BC-assisted system energy efficiency optimization method in MEC network

Abstract

This invention discloses a method for optimizing system energy efficiency in a BC-assisted MEC network, primarily addressing the existing issues of insufficient energy supply and limited computing power in IoT nodes. The solution includes: 1) establishing a BC-assisted MEC network system model; 2) users offloading computing tasks to multiple edge servers using time-sharing access, respectively, via backscattering and active transmission; 3) maximizing the system energy efficiency of the BC-assisted MEC network by identifying active and passive offloading times, the MEC server's computing frequency and time, the user's transmit power and offloading decision variables, and the BC coefficient; and 4) setting system operating parameters based on the optimal results for optimization. This invention can effectively improve user computing efficiency, optimize network performance, and is applicable to edge computing systems.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communication technology for the Internet of Things, and further relates to edge computing technology. Specifically, it provides a method for optimizing system energy efficiency in a mobile edge computing (MEC) network assisted by backscatter communication (BC), which can be used for multi-user offloading tasks in edge computing systems. Background Art

[0002] The Internet of Things (IoT) aims to interconnect everything. Its basic architecture consists of three layers: the perception layer, the network layer, and the application layer. Physical sensors in the perception layer collect useful information from the environment, convert it into digital form, and assign unique addresses to all objects. The network layer effectively connects the application layer and the perception layer. The IoT provides personalized services at the application layer based on user needs, offering high-level intelligent solutions for applications in various fields, such as disaster monitoring, smart homes, connected vehicles, healthcare, production control, and education. An increasing number of computationally intensive and latency-sensitive applications are being deployed on various IoT devices. However, due to the small size and low cost of IoT nodes, their energy reserves and computing power are very limited, hindering their widespread adoption. Due to the size and cost constraints of IoT nodes, they cannot be equipped with large-capacity batteries or powerful CPU processors. Therefore, in the era of fifth-generation mobile communication technology (5G), insufficient energy supply and limited computing power of IoT nodes are two major challenges for the development of the IoT.

[0003] Energy Harvesting (EH) and Wireless Power Transfer (WPT) are two wireless charging technologies that have received widespread attention in recent years. EH technology allows devices to absorb energy in space such as solar energy, wind energy, and electromagnetic energy, thereby extending the operating time of the device and the operating life of the network. However, the energy acquisition process of devices with EH technology is discontinuous, uncontrollable, and random. WPT technology can more stably electromagnetically charge devices by deploying dedicated energy stations in the network and sending radio frequency (RF) signals. WPT technology uses a half-duplex transmission method to charge the device first. After the device power reaches a certain level, the harvested energy is used to forward tasks. The commonly used protocol is the Harvest-Then-Transmit (HTT) protocol, and the HTT protocol is also called active transmission.

[0004] Backscatter Communication (BC) technology is a passive transmission technology with ultra-low or zero power consumption for device nodes. Devices using BC reflect received RF signals by adjusting antenna impedance, with some of the RF signal's energy reflected and some absorbed by the device. During the reflection phase, the device transmits information by modulating its own transmit signal onto the RF signal. After reflection, the absorbed energy can be used for active transmission. Compared to EH and WPT technologies, BC is more stable and energy-efficient, making it well-suited for IoT applications.

[0005] Mobile Edge Computing (MEC) is a computing-assisted technology that deploys edge servers at the edge of the network, directly connecting them to devices and reducing transmission latency and energy consumption. Devices offload computing tasks to MEC servers via wireless links. The MEC servers then return the results, effectively reducing user latency and improving computing power.

[0006] Zargari S, Tellambura C, Herath S. et al., in their published paper “Energy-efficient hybrid offloading for backscatter-assisted wirelessly powered MEC with reconfigurable intelligent surfaces.” (IEEE Transactions on Mobile Computing, 2022: 1-1), consider a system model with one energy station, one MEC server, one IRS, and multiple wireless devices, and optimize the BC time, active transmission time, BC coefficient, transmit power of the energy station and wireless device, local computing time and computing frequency, and phase shift of the IRS to maximize the energy efficiency of the entire system. Xu S, Du Y, Liu J, et al., in their published paper “Intelligent reflecting surface based backscatter communication for data offloading.” (IEEE Transactions on Communications, 2022, 70(6): 4211-4221.), consider one energy station, one wireless device, and multiple MEC servers. The IRS acts as a wireless device with beamforming capabilities, maximizing the total computational bits used in the wireless device's beamforming and local computation phases by optimizing the energy station's beamforming vector, the IRS's phase shift vector, the time allocation for the EH and BC phases, and the local computation time. Relatively little work has been conducted on the integration of beamforming, active transmission, and MEC technologies, and existing work has only considered scenarios involving a single MEC server. However, in real systems, multiple MEC servers may be deployed, making these approaches difficult to address. Summary of the Invention

[0007] The present invention aims to address the shortcomings of the above-mentioned existing technologies and propose a BC-assisted solution for maximizing system energy efficiency in MEC networks. By maximizing system energy efficiency, the solution identifies active and passive offloading times, the computation frequency and computation time of the MEC server, the user's transmit power and offloading decision variables, and the BC coefficient. This solution solves the problems of insufficient energy supply and limited computing power of IoT nodes, effectively improving user computing efficiency.

[0008] The basic concept behind this invention is as follows: First, the energy station sends an energy-carrying wireless signal to each user. All users offload their tasks to the MEC server via the BC transmission circuit using a time-determined multiple access (TDMA) scheme within their assigned time slots, allowing all time slots to absorb energy. Subsequently, the users take turns actively transmitting using the collected energy, offloading their tasks to the MEC server. Finally, assuming that the user tasks have been offloaded during the BC and active transmission phases, all MEC servers execute the received user tasks in parallel.

[0009] To achieve the above object, the technical solution of the present invention includes the following steps:

[0010] (1) Build a communication network consisting of an energy station, K users, and M MEC servers, and let Ω K ={1,2,…,K} represents the user set, Ω M = {1, 2, …, M} represents the set of MEC servers. Each user includes an energy absorption module, a signal transceiver module, and a backscatter communication (BC) circuit. Each user is equipped with an antenna that has a conventional fixed impedance mode and multiple impedance modes for BC, and can switch between the two modes.

[0011] (2) The energy station sends a wireless signal carrying energy to each user. All users use time division multiple access (TDMA) and offload user tasks to the MEC server through the BC circuit within their allocated time slots to obtain the energy collected by the kth user. and the rate of offloading via BC

[0012] (3) After the energy station stops sending energy, users take turns to actively transmit using the collected energy, offloading user tasks to the MEC server; the offloading rate of the kth user's active transmission is obtained Energy consumed by active transmission and

[0013] (4) Assuming that the user task has been offloaded, all MEC servers execute the received user tasks in parallel, let f m,k represents the computing frequency provided by the m-th MEC server for the k-th user, c k represents the task complexity, then the calculation rate of the m-th MEC server calculating the k-th user upload task is Get the energy consumption E of the mth MEC server to calculate the kth user offloading task m,k , further calculate the total amount of tasks R that users uninstall sum And the total energy consumption of the system E sum ;

[0014] (5) Constructing the optimal task computing efficiency E max expression:

[0015]

[0016] Among them, the optimization variable is BC unloading time Active transmission time BC coefficient User's transmit power Unloading decision variables Computation frequency of MEC server and execution time t M ;

[0017] Set the constraints as follows:

[0018] Constraints on the receive power partitioning factor; represents the computing task constraint of the kth user, where Indicates the amount of tasks that the kth user needs to complete; 0≤p k ≤p max , represents the transmission power constraint of the kth user and the computation frequency constraint of the MEC server, where f max and P max The maximum available CPU frequency of MEC and the maximum transmit power of the user are respectively; represents the computing task constraints of the MEC server, where L k is the task amount of the kth user; represents the energy consumption constraint of the kth user; represents the working time constraint, T is the total time; represents the uninstall decision variable constraint of the kth user;

[0019] (6) E is converted into max The expression is converted into a convex expression, and then the Lagrange dual algorithm and subgradient algorithm are used to solve the convex expression to obtain the optimal task calculation efficiency E max and its corresponding optimal optimization variables, which include the optimal BC time (t b ) * , optimal active transmission time (t a ) * , optimal BC coefficient β * , optimal user transmit power p * , optimal MEC CPU frequency f * and optimal MEC task computing time and the optimal unloading variable α * ;

[0020] (7) The system selects operating parameters based on the optimal optimization variables to maximize the system energy efficiency.

[0021] Compared with the prior art, the present invention has the following advantages:

[0022] First, because the present invention utilizes wireless power transmission (WPT) technology, by deploying dedicated energy stations within the network to transmit radio frequency signals, it provides more stable electromagnetic charging for devices. WPT technology uses half-duplex transmission, first charging the device. Once the device's power reaches a certain level, it then uses the harvested energy to forward tasks, using the Harvest-Then-Transmit (HTT) protocol to provide energy to IoT users. Furthermore, using backscatter communication (BC) technology, the device adjusts the antenna impedance to reflect the received RF signal, thereby solving the problem of insufficient energy supply for IoT nodes.

[0023] Second, in real-world scenarios, network systems often face massive user volumes, and a single MEC server cannot provide sufficient computing power, making it difficult to solve practical problems. Compared to existing methods, this invention introduces backscatter communication technology and considers multi-user, multi-MEC server scenarios. In this scenario, it optimizes system performance, solves the problem of limited computing power, and effectively improves user computing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 Schematic diagram of an application scenario of the method of the present invention;

[0025] Figure 2 Schematic diagram of the time slot structure of the present invention;

[0026] Figure 3 Flowchart for realizing the method of the present invention;

[0027] Figure 4 The calculation time t of the MEC server in the method of the present invention is M Simulation results of the impact on energy efficiency;

[0028] Figure 5 is the user task amount L in the method of the present invention k Simulation results of the impact of time slot length T on energy efficiency;

[0029] Figure 6 is the task complexity c in the method of the present invention k Simulation results of the impact of system bandwidth B on energy efficiency;

[0030] Figure 7 This is a simulation result diagram of the impact of path loss on energy efficiency in the method of the present invention; DETAILED DESCRIPTION

[0031] The following describes in detail the implementation process of the technical solution of the present invention with reference to the accompanying drawings:

[0032] Example 1: Reference Figure 3 The present invention provides a BC-assisted method for optimizing system energy efficiency in MEC networks. The specific implementation steps are as follows:

[0033] Step 1: Reference Figure 1 , build a communication network consisting of an energy station, K users and M MEC servers, let Ω K ={1,2,…,K} represents the user set, Ω M = {1, 2, …, M} represents the set of MEC servers. Each user includes an energy absorption module, a signal transceiver module, and a backscatter communication (BC) circuit. Each user is equipped with an antenna that has a conventional fixed impedance mode and multiple impedance modes for BC, and can switch between the two modes.

[0034] Reference Figure 2 The time slot structure of the present invention includes four stages. The first stage represents BC and charging. The energy station sends a wireless signal carrying energy to each user. All users use TDMA to offload user tasks to the MEC server through the BC transmission circuit within their respective allocated time slots. All time slots can absorb energy. represents the time when the kth user passively offloads the task to the MEC server through the BC circuit. The second stage represents active transmission, the energy station stops sending energy, and the users take turns actively transmitting using the collected energy to offload the task to the MEC server, where, It represents the time when the kth user actively offloads the task to the MEC server. The third stage represents task processing. Assuming that the user task has been offloaded in the BC and active transmission stages, all MEC servers execute the received user tasks in parallel for a time of t M The fourth stage represents the result transmission. Considering that the calculation result is relatively small, the time it takes for the MEC server to transmit the calculation result to the user can be ignored.

[0035] Step 2: The energy station sends a wireless signal carrying energy to each user. All users use time division multiple access (TDMA) to offload user tasks to the MEC server through the BC circuit within their allocated time slots. All time slots can absorb energy. The energy collected by the kth user is calculated according to the following formula: and the rate of offloading via BC

[0036]

[0037]

[0038] Among them, η k represents the energy absorption efficiency of the kth user, α k,m ∈{0,1} is the unloading variable, α k,m =1 means that the kth user offloads data to the mth MEC server, α k,m =0 means that the kth user does not offload data to the mth MEC server; B k,m represents the bandwidth from the kth user to the mth MEC server, β k represents the BC coefficient, P B Indicates the sending power of the energy station, represents the channel power gain between the energy station and the kth user, represents the channel power gain from the kth user to the mth MEC server, represents the noise power of the MEC server; represents the total time for k users to offload tasks to the MEC server through the BC circuit; It represents the time when the kth user offloads the task to the MEC server through the BC circuit.

[0039] Step 3: After the energy station stops sending energy, users take turns to actively transmit using the collected energy, offloading user tasks to the MEC server; the offloading rate of the kth user's active transmission is obtained Energy consumed by active transmission and The offloading rate of the kth user's active transmission and the energy consumed by active transmission They are calculated as follows:

[0040]

[0041]

[0042] Among them, p k represents the transmission power of the kth user; It represents the time when the kth user actively offloads the task to the MEC server.

[0043] Step 4: Assume that the user task has been offloaded, that is, the user has completed the offloading in the BC and active transmission phases of steps 2 and 3, and all MEC servers execute the received user tasks in parallel. Let f m,k represents the computing frequency provided by the m-th MEC server for the k-th user, c krepresents the task complexity, then the calculation rate of the m-th MEC server calculating the k-th user upload task is Get the energy consumption E of the mth MEC server to calculate the kth user offloading task m,k , further calculate the total amount of tasks R that users uninstall sum And the total energy consumption of the system E sum ;

[0044] The mth MEC server calculates the energy consumption E of the kth user offloading task m,k It is calculated as follows:

[0045]

[0046] Among them, γ m represents the effective capacitance coefficient of the mth MEC server CPU, t M Indicates the time the MEC server takes to calculate user tasks.

[0047] The calculation obtains the total amount of tasks R that the user has uninstalled sum And the total energy consumption of the system E sum , the formula is as follows:

[0048]

[0049]

[0050] Among them, ω is the weight coefficient of system energy consumption.

[0051] Step 5: Construct the optimal task computing efficiency E max expression:

[0052]

[0053] Among them, the optimization variable is BC unloading time Active transmission time BC coefficient User's transmit power Unloading decision variables Computation frequency of MEC server and execution time t M ;

[0054] Set the constraints as follows:

[0055] Constraints on the received power partitioning factor; represents the computing task constraint of the kth user, where Indicates the amount of tasks that the kth user needs to complete; 0≤p k ≤p max , represents the transmission power constraint of the kth user and the computation frequency constraint of the MEC server, where f max and P max The maximum available CPU frequency of MEC and the maximum transmit power of the user are respectively; represents the computing task constraints of the MEC server, where L k is the task amount of the kth user; represents the energy consumption constraint of the kth user; represents the working time constraint, T is the total time; represents the uninstall decision variable constraint of the kth user;

[0056] Step 6: Use variable substitution, BCD, and Dinkelbach algorithm to convert E max The expression is converted into a convex expression, and then the Lagrange dual decomposition algorithm and subgradient algorithm are used to solve the convex expression to obtain the optimal task computation efficiency E max and its corresponding optimal optimization variables, which include the optimal BC time (t b ) * , optimal active transmission time (t a ) * , optimal BC coefficient β * , optimal user transmit power p * , optimal MEC CPU frequency f * and optimal MEC task computing time and the optimal unloading variable α * .

[0057] Step 7: The system selects operating parameters based on the optimal optimization variables to optimize system performance.

[0058] Example 2: The overall implementation steps of this embodiment are the same as those of Example 1. In step 6, E is replaced by variable substitution, BCD, and Dinkelbach algorithm. max The process of converting the expression into a convex expression and then solving the convex expression using the Lagrange dual decomposition algorithm and the subgradient algorithm is further described. The process includes the following steps:

[0059] (6.1) Since the decision variable constraint is an integer constraint, it is a non-convex constraint. First, k,m Relax and set α k,m ∈{0,1} is relaxed to 0≤α k,m ≤1. At the same time, since the objective function is in fractional form and the constraint variables are coupled, Formula <1.1> is a non-convex problem. To solve it, the Dinkelbach algorithm is first used to remove the fractional structure in the objective function and transform the objective function into:

[0060]

[0061] Among them, δ>0 is an introduced variable, and its initial value δ=δ0 is specified;

[0062] (6.2) Since the variable and β k 、 With p k Coupling, for this, introduce variables With variables This eliminates variable coupling and updates the constraints. Formula <1.1> becomes:

[0063]

[0064] constraint became Constraint 0≤p k ≤p max , became constraint became constraint became

[0065] (6.3) Use the BCD algorithm to solve formula <1.2>.

[0066] (6.4) Fix α and use the Lagrange dual method to solve for t b , t a ,m,q,f, use one-dimensional search algorithm to update t M Fix α and change formula <1.2> to:

[0067]

[0068] (6.4.1) Formula <1.3> is a convex problem and is solved using the Lagrangian dual method. The Lagrangian function of Formula <1.3> is:

[0069]

[0070] Among them, μ, κ, ξ, are dual variables, and The Lagrange dual function of formula <1.3> is:

[0071]

[0072] The constraints are

[0073] (6.4.2) Use the dual decomposition method to decompose formula <1.4> into the following sub-formulas, as follows:

[0074] First, for variable f m,k The corresponding sub-formula is:

[0075]

[0076] The constraints are Formula <1.4.1> is a convex problem. Using the KKT condition, the optimal solution can be obtained as:

[0077]

[0078] for variable m k , q k , t M The corresponding sub-formula is:

[0079]

[0080] The constraints are Formula <1.4.2> is a convex problem. The optimal solution obtained by using convex optimization tools is:

[0081]

[0082] (6.4.3) After solving each sub-formula, we need to solve the dual problem of formula <1.4>, which is:

[0083] min g(μ,κ,ξ) <1.5>

[0084] The constraints are Since the optimal values ​​of formula <1.5> and formula <1.3> are the same when the Slater condition is satisfied, the subgradient method can be used to solve formula <1.5> and optimize the dual variable. k , κ k and ξ k The subgradient update expression is as follows:

[0085]

[0086]

[0087]

[0088] in, and Both represent the iterative update step size.

[0089] (6.5) Order and is the optimal value of the above solution, the offloading decision variable α of the k-th user offloading to the m-th MEC server k,m The optimization problem can be transformed into:

[0090]

[0091] The constraints are Formula <1.6> is a convex problem, which can be solved with the help of the convex optimization toolkit.

[0092] (6.6) When δ is given, after formula <1.2> is solved, the value of δ is updated:

[0093]

[0094] (6.7) Repeat steps (6.1)-(6.6) until the algorithm converges and the optimal energy efficiency δ is obtained. * .

[0095] The effect of the present invention is further described below in conjunction with simulation experiments:

[0096] A. Simulation Conditions

[0097] Computer simulation software is used for simulation. Unless otherwise specified, the channel power gain settings adopted by the present invention are as follows: Assume that all channels between points x and y are Generate, where ρ0 = -30dB is the path loss at reference distance d = 1m, d x,y Represents the distance between two points, α x,y represents the path loss coefficient, g x,y The specific simulation parameters are summarized in Table 1.

[0098] Table 1 Simulation parameters

[0099]

[0100]

[0101] B. Simulation Content

[0102] Simulation 1: MEC server computing time t M The impact on energy efficiency, the simulation results are as follows Figure 4 As shown;

[0103] Simulation 2: User task volume L k The impact of time slot T on energy efficiency, the simulation results are as follows Figure 5 As shown;

[0104] Simulation 3: Task complexity c k The impact of system bandwidth B on energy efficiency, the simulation results are as follows Figure 6 As shown;

[0105] Simulation 4: The impact of path loss on energy efficiency. The simulation results are as follows: Figure 7 As shown;

[0106] C. Simulation Results

[0107] Depend on Figure 4 It can be seen that t M Perform one-dimensional search, as the calculation time t of MEC server M As t increases, the energy efficiency first increases and then decreases, and there is an optimal value A, so the optimal t can be effectively obtained through one-dimensional search. M , when t M When it is equal to 0.7, the energy efficiency is maximum.

[0108] Depend on Figure 5 It can be seen that as the user task volume L k As the time slot length T increases, the energy efficiency decreases. There are two specific reasons. When the user task volume L k Increased time slot lengths (T) increase the number of tasks users must complete. Completing tasks within a specified timeframe requires increasing the computational speed, which in turn increases the user's task offload power and the CPU's computational frequency. This increases energy consumption and, consequently, energy efficiency. When the time slot length (T) increases, users have more time to send tasks, gaining greater temporal freedom and a larger decision space, enabling them to find better parameter settings and achieve higher energy efficiency.

[0109] Depend on Figure 6 It can be seen that as the task complexity c k As the bandwidth B between the user and the MEC server increases, the energy efficiency decreases. The specific reasons are divided into two aspects. When the task complexity c k When bandwidth B increases, the MEC server's computing power decreases, requiring more CPU cycles to complete task calculations. To complete tasks within the specified time, the MEC server requires a higher computing frequency, resulting in higher energy consumption and lower energy efficiency. When bandwidth B increases, according to Shannon's equation, users require less power to complete task offloads, resulting in lower energy consumption and, in turn, higher system energy efficiency.

[0110] Depend on Figure 7As can be seen, v1 represents the path loss between the energy station and the user, and v2 represents the path loss between the user and the MEC server. Increases in either v1 or v2 result in reduced energy efficiency. The specific reasons are as follows: an increase in v1 and v2 indicates a gradual deterioration of the channel, with more severe channel fading. The energy station requires more energy for recharging, and the user also requires more energy for task offloading. As the path loss v1 between the energy station and the user increases, channel fading becomes more severe, and the power received by the user decreases, resulting in a decrease in the passive offloading rate used for BC. Simultaneously, the energy collected by the user for active transmission also decreases, and the rate of active offloading also decreases, leading to a decrease in energy efficiency. When the path loss v2 between the user and the MEC server increases, channel fading becomes more severe, and the user's offloading rate decreases. At the same time, the MEC server receives fewer tasks, leading to a decrease in energy efficiency.

[0111] The above simulation analysis proves the correctness and effectiveness of the method proposed in the present invention.

[0112] Parts of the present invention that are not described in detail belong to common knowledge among those skilled in the art.

[0113] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Obviously, for professionals in this field, after understanding the content and principles of the present invention, they may make various modifications and changes in form and details without departing from the principles and structure of the present invention. However, these modifications and changes based on the ideas of the present invention are still within the scope of protection of the claims of the present invention.

Claims

1. A method for maximizing system energy efficiency in a BC-assisted MEC network, characterized in that: The steps include: (1) Build a communication network consisting of an energy station, K users, and M MEC servers, and let Ω K ={1,2,…,K} represents the user set, Ω M = {1, 2, …, M} represents the set of MEC servers. Each user includes an energy absorption module, a signal transceiver module, and a backscatter communication (BC) circuit. Each user is equipped with an antenna that has a conventional fixed impedance mode and multiple impedance modes for BC, and can switch between the two modes. (2) The energy station sends a wireless signal carrying energy to each user. All users use time division multiple access (TDMA) and offload user tasks to the MEC server through the BC circuit within their allocated time slots to obtain the energy collected by the kth user. and the rate of offloading via BC (3) After the energy station stops sending energy, users take turns to actively transmit using the collected energy, offloading user tasks to the MEC server; the offloading rate of the kth user's active transmission is obtained Energy consumed by active transmission and (4) Assuming that the user task has been offloaded, all MEC servers execute the received user tasks in parallel, let f m,k represents the computing frequency provided by the m-th MEC server for the k-th user, c k represents the task complexity, then the calculation rate of the m-th MEC server calculating the k-th user upload task is Get the energy consumption E of the mth MEC server to calculate the kth user offloading task m,k , further calculate the total amount of tasks R that users uninstall sum And the total energy consumption of the system E sum ; (5) Constructing the optimal task computing efficiency E max expression: Among them, the optimization variable is BC unloading time Active transmission time BC coefficient User's transmit power Unloading decision variables Computation frequency of MEC server and execution time t M ; Set the constraints as follows: Constraints on the receive power partitioning factor; represents the computing task constraint of the kth user, where Indicates the amount of tasks that the kth user needs to complete; represents the transmission power constraint of the kth user and the computation frequency constraint of the MEC server, where f max and P max The maximum available CPU frequency of MEC and the maximum transmit power of the user are respectively; represents the computing task constraints of the MEC server, where L k is the task amount of the kth user; t M Indicates the time it takes for the MEC server to calculate the user task; represents the energy consumption constraint of the kth user; Represents working time constraints, T is the total time; P B Indicates the sending power of the energy station, represents the channel power gain between the energy station and the kth user; η k represents the energy absorption efficiency of the kth user; represents the total time for k users to offload tasks to the MEC server through the BC circuit; represents the k-th user's uninstall decision variable constraint; where α k,m ∈{0,1} is the uninstall variable; (6) E is converted into max The expression is converted into a convex expression, and then the Lagrange dual decomposition algorithm and subgradient algorithm are used to solve the convex expression to obtain the optimal task computation efficiency E max and its corresponding optimal optimization variables, which include the optimal BC time (t b ) * , optimal active transmission time (t a ) * , optimal BC coefficient β * , optimal user transmit power p * , optimal MEC CPU frequency f * and optimal MEC task computing time and the optimal unloading variable α * ; (7) The system selects operating parameters based on the optimal optimization variables to maximize the system energy efficiency.

2. The method according to claim 1, wherein: The energy collected by the kth user in step (2) and the rate of offloading via BC They are calculated as follows: Among them, η k represents the energy absorption efficiency of the kth user, α k,m ∈{0,1} is the unloading variable, α k,m =1 means that the kth user offloads data to the mth MEC server, α k,m =0 means that the kth user does not offload data to the mth MEC server; B k,m represents the bandwidth from the kth user to the mth MEC server, β k represents the BC coefficient, P B Indicates the sending power of the energy station, represents the channel power gain between the energy station and the kth user, represents the channel power gain from the kth user to the mth MEC server, represents the noise power of the MEC server; represents the total time for k users to offload tasks to the MEC server through the BC circuit; It represents the time when the kth user offloads the task to the MEC server through the BC circuit.

3. The method according to claim 2, wherein: The unloading rate of the kth user's active transmission in step (3) and the energy consumed by active transmission They are calculated as follows: Among them, p k represents the transmission power of the kth user; It represents the time when the kth user actively offloads the task to the MEC server.

4. The method according to claim 1, wherein: The mth MEC server in step (4) calculates the energy consumption E of the kth user offloading task m,k It is calculated as follows: Among them, γ m represents the effective capacitance coefficient of the mth MEC server CPU, t M Indicates the time the MEC server takes to calculate user tasks.

5. The method according to claim 4, characterized in that: The total amount of tasks R that the user has uninstalled is calculated in step (4). sum And the total energy consumption of the system E sum , the formula is as follows: Among them, ω is the weight coefficient of system energy consumption.

Images