;

[0097] The local encryption time of the user's computing task is ; is any one element in the set ; is the set of resource units in the user's local device that can participate in encryption calculation; is the computational complexity of the cryptographic algorithm corresponding to the computing task y of the user o;

[0098] The local encryption energy consumption of the user's computing task is ; the remote decryption time of the user's computing task is ;

[0099] The security vulnerability cost of the user's computing task is , wherein represents the financial loss caused by the failure of the user's computing task;

[0100] A security model is constructed based on the probability of failure when using the encryption algorithm, the local encryption time of the user's computing task, the local encryption energy consumption of the user's computing task, the remote decryption time of the user's computing task, and the security vulnerability cost of the user's computing task.

[0101] The specific process of constructing the optimization problem is as follows:

[0102] (4);

[0103] In the formula, is minimized; is the objective function; denotes the set of all users' association decision variables with the base stations; ; is the universal quantifier; denotes the set of all users' mmWave subchannel selection decision variables, ; denotes the set of users' decision variables whether to perform the computing task at the base station, ; denotes the set of users' decision variables whether to select the cryptographic algorithm for the task, , is the decision indicator whether the user o's computing task y selects the cryptographic algorithm u; denotes the set of all users' computing capabilities, ; denotes the set of all users' transmit powers, ; is the user o's local computing overhead; is subject to; is the constraint that the user's task processing latency does not exceed the maximum execution time of the computing task; is the user o's computing task y's processing latency; is the constraint that for each user o, the sum of the association indices with all base stations n is 1; is the constraint that for each user o, the sum of the selection of all mmWave subchannel s variables is 1; is the constraint that for each user o's each computing task y, the sum of the task execution variables over all base stations n does not exceed 1; is the constraint that for each user o's each computing task y, the sum of the selection of all cryptographic algorithm selection variables is 1; is the constraint that the user o's overall security vulnerability cost does not exceed the maximum allowed security vulnerability cost; is the user security vulnerability cost; is the maximum allowed user security vulnerability cost; is the constraint that the user's minimum and maximum allowed computing capabilities; denotes a positive number set to be small enough; is the user maximum allowed computing capability; is the constraint that the user o's transmit power lower and upper bounds; is the user transmit power upper bound; is the constraint that the user's association indices with the base stations are binary variables taking 0 or 1; is the constraint that the user's selection of mmWave subchannel variables are binary variables taking 0 or 1; The variable for the user to calculate the task executed at the base station is a binary variable, taking 0 or 1 as a constraint condition; The variable for the user to select the cryptographic algorithm is a binary variable, taking 0 or 1.

[0104] The specific process of outputting the position of the global optimal solution in step S2 is:

[0105] Step S2.1: Population initialization according to the constraint condition of the optimization problem, a set of initial solution set is randomly generated within the constraint range of population initialization; Tent chaotic mapping method is used to disturb and redistribute the initial solution set, and the initial population is constructed;

[0106] The maximum number of iterations of the improved grey wolf algorithm , and the current iteration number is set to 1;

[0107] Define a virtual user set ; ; For any specific virtual user in the virtual user set ; The total number of virtual users is

[0108] Step S2.2: the single grey wolf of the initial population is , the total number of grey wolves in the initial population is , ; Replace with ; The user associated with the base station corresponding to the grey wolf is an index set; Indicates the base station index when the user is associated with the grey wolf; The set of millimeter wave subchannels selected by the user corresponding to the grey wolf is , which indicates the millimeter wave subchannel index when the user is associated with the grey wolf; The set of base stations serving the virtual user corresponding to the grey wolf is , which indicates the base station index serving the virtual user when the user is associated with the grey wolf; The set of cryptographic algorithms selected by the virtual user corresponding to the grey wolf is , which indicates the encryption algorithm index selected by the virtual user when the user is associated with the grey wolf; The set of local computing capabilities of the user corresponding to the grey wolf is , which indicates the local computing capability when the user is associated with the grey wolf; The set of transmit powers of the user corresponding to the grey wolf is , which indicates the transmit power when the user is associated with the grey wolf;

[0109] Step S2.3: construct the grey wolf The fitness function is expressed as:

[0110] (5);

[0111] In the formula, For users The penalty factor; For users computational tasks The penalty factor; This represents the fitness function value;

[0112] Step S2.4: Randomly assign an initial value within the interval [0,1] of the initial population. By repeatedly applying this initial value, a chaotic sequence can be generated. Chaotic sequence Mapping to the solution space, in which the chaotic sequence is represented Converting to the position of the gray wolf in the gray wolf algorithm, the gray wolf position update process is represented as follows:

[0113] (6);

[0114] In the formula, For the Tent chaotic mapping method The value of the next iteration; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the user's side, corresponding to the first The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; In the t-th iteration, the v-th gray wolf is in the t-th iteration. For each user, the initial position of the parameter corresponding to the k-th decision variable; This represents the upper limit of the value of the k-th decision variable parameter; This represents the lower bound of the value of the parameter of the k-th decision variable; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the [number] user, corresponding to the [number]th [user] The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the [number] user, corresponding to the [number]th [user] The initial positions of the parameters of each decision variable; upper limit of the value of the th decision variable parameter; upper limit of the value of the th decision variable parameter; lower limit of the value of the th decision variable parameter; lower limit of the value of the th decision variable parameter; initial position of the th decision variable parameter corresponding to the th user in the th iteration; upper limit of the value of the th decision variable parameter; lower limit of the value of the th decision variable parameter; upper limit of the value of the th decision variable parameter; lower limit of the value of the th decision variable parameter; lower limit of the value of the th decision variable parameter; lower limit of the value of the th decision variable parameter; initial position of the th decision variable parameter corresponding to the th user in the th iteration; upper limit of the value of the th decision variable parameter; lower limit of the value of the th decision variable parameter; upper limit of the value of the th decision variable parameter; lower limit of the value of the th decision variable parameter; lower limit of the value of the th decision variable parameter; lower limit of the value of the th decision variable parameter;

[0115] calculating the fitness values of all grey wolves in the grey wolf population using the fitness function , , , , ,

[0116] If the current iteration number is less than or equal to the maximum iteration number , the current iteration number is less than or equal to the maximum iteration number , the search prey stage, the surround prey stage, the attack prey stage and the global optimal wolf updating operation stage are performed on the grey wolf population;

[0117] If the current iteration number is greater than the maximum iteration number , the global optimal wolf in the target population is output as the global optimal solution.

[0118] The specific process of the search prey stage is as follows:

[0119] In the search prey stage, the grey wolf population is classified according to the fitness values, wherein the grey wolves with the optimal fitness are divided into wolves, wolves,​​​​ Wolf; Wolf, Wolf, All wolves are alpha wolves; during the prey-hunting optimization phase, gray wolves within the pack update their own positions based on the alpha wolf's optimal position using the following formula:

[0120] (7);

[0121] In the formula, for The distance from the optimal position of the leader wolf; It is a random factor; For the vth gray wolf in the th In a scenario with one user, the optimal position of the leader wolf in the user-base station correlation index; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with a single user, the optimal position of the leader wolf in the millimeter-wave sub-channel index; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with one user, the optimal position of the leader wolf in the service base station set; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with multiple users, the optimal position of the leader wolf in the cryptographic algorithm index; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with a single user, the optimal position of the leader wolf in terms of local computing power; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with one user, the optimal position of the leader wolf in terms of transmit power; For the first In the iteration, the v-th wolf is in the ... In a scenario with one user, the updated position of the user-base station correlation index; The convergence factor; For the first In the iteration, the v-th wolf is in the ... In a scenario with one user, the updated position of the sub-channel index; For the first In the iteration, the v-th wolf is in the ... updated position of the service base station set for the vth wolf in the scenario of the ith user; updated position of the password algorithm index for the vth wolf in the scenario of the ith user; updated position of the local computing capability for the vth wolf in the scenario of the ith user; updated position of the transmit power for the vth wolf in the scenario of the ith user.

[0122] The specific process of the surrounding prey stage is as follows:

[0123] In the surrounding prey stage, the step length of the grey wolf is controlled by and ; wherein, ; is a nonlinear convergence factor; ; represents generating a first random number between 0 and 1; is a natural constant; when , it indicates that the grey wolf approaches the prey; when , it indicates that the grey wolf moves away from the prey;

[0124] wherein, , represents generating a second random number between 0 and 1; when , it indicates that the influence of the obstacle on the grey wolf is small; when , it indicates that the influence of the obstacle on the grey wolf is large;

[0125] When the grey wolf approaches the prey and the influence of the obstacle on the grey wolf is small, the attack prey stage is executed; when the grey wolf moves away from the prey and the influence of the obstacle on the grey wolf is large, the fitness value is recalculated.

[0126] The specific process of the attack prey stage is as follows:

[0127] Let represent , , ; let represent , , ; let represent , , ; and let represent​​​​​​ , , ;Will express , , ;Will express , , ;

[0128] The attack on prey phase indicates:

[0129] (8);

[0130] In the formula, For alpha wolf and exist Dimensional distance; It is the first independent random factor; For wolf α, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For β wolf and exist Dimensional distance; It is the second independent random factor; For wolf β, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For δ Wolf and exist Dimensional distance; It is the third independent random factor; For δ wolf, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For alpha wolf and exist Dimensional distance; For wolf α, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For β wolf and exist Dimensional distance; for In the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For δ Wolf and exist Dimensional distance; for In the t-th iteration, the v-th gray wolf targets the o-th user. position in the decision dimension; for alpha wolf and at distance in the dimension; for alpha wolf and position in the decision dimension; for beta wolf and at distance in the dimension; for beta wolf and position in the decision dimension; for beta wolf and at distance in the dimension; for beta wolf and position in the decision dimension; for beta wolf and at distance in the dimension; for beta wolf and position in the decision dimension; for beta wolf and at distance in the dimension; for beta wolf and position in the decision dimension; for beta wolf and at distance in the dimension; for beta wolf and position in the decision dimension; for beta wolf and at distance in the dimension; for beta wolf and position in the decision dimension; for beta wolf and at distance in the dimension; for beta wolf and position in the decision dimension; for beta wolf and at distance in the dimension; for beta wolf and position in the decision dimension; for beta wolf and For δ wolf, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For alpha wolves and Distance in the f-dimensional dimension; Let α be the position of the vth gray wolf targeting the oth user on the f decision dimension at the tth iteration. For β wolf and Distance in the f-dimensional dimension; Let β be the position of the vth gray wolf targeting the oth user on the f decision dimension during the t-th iteration. For δ Wolf and Distance in the f-dimensional dimension; Let δ be the position of the vth gray wolf targeting the oth user on the f decision dimension during the tth iteration. For the alpha wolf in the first In the next iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; This represents the step size factor corresponding to the α wolf; For wolf β, in the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; This is the step size factor corresponding to the β wolf; For δ wolf, in the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; This is the step size factor corresponding to the δ wolf; For wolf α, in the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration; For the position of the vth grey wolf for the oth user in the decision dimension at the t+1th iteration. Step S2.5.3: The specific process of the global optimal grey wolf updating operation stage is as follows:

[0131] The global optimal grey wolf updating operation stage is represented as:

[0132]

[0133] (9); ,

[0134] , , , , , , , as the global optimal solution.

[0135] The specific process of outputting the perturbed global optimal solution is as follows:

[0136] Step S3.1: After obtaining the global optimal solution, a stagnation judgment mechanism is introduced. When the global optimal solution does not change after 100 consecutive iterations, it is determined that the improved grey wolf algorithm is in stagnation, and the adaptive Gaussian Cauchy mutation strategy of the stagnation judgment mechanism is started to disturb the current global optimal solution, which is represented as:

[0137] (10);

[0138] wherein, is the disturbed global optimal position of the vth grey wolf in the scene of the oth user, user-base station association index; is the weight coefficient of the Cauchy distribution disturbance term; is a random number subject to Cauchy distribution; is the weight coefficient of the Gaussian distribution disturbance term; is a random number subject to Gaussian distribution; is the disturbed global optimal position of the vth grey wolf in the scene of the oth user, millimeter wave subchannel index; is the disturbed global optimal position of the vth grey wolf in the scene of the oth user, service base station set; is the disturbed global optimal position of the vth grey wolf in the scene of the oth user, cryptographic algorithm index; is the disturbed global optimal position of the vth grey wolf in the scene of the oth user, local computing capacity; is the disturbed global optimal position of the vth grey wolf in the scene of the oth user, transmit power;

[0139] wherein, , ; is gradually reduced from 1 to 0, and is gradually increased from 0 to 1; it is explained that in the early stage of iteration, the Cauchy distribution disturbance term plays a greater role, expanding the search range of the initial population, and in the later stage of iteration, the Cauchy distribution disturbance term gradually weakens, and the Gaussian distribution variation plays a major role, allowing the grey wolf to search in a small range, enhancing the local search ability of the initial population, and allowing the algorithm to quickly converge.

[0140] Step S3.2: the value of the current iteration number t of the improved grey wolf algorithm is increased by 1;

[0141] Step S3.3: after updating the value of t, the following are obtained: , , , , , as the global optimal solution after disturbance of the grey wolf; when the current iteration number t reaches the maximum iteration number, the global optimal solution after disturbance is output.​

[0142] In which, the effect of the embodiment of the present application can be further illustrated by simulation.

[0143] The simulation conditions are set as follows: the base stations and users are randomly distributed in a cellular network with a radius of 500m; 21 base stations and 20 users are considered, each user has 3 sub-tasks; the system bandwidth is 20MHz, 6 encryption algorithms; the maximum computing capacity of the user equipment is 2GHz, and the maximum transmission power of the user terminal is 23dBm.

[0144] Figure 2 The present application discloses a schematic diagram of the influence of the number of user iterations on the fitness value. The fitness of the grey wolf algorithm GWO and the improved grey wolf algorithm AGWO both increases with the increase of the number of user iterations, but the fitness value of the improved grey wolf algorithm AGWO is always higher than that of the grey wolf algorithm GWO.

[0145] Figure 3 The present application discloses a schematic diagram of the influence of the density of user task amount on energy consumption. The density of user task amount changes from 0.01 to 0.05, the total energy consumption of the grey wolf algorithm GWO and the improved grey wolf algorithm AGWO both increases with the increase of the user task amount, the total energy consumption of the improved grey wolf algorithm AGWO is always lower than that of the grey wolf algorithm GWO, achieving the goal of reducing energy consumption.

[0146] Figure 4 The present application discloses a schematic diagram of the influence of the density of user task amount on total cost. The density of user task amount changes from 0.01 to 0.05, the total energy consumption of the grey wolf algorithm GWO and the improved grey wolf algorithm AGWO both increases with the increase of the user task amount, the total cost of the improved grey wolf algorithm AGWO is always lower than that of the grey wolf algorithm GWO, achieving the goal of reducing cost.

[0147] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for secure mobile edge computing and caching in ultra-dense millimeter-wave networks, characterized in that, Includes the following steps: Step S1: Obtain basic network information of user terminal devices, base stations, channels, tasks and security services in the multi-access mobile edge computing network with cache unloading assistance to construct the network system; The network system includes a network model, a communication model, a caching and computing model, and a security model; an optimization problem is constructed under the constraints of the network system. Step S2: Introduce the Tent chaotic mapping method, nonlinear convergence factor, and adaptive Gaussian-Cauchy mutation strategy into the traditional gray wolf algorithm to form an improved gray wolf algorithm; use the improved gray wolf algorithm to search the optimization problem and output the global optimal solution; the search process includes the prey search stage, the prey surround stage, the prey attack stage, and the global optimal gray wolf update operation stage. Step S3: When the global optimal solution remains unchanged after 100 consecutive iterations of the improved Grey Wolf algorithm, the current global optimal solution is perturbed by the adaptive Gaussian-Cauchy mutation strategy with the stagnation judgment mechanism, and the perturbed global optimal solution is output. Step S4: Perform multi-access mobile edge computing network offloading optimization configuration based on the perturbed global optimal solution; Construct an optimization problem, representing: ； In the formula, To minimize; The objective function is... This represents the set of decision variables relating all users and base stations; ; It is a universal quantifier; This is a decision indicator for whether a user is associated with a base station; n is the base station; 'o' represents the base station index set; 'o' represents the user. For user index set; This represents the set of millimeter-wave sub-channel selection decision variables for all users. ; The millimeter-wave sub-channel decision indicator selected by the user; s represents the millimeter-wave sub-channel; For millimeter-wave sub-channel index set; The set of decision variables representing whether a user's computational task is executed at the base station. ; The offloading decision indicator for the user's computing tasks at the base station; For computational tasks; For the calculation task index set; This represents the set of decision variables for whether a user task should select a cryptographic algorithm. , A decision indicator for whether the user's computing task should select the cryptographic algorithm of the user's local device; u represents the cryptographic algorithm of the user's local device; A set of cryptographic algorithm indexes for the user's local device; This represents the set of computing power of all users. ; The computing power of user equipment; This represents the set of all users' transmit power. ; The transmit power of user o; Local computational overhead for user o; Limited by; The constraint is that the user's task processing latency does not exceed the maximum execution time of the computation task; The processing latency of user o's computation task y; The maximum allowed execution time for user o's computation task y; For each user o, the sum of the association exponents with all base stations n is 1; For each user o, select a constraint that the sum of all millimeter-wave sub-channel s variables is 1; For each computational task y of each user o, the sum of the task execution variables on all base stations n shall not exceed 1; For each computational task y of each user o, select a constraint that the sum of the selection variables of all cryptographic algorithms is 1; The total cost of security vulnerabilities for user o must not exceed the constraint of the maximum permissible cost of security vulnerabilities; Cost of user security vulnerabilities; To maximize the cost of allowing users to exploit security vulnerabilities; Constraints on the minimum and maximum allowed computing power for users; This indicates that the value should be a sufficiently small positive number. The maximum computing power allowed by the user; The constraints are the lower and upper limits of the transmit power for user o; This represents the upper limit of the user's transmit power. The correlation index between users and base stations is a binary variable, with a constraint that it can take the form of 0 or 1. The variable for selecting the millimeter-wave sub-channel for the user is a binary variable, with constraints that take the form of 0 or 1; The variables used to perform user computation tasks at the base station are binary variables, with constraints that take the form of 0 or 1. The variable used to select a cryptographic algorithm for a user is a binary variable, with constraints that allow it to take the form of 0 or 1.

2. The method for secure mobile edge computing and caching in ultra-dense millimeter-wave networks according to claim 1, characterized in that: The specific process of outputting the global optimal solution in step S2 is as follows: Population initialization is performed based on the constraints of the optimization problem. Within the constraints of population initialization, an initial solution set is randomly generated. The Tent chaotic mapping method is used to perturb and redistribute the initial solution set to construct the initial population. Maximum number of iterations of the improved Grey Wolf algorithm and the current iteration number Set to 1; Define virtual user set ; ; For virtual user set Any specific virtual user in the process; The total number of virtual users; The initial population of individual gray wolves was The initial population of gray wolves was , ;Will Replace with ; This is the set of user and base station related indices corresponding to the Grey Wolf; This represents the base station index when a user is associated with Grey Wolf; The set of millimeter-wave sub-channels selected for the users corresponding to Grey Wolf. This represents the millimeter-wave sub-channel index when a user is associated with the Grey Wolf. This refers to the set of base stations serving virtual users, corresponding to the Grey Wolf. This represents the base station index that serves virtual users when a user is associated with the Grey Wolf. The set of cryptographic algorithms selected for the virtual user corresponding to Grey Wolf. This represents the index of the encryption algorithm selected by the virtual user when the user is associated with Grey Wolf; This refers to the set of local computing capabilities for users corresponding to the Grey Wolf. This indicates the user's local computing power when associated with the Grey Wolf; This represents the set of transmit powers for the users corresponding to the Grey Wolf. This indicates the transmit power when the user is associated with the Grey Wolf; Gray wolves in the initial population The fitness function is expressed as: ； In the formula, For users The penalty factor; For users computational tasks The penalty factor; Represents the fitness function; Given a random initial value within the interval [0,1] of the initial population, a chaotic sequence is generated by repeatedly applying this initial value. Chaotic sequence Mapping to the solution space, in which the chaotic sequence is represented Converting to the position of the gray wolf in the gray wolf algorithm, the gray wolf position update process is represented as follows: ； In the formula, For the Tent chaotic mapping method The value of the next iteration; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the user's side, corresponding to the first The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; In the t-th iteration, the v-th gray wolf is in the t-th iteration. For each user, the initial position of the parameter corresponding to the k-th decision variable; This represents the upper limit of the value of the k-th decision variable parameter; This represents the lower bound of the value of the parameter of the k-th decision variable; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the [number] user, corresponding to the [number]th [user] The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the [number] user, corresponding to the [number]th [user] The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the [number] user, corresponding to the [number]th [user] The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; In the t-th iteration, the v-th gray wolf is in the t-th iteration. On the [number] user, corresponding to the [number]th [user] The initial positions of the parameters of each decision variable; For the first The upper limit of the values ​​of each decision variable parameter; For the first The lower bound of the values ​​of each decision variable parameter; Using the fitness function Calculate all gray wolves in the gray wolf population The fitness value; based on , , , , , The fitness values ​​are determined, and the gray wolf with the highest fitness value is selected. As the best gray wolf in the entire game; Determine the current iteration number Is it less than or equal to the maximum number of iterations? Current iteration number Less than or equal to the maximum number of iterations Then, the gray wolf population will undergo a prey search phase, a prey encirclement phase, a prey attack phase, and a global optimal gray wolf update phase. Current iteration number Greater than the maximum number of iterations Then, the globally best gray wolf in the target population is output as the global optimal solution.

3. The method for secure mobile edge computing and caching in ultra-dense millimeter-wave networks according to claim 2, characterized in that: The specific process of performing the prey-hunting phase is as follows: During the prey-hunting phase, the gray wolf population was classified according to its fitness value, with the gray wolves having the highest fitness being divided into... Wolf, Wolf, Wolf; Wolf, Wolf, All wolves are alpha wolves; during the prey-hunting optimization phase, gray wolves within the pack update their own positions based on the alpha wolf's optimal position using the following formula: ； In the formula, for The distance from the optimal position of the leader wolf; It is a random factor; For the vth gray wolf in the th In a scenario with one user, the optimal position of the leader wolf in the user-base station correlation index; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with a single user, the optimal position of the leader wolf in the millimeter-wave sub-channel index; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with one user, the optimal position of the leader wolf in the service base station set; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with multiple users, the optimal position of the leader wolf in the cryptographic algorithm index; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with a single user, the optimal position of the leader wolf in terms of local computing power; for The distance from the optimal position of the leader wolf; For the vth gray wolf in the th In a scenario with one user, the optimal position of the leader wolf in terms of transmit power; For the first In the iteration, the v-th wolf is in the ... In a scenario with one user, the updated position of the user-base station correlation index; The convergence factor; For the first In the iteration, the v-th wolf is in the ... In a scenario with one user, the updated position of the sub-channel index; For the first In the iteration, the v-th wolf is in the ... In a scenario with one user, the updated location of the service base station set; For the first In the iteration, the v-th wolf is in the ... In a scenario with multiple users, the updated position of the password algorithm index; For the first In the iteration, the v-th wolf is in the ... In a scenario involving individual users, the updated location of local computing power; For the first In the iteration, the v-th wolf is in the ... In a scenario with individual users, the updated position of the transmit power.

4. The method for secure mobile edge computing and caching in ultra-dense millimeter-wave networks according to claim 3, characterized in that: The specific process of encircling the prey is as follows: During the encirclement phase, utilize and To control the gray wolf's stride; among which, ; It is a nonlinear convergence factor; ; This indicates that a first random number between 0 and 1 will be generated; It is a natural constant; when This indicates that the gray wolf is approaching its prey; when This indicates that the gray wolf is keeping away from its prey; in, , This indicates generating a second random number between 0 and 1. This indicates that the obstacle has little impact on the gray wolf; when This indicates that the obstacle has a significant impact on the gray wolf; When the gray wolf approaches the prey and the obstacle has little effect on the gray wolf, it enters the prey attack phase; when the gray wolf moves away from the prey and the obstacle has a large effect on the gray wolf, the fitness value is recalculated.

5. The method for secure mobile edge computing and caching in ultra-dense millimeter-wave networks according to claim 4, characterized in that: The specific process of executing the attack on prey phase is as follows: Will express , , ;Will express , , ;Will express , , ;Will express , , ;Will express , , ;Will express , , ; The attack on prey phase indicates: ； In the formula, For alpha wolf and exist Dimensional distance; It is the first independent random factor; For wolf α, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For β wolf and exist Dimensional distance; It is the second independent random factor; For wolf β, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For δ Wolf and exist Dimensional distance; It is the third independent random factor; For δ wolf, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For alpha wolf and exist Dimensional distance; For wolf α, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For β wolf and exist Dimensional distance; for In the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For δ Wolf and exist Dimensional distance; for In the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For alpha wolf and exist Dimensional distance; For wolf α, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For β wolf and exist Dimensional distance; for In the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For δ Wolf and exist Dimensional distance; for In the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For alpha wolf and exist Dimensional distance; for In the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For β wolf and exist Dimensional distance; For wolf β, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For δ Wolf and exist Dimensional distance; For δ wolf, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For alpha wolf and exist Dimensional distance; For wolf α, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For β wolf and exist Dimensional distance; For wolf β, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For δ Wolf and exist Dimensional distance; For δ wolf, in the t-th iteration, the v-th gray wolf targets the o-th user. Position in the decision-making dimension; For alpha wolf and Distance in the f-dimensional dimension; Let α be the position of the vth gray wolf targeting the oth user on the f decision dimension at the tth iteration. For β wolf and Distance in the f-dimensional dimension; Let β be the position of the vth gray wolf targeting the oth user on the f decision dimension during the t-th iteration. For δ Wolf and Distance in the f-dimensional dimension; Let δ be the position of the vth gray wolf targeting the oth user on the f decision dimension during the tth iteration. For the alpha wolf in the first In the next iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; This represents the step size factor corresponding to the α wolf; For wolf β, in the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; This is the step size factor corresponding to the β wolf; For δ wolf, in the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; This is the step size factor corresponding to the δ wolf; For wolf α, in the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension; for In the (t+1)th iteration, the vth gray wolf targets the oth user. Position in the decision-making dimension.

6. The method for secure mobile edge computing and caching in ultra-dense millimeter-wave networks according to claim 5, characterized in that: The specific process of performing the global optimal gray wolf update operation phase is as follows: The update phase of the globally optimal gray wolf is represented as follows: ； Will , , , , , As the globally optimal solution.

7. The method for secure mobile edge computing and caching in ultra-dense millimeter-wave networks according to claim 6, characterized in that: The specific process of outputting the perturbed global optimal solution is as follows: After obtaining the global optimum, a stagnation detection mechanism is introduced. If the global optimum remains unchanged after 100 consecutive iterations, the improved Grey Wolf algorithm is considered to have stagnated. The adaptive Gaussian-Cauchy mutation strategy of the stagnation detection mechanism is then activated to perturb the current global optimum, as shown below: ； In the formula, For the v-th gray wolf in the scenario of the o-th user, the globally optimal position after perturbation of the user-base station correlation index; These are the weighting coefficients for the Cauchy distribution perturbation term; These are random numbers that follow a Cauchy distribution. These are the weighting coefficients for the Gaussian distribution perturbation term; These are random numbers that follow a Gaussian distribution. For the v-th gray wolf in the scenario of the o-th user, the globally optimal position of the millimeter-wave sub-channel index after perturbation; For the v-th gray wolf in the scenario of the o-th user, the globally optimal position of the service base station set after perturbation; For the v-th gray wolf in the scenario of the o-th user, the globally optimal position of the cryptographic algorithm index after perturbation; For the v-th gray wolf in the scenario of the o-th user, the globally optimal position after perturbation of local computing power; Let v be the globally optimal position of the v-th gray wolf after perturbation of the transmit power in the scenario of the o-th user; The current iteration number of the improved gray wolf algorithm Increment the value by 1; After the t-value is updated, , , , , , As the globally optimal solution after the gray wolf perturbation; When the current iteration number t reaches the maximum iteration number, output the perturbed global optimal solution.

Images