Energy efficiency optimization method of D2D auxiliary wireless energy supply mobile edge computing system
Through the energy efficiency optimization method of D2D assisted wireless energy-supply mobile edge computing system, the nonlinear energy harvesting model and Lyapunov optimization technology are used to solve the problem of insufficient energy supply and computing power in the mobile edge computing system, and the maximum system energy efficiency is achieved and stability, suitable for multi-user environments.
Patent Information
- Application Number
- CN202411497738.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-25
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing mobile edge computing system is difficult to effectively integrate D2D communication, wireless power supply and mobile edge computing technologies in complex multi-user environments, resulting in limited energy supply, insufficient computing power, and signal transmission blockage, and long-term performance has not been fully considered.
The energy efficiency optimization method of D2D assisted wireless power supply mobile edge computing system is adopted, based on the nonlinear energy harvesting model, and directly transmit data between users through D2D communication technology. The system energy efficiency optimization problem is decomposed using the Liyapunov optimization and punishment dual decomposition method. Solve subproblems to maximize system energy efficiency.
It effectively reduces the data load of the core network of the communication system, improves the system's energy efficiency, ensures the stability of the system in a dynamic network environment and the fairness of resource allocation, and is suitable for mobile devices with energy limitations.
Smart Images

Figure CN119012274A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of wireless communication technologies, and in particular to an energy efficiency optimization method for a D2D assisted wireless powered mobile edge computing system. Background Art
[0002] With the development of emerging technologies such as the Internet of Things, smart cities, and driverless cars, the number and types of mobile devices have exploded. These devices are usually limited in computing power and energy, so efficient computing resource allocation and energy management strategies are needed to meet their growing needs. Mobile edge computing (MEC), as an emerging computing architecture, can significantly reduce latency, improve computing efficiency, and meet the needs of real-time applications by sinking computing resources from the remote cloud to the edge of the network. However, MEC systems still face challenges such as limited energy supply, insufficient computing power, and signal transmission congestion.
[0003] To solve these problems, wireless power transmission (WPT) and device-to-device (D2D) communication have gradually become the focus of research. Wireless power supply technology allows devices to obtain energy from base stations or other devices through wireless signals, extending their working time and reducing their dependence on batteries. At the same time, D2D communication allows devices to communicate directly without relaying through base stations, which not only reduces communication delays, but also reduces the load on the core network and improves spectrum utilization.
[0004] Although MEC, WPT and D2D technologies each have significant advantages, how to effectively integrate these technologies in a complex multi-user environment to maximize system energy efficiency and improve communication performance is still a difficult problem to be solved. Therefore, proposing an energy efficiency optimization algorithm that can integrate D2D communication, wireless power supply and mobile edge computing technology to meet the computing and communication needs of low-power devices has become an important research direction in the current field of wireless communication technology.
[0005] In addition, the research on the mobile edge computing environment of wireless power supply only considers the short-term performance of the system, and often ignores the long-term performance; and the energy collection model of the existing wireless power supply MEC system mostly adopts the linear energy collection model, but the working efficiency of the energy collector is not linear during the wireless energy transmission process. Therefore, it is urgent to use the nonlinear energy collection model in the energy efficiency optimization algorithm to better capture the actual working characteristics of the energy collector, quickly and accurately obtain the optimal solution, optimize the system energy efficiency, and save energy. Summary of the invention
[0006] The purpose of the present invention is to provide an energy efficiency optimization method for a D2D-assisted wireless power supply mobile edge computing system, which can reduce the data load of the core network of the communication system and effectively improve the system energy efficiency.
[0007] To achieve the above object, the present invention provides an energy efficiency optimization method for a D2D-assisted wireless power supply mobile edge computing system, comprising the following steps: S1. Based on the nonlinear energy harvesting model, a D2D-assisted wireless energy supply MEC system model is established; S2, through D2D communication technology, remote users and nearby users can directly transmit data, and based on mathematical models, the system energy efficiency is used as the objective function to establish a joint optimization problem; S3. Based on Lyapunov optimization, the joint optimization problem is deterministically transformed, and the nonlinear subproblems introduced by the nonlinear energy harvesting model are linearly transformed; S4. Based on the penalty dual decomposition method, the transformed system energy efficiency optimization problem is decomposed into sub-problems for solution.
[0008] Preferably, step S2 includes: S21, establishing a user D2D transmission offloading decision model according to the task transmission of the near user and the far user; S22, establishing a user computing model, including a local computing model, a remote user unloading model, and a local user unloading model; S23, establishing a user energy model based on a piecewise linear energy harvesting model; S24, based on tasks and energy, establish queue models respectively; S25. Define long-term network energy efficiency and establish a mathematical formula to describe the original optimization problem.
[0009] Preferably, in step S25, the original optimization problem takes system energy efficiency as the objective function, which is expressed as follows: ; In the formula, For system energy efficiency, for Energy consumption of time-slotted systems, For nearby users The amount of tasks processed in the time slot, , They are , The limit of the time average expectation value of The original optimization problem is:
[0010] ; In the formula, is the number of single-antenna users, is the number of recent users, the subscript is Represents a remote user, with the subscript Indicates the nearest user. express The unloading decision of the time slot, express The transmission time of the time slot user sending information to the IRS, express The transmission power of the time slot user sending information to the IRS, Indicates user In time slot The local CPU frequency, represents the equivalent baseband channel from HAP to remote users, represents the equivalent baseband channel from HAP to remote users, Indicates Threshold of piecewise linear function, , are the transmission power and transmission time of the HAP broadcast wireless energy signal, represents the coefficient of the chip structure, Indicates user In time slot The task queue, Indicates user In time slot energy queue; among them, C1, C2 and C3 are discrete constraints generated by the user D2D transmission offloading strategy, C4 and C5 are the transmission time constraints that need to be met, C6 is the user CPU frequency constraint, C7 is the task and energy queue stability constraint, C8 is the user transmission power constraint, and C9 and C10 are power constraints generated by the model using a piecewise linear energy harvesting model.
[0011] Preferably, step S3 includes transforming the original optimization problem based on Lyapunov optimization and processing the fractional form of the objective function to obtain the transformed optimization problem, which is expressed as follows: ; In the formula, is the weight parameter, Indicates user time slot The amount of tasks being processed.
[0012] Preferably, discrete variables are introduced to linearize the piecewise linear function introduced in the nonlinear energy harvesting model. The transformed optimization problem expression is: ; In the formula, is the introduced discrete variable.
[0013] Preferably, in step S4, solving the converted system energy efficiency optimization problem includes: S41, using the augmented Lagrangian method to process the equality constraints in the optimization problem, i.e., the constraints brought by the piecewise linear function and the discrete constraints generated in the D2D offloading strategy, to obtain the augmented Lagrangian problem; S42, by using an effective block coordinate descent method, the augmented Lagrangian problem is divided into a D2D offloading strategy optimization subproblem, a transmission power and time optimization subproblem, and an auxiliary variable and local CPU frequency optimization subproblem, and they are solved in sequence; S43. An outer loop algorithm based on the penalty dual decomposition method updates the penalty factor and Lagrange multiplier to solve the optimal system energy efficiency.
[0014] Preferably, in step S42, the D2D offloading strategy optimization sub-problem is as follows: ; In the formula, , , , and is the Lagrange multiplier, is the penalty factor.
[0015] Preferably, in step S42, the transmission power and time optimization sub-problems are specifically as follows: ; Among them, transmission power and transmission time are highly coupled variables. By introducing auxiliary variables and To solve.
[0016] Preferably, in step S42, the auxiliary variables and local CPU frequency optimization sub-problems are specifically as follows: ; In the formula, is an auxiliary variable, is the time slot length, For users In time slot energy consumption.
[0017] Therefore, the present invention adopts the energy efficiency optimization method of the above-mentioned D2D-assisted wireless power supply mobile edge computing system, which has the following technical effects: (1) By jointly optimizing task offloading, transmission time, power control, and CPU frequency, the algorithm effectively reduces system energy consumption and maximizes energy efficiency, which is particularly suitable for energy-constrained mobile devices.
[0018] (2) The introduction of D2D communication technology enables devices to communicate directly, reducing the data load of the core network and significantly improving spectrum utilization and system throughput.
[0019] (3) Through Lyapunov optimization theory, the stability of the system in a dynamic network environment is ensured, avoiding performance instability problems caused by network fluctuations or load changes.
[0020] (4) The penalty dual decomposition method is used to decompose complex optimization problems, which greatly improves the solution speed and can quickly and accurately find the optimal solution to meet the real-time processing requirements of the system.
[0021] (5) It can perform collaborative optimization in multi-user scenarios to ensure the fairness and efficiency of task offloading and energy allocation, thereby improving the overall system performance. It is suitable for resource-constrained devices and scenarios with high energy-saving requirements.
[0022] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Figure 1 It is a data transmission schematic diagram in an embodiment of an energy efficiency optimization method of a D2D-assisted wireless power supply mobile edge computing system; Figure 2 It is a schematic diagram of the PDD framework algorithm in an embodiment of the energy efficiency optimization method of the D2D-assisted wireless power supply mobile edge computing system. DETAILED DESCRIPTION
[0024] The present invention can be explained in more detail by the following examples. The purpose of disclosing the present invention is to protect all changes and improvements within the scope of the present invention. The present invention is not limited to the following examples.
[0025] like Figure 1 As shown, the energy efficiency optimization method of the D2D-assisted wireless power supply mobile edge computing system provided by the present invention includes research and application of mobile edge computing (MEC), device-to-device (D2D) communication, wireless power transmission (WPT), and energy efficiency optimization algorithm. The specific steps are as follows: S1. Based on the nonlinear energy collection model, a D2D-assisted wireless power supply MEC system model was established. By introducing D2D communication technology, more efficient communication services can be provided for the wireless power supply MEC system, and the overall performance of the system can be improved. The details are as follows: Consider a system consisting of a hybrid access point (HAP) equipped with an RF energy transmitter and a MEC server. Single Antenna User , Among them, remote users are express, , recent users by express, Access points can broadcast radio signals to users near and far through WPT to provide stable and sustainable energy supply. time slots, use The length of the time slot is . In addition, each time block is divided into four stages, namely, the energy transmission stage, the far user unloading stage, the near user unloading stage, and the data return stage. First, in the energy transmission stage, the HAP broadcasts energy signals to far and near users downlink, and the far and near users equipped with energy transmission modules collect the energy signals and convert them into energy. In this process, due to distance problems, the energy and data signal strength obtained by far users will be less than those of near users. In order to avoid mutual interference between users, this embodiment adopts TDMA communication technology. In the far user unloading stage, each far user chooses to offload the task to the HAP or transmit it to the near user through D2D communication in turn. In the near user unloading stage, the near users unload directly to the HAP through uplink transmission in turn. In the data return stage, due to the low latency of edge computing, the download speed will be much faster than the unloading speed, so the duration of this stage can be ignored in this model.
[0026] By introducing device-to-device (D2D) communication technology, direct data transmission between devices can be achieved, allowing users who are far away from the base station to offload tasks to users who are closer to the base station, solving the "double near-far effect" problem in the wireless power supply MEC system, thereby reducing the data load of the core network of the communication system and effectively improving the system's spectrum efficiency and throughput.
[0027] S2. According to the data transmission and offloading process of far users and near users, a mathematical model is used to establish a joint optimization problem to maximize the system energy efficiency, which is suitable for energy-constrained mobile devices.
[0028] S21. Establish a user D2D transmission offloading decision model, as follows: During a specific time slot t, the far user chooses to offload the task to the HAP or transmit it to the near user via D2D communication. Indicates uninstall decision: when When In time slot Choose to offload the task directly to the HAP; when When Transmitted to nearby users via D2D communication .
[0029] In order to prevent a near user from receiving computing task transmissions from multiple far users, a far user can only perform D2D transmission with one near user in the same time slot, and a near user can only receive D2D transmissions from one far user. Therefore, the following constraints need to be met: ; S22. Establish a user calculation model, as follows: (1) Local computing model Because the user The computing module and energy collection module of the mobile user work independently, so the user The time slot length can be The entire time slot Perform local calculations within the Indicates user The number of CPU cycles required to process a bit of work, Indicates user In time slot The local CPU frequency in can now be calculated In time slot The amount of tasks processed by local computing : ; (2) Remote User Uninstallation Model when When remote users We choose to offload the computational tasks directly to the HAP. The equivalent baseband channel from the HAP to the remote user is expressed as , assuming that the remote user is in the time slot With transmission power and transmission time Send information to the IRS, then the user is blocked In the The amount of tasks offloaded to HAP in time slots It is expressed as: ; in, Indicates the system transmission bandwidth, is the Gaussian white noise power at the HAP.
[0030] when When Choose to use D2D communication to offload the computing task to a mobile assistant. The equivalent baseband channel between the blocked user and the mobile assistant is used Indicates that the blocked user In the The amount of tasks transmitted through D2D in each time slot It is expressed as: ; In the formula, is the Gaussian white noise power at the mobile assistant.
[0031] when When remote users The computing task is transmitted to the near user through D2D communication. The equivalent baseband channel from the far user to the near user is Indicates that users far away In the The amount of D2D transmission tasks in the time slot for: ; In the formula, is the Gaussian white noise power near the user.
[0032] Therefore, remote users In the The total number of tasks processed in a time slot for: ; In the formula, For remote users In the The amount of local computing tasks in a time slot.
[0033] (3) Near-user uninstall model Unlike far users, near users The computation task is directly offloaded to the HAP. The equivalent baseband channel from the HAP to the remote user can be expressed as , in uplink transmission, near the user In time slot Transmit power and transmission time Send information to HAP. Therefore, the user In the The amount of tasks unloaded in a time slot for: ; Since the remote user will transfer the computing task to the nearby user through D2D communication, the nearby user The amount of tasks received through D2D communication transmission It can be expressed as: ; Nearby users The total number of tasks processed in a time slot for: ; In the formula, For nearby users In the The amount of local computing tasks in a time slot.
[0034] In summary, the total amount of tasks processed by the users in the time slot is for: ; S23. Establish a user energy model, as follows: Existing energy harvesting (EH) models are generally divided into two categories: linear EH models and nonlinear EH models. In practical applications, due to the nonlinear characteristics of components such as diodes, inductors and capacitors in EH circuits, power conversion is usually a nonlinear function of received power. Therefore, in this embodiment, a practical nonlinear EH model is adopted, that is, The piecewise linear EH model of the segment is expressed as follows: ; In the formula, express and The functional relationship of Indicates the RF power received by the user after channel attenuation; ( , ) represents the threshold of each segment of the piecewise linear function; and Represent the slope and intercept of each linear function segment, respectively.
[0035] Since there is a sensitivity threshold and saturation voltage in the EH circuit, when the input RF power is less than the sensitivity threshold, the harvested energy is 0; when the input RF power exceeds the saturation voltage, the EH circuit is saturated and the energy output cannot be further increased. Therefore, it can be obtained that hour, ;when hour, , ,in Indicates the saturation voltage.
[0036] In time slot In the energy transmission phase, HAP broadcasts wireless energy signals with a transmission power of , the transmission time is Therefore, the energy consumed by HAP in this time slot can be expressed as: ; Through the piecewise linear EH model, the energy obtained by the user in the energy collection stage can be expressed as: ; For users, their energy consumption is divided into two parts, namely local computing energy consumption and link transmission energy consumption.
[0037] user In time slot Local computing energy consumption It is expressed as: ; In the formula, A coefficient that depends on the chip structure and is determined by actual measurements.
[0038] user Uplink transmission energy consumption It is expressed as: ; user In time slot The energy consumed is expressed as: ; In the whole time block, the users have energy collection and consumption. Assuming that they have no energy storage at the beginning and the collected energy is greater than the consumed energy, the user In time slot The final energy obtained It is expressed as: ; S24. Establish a queue model, as follows: use Indicates near and far users The data task queue is used Indicates The size of the computational task that reaches the user at the end of the time slot. Without loss of generality, assume that and are independent and identically distributed in all time slots, and the average arrival rate is , . Therefore, their update rules are as follows: ; Similarly, not only are there task queues, but energy can also be represented in the form of queues. Represents mobile users In time slot The energy queue of is updated as follows: ; Obviously, the user The energy consumed in a time slot cannot be greater than the time slot Battery remaining power and time slot before starting The sum of the energy collected by the users, so there is a constraint: ; S25. The goal of this embodiment is to improve the long-term energy efficiency of the entire network while ensuring the stability of different types of task queues. Based on this, this embodiment defines the objective function as long-term network energy efficiency and provides definitions of network stability and network energy efficiency, thereby establishing a mathematical formula to describe the optimization problem, as follows:
[0039] Definition 1: A data queue with bounded average backlog time is stable.
[0040] ; Definition 2: If each data queue in the network is stable, then the entire network queue is also stable.
[0041] Definition 3: System energy efficiency is defined as the ratio between the long-term total energy consumption of the entire network and the total computing task size completed in a period of time.
[0042] ; in, , Time slots Total system energy consumption and users The amount of tasks being processed, , Respectively , The limit of the time average expectation value.
[0043] It should be noted that the definition of network energy efficiency is very different from the common definition. This embodiment takes the concept of time into consideration and defines it from a long-term perspective. It can describe both energy efficiency and average delay.
[0044] In summary, the original optimization problem of the long-term energy efficiency of the D2D-assisted wireless power supply mobile edge computing system is expressed as:
[0045] In the above problem, C1, C2 and C3 are discrete constraints brought by discrete variables generated by the user cooperative D2D transmission offloading strategy, C4 and C5 are the transmission time constraints that need to be met, C6 is the user CPU frequency constraint, C7 is the task and energy queue stability constraint, C8 is the user transmission power constraint, and C9 and C10 are the power constraints generated by the piecewise linear energy harvesting model adopted by the model.
[0046] This is a joint optimization problem involving multiple variables, including discrete variables, and the constraints are nonlinear, so it is a mixed integer nonlinear programming problem with multivariable coupling. In addition, since this problem is optimized in a long-term random environment, the complexity of the solution is increased. Compared with the traditional MINLP problem, this problem is more difficult. Not only does it need to deal with deterministic constraints and continuous variables, but it also involves the coupling of discrete variables and nonlinear constraints, making the problem more challenging. Therefore, this problem is regarded as an NP-hard problem, that is, as the scale increases, the difficulty of solving it increases exponentially, and it is impossible to obtain the optimal solution within a limited time through traditional methods.
[0047] S3. It can be noted that the objective function of this problem is in the form of a non-convex fraction, which may lead to difficulties in solving. Based on this, the original optimization problem is transformed as follows: Using the properties of fractions to change the form of the objective function, first, theorem 1 is proved: the system energy efficiency is optimal if and only if .
[0048] Then, according to Theorem 1, the fractional form of the objective function is processed and the original optimization problem is equivalently transformed into: ; in, The optimal solution to the problem is still difficult to deal with in the optimization problem after equivalent transformation, that is, The value of is unknown. Indicates that the system is in time slot The total energy consumption before the start, Indicates that the system is in time slot The total amount of tasks to be processed before starting, The system is in the time slot The total energy efficiency before the start. Therefore, we get Before we begin, the system energy efficiency is expressed as: ; in, .
[0049] use replace , thus further transforming the optimization problem into:
[0050] Although the above method has processed the fractional terms in the original optimization problem, the converted problem still belongs to long-term optimization. In order to convert it into a short-term deterministic problem, this embodiment uses Lyapunov optimization to perform deterministic transformation on the converted problem again to ensure the stability of the system in a dynamic network environment and avoid performance instability caused by network fluctuations or load changes.
[0051] Among them, the Lyapunov function is defined as: ; The Lyapunov drift plus penalty function is defined as: ; In the formula, is an adjustable weight parameter that indicates the relative importance between system energy efficiency and latency. To ensure network stability, Minimizing, which also means minimizing the energy efficiency of the system.
[0052] According to the following formula, the upper bound of the optimization problem is derived: ; in, ; In the formula, and They are and The upper bound of .
[0053] According to the principle of expected opportunity minimization, the right side of the inequality is minimized to find the optimal solution for the system energy efficiency of each time slot. Therefore, the problem is transformed into the following form: ; In the formula, is the uplink / downlink IRS phase shift vector set.
[0054] In addition, since the piecewise linear EH function is essentially a nonlinear function, it is difficult to handle directly. Therefore, a set of discrete variables needs to be introduced Linearize it and express it as: ; In the formula, and Function coefficients representing the piecewise linear energy harvesting model.
[0055] After linearizing the EH model, the optimization problem is further expressed as:
[0056] S4, such as Figure 2As shown, this embodiment adopts a penalized dual decomposition (PDD) framework to solve the converted optimization problem. The penalized dual decomposition method can transform complex nonlinear optimization problems into easy-to-handle sub-problems, significantly improving the solution speed of the algorithm, so that the system can quickly find the optimal solution within a limited time to meet the needs of real-time processing. Specifically, the following steps are included: S41. Using the augmented Lagrangian (AL) method, the equality constraints in the optimization problem, i.e., the constraints brought by the piecewise linear function and the discrete constraints generated in the D2D offloading strategy, are processed to obtain the augmented Lagrangian problem, which is as follows: Due to the discrete variables in the optimization problem after Lyapunov optimization transformation, the optimization problem becomes difficult to handle. Therefore, this embodiment introduces auxiliary variables , , the constraints C8 and C11 are equivalently transformed into:
[0057] Then, using the AL method, the transformed equation constraints are penalized into the objective function. At this time, the new AL problem is expressed as: ; In the formula, , , , and is the Lagrange multiplier, is the penalty factor. In the PDD framework, the Lagrangian multiplier and the penalty factor remain fixed in the inner loop and are not updated; their updates are only performed in the outer loop of PDD. As the updates in the outer loop continue until convergence, the augmented Lagrangian problem will tend to be equivalent to the original problem.
[0058] S42. Since the variables and constraints in the AL problem are highly coupled, it is difficult to process directly. Therefore, this embodiment uses an effective block coordinate descent (BCD) method to divide the augmented Lagrangian problem into D2D offloading strategy optimization subproblem, transmission power and time optimization subproblem, auxiliary variable and local CPU frequency optimization subproblem, and solves them in sequence, as follows: (1) When other variables are fixed, the D2D offloading strategy optimization sub-problem is expressed as:
[0059] Among them, the equality constraint has been transferred to the objective function through the AL method. At this time, the variables that need to be optimized are the D2D offloading strategy and the piecewise function selection of the piecewise linear EH model. It can be found that this subproblem is a convex optimization problem with inequality constraints. Due to the properties of convex sets, all solutions that meet the constraints constitute a convex set, which makes the problem have a good solution space structure in the set and can be solved using various convex optimization tools and classic algorithms, such as CVX and CPLEX toolboxes.
[0060] (2) For the transmission power and time optimization sub-problems, the two sets of variables are highly coupled, as shown below:
[0061] In order to deal with the difficulties caused by highly coupled variables, a set of auxiliary variables are introduced and , it is easy to find yes If it is a concave function, then its perspective function Too and According to the properties of the perspective function, it can be proved that the transformed sub-problem is a convex problem and can be solved by CVX and CPLEX toolboxes.
[0062] (3) For the auxiliary variable and local CPU frequency optimization sub-problem, the specific representation is as follows:
[0063] Since the auxiliary variable , and mobile user local CPU frequency They do not interfere with each other and are independent of each other, so they can be solved separately.
[0064] for and For example, they are typical unconstrained convex quadratic programming problems, and the closed-form solution can be directly obtained as: ; and The optimal solution of is located at its boundary or stationary point, and its closed-form solution can be expressed as: ; S43. The outer loop algorithm based on the penalized dual decomposition (PDD) method continuously updates the penalty factor and Lagrange multiplier in each loop iteration to solve the optimal system energy efficiency.
[0065] Among them, the penalty factor Available through , renew, is a step size control parameter.
[0066] The update of the AL multiplier is as follows:
[0067] The optimization problem uses violation indicators to control the validity of the equality constraints in the problem. As shown below: ; Therefore, the present invention adopts the energy efficiency optimization method of the above-mentioned D2D-assisted wireless power supply mobile edge computing system, so that data can be transmitted directly between devices without the intervention of core management equipment, which can reduce the data load of the core network of the communication system and effectively improve the spectrum efficiency and throughput of the system.
[0068] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.
Claims
1. A method for optimizing energy efficiency of a D2D-assisted wireless power supply mobile edge computing system, characterized in that: The following steps are involved: S1. Based on the nonlinear energy harvesting model, a D2D-assisted wireless energy supply MEC system model is established; S2, through D2D communication technology, remote users and nearby users can directly transmit data, and based on mathematical models, the system energy efficiency is used as the objective function to establish a joint optimization problem; S3. Based on Lyapunov optimization, the joint optimization problem is deterministically transformed, and the nonlinear subproblems introduced by the nonlinear energy harvesting model are linearly transformed; S4. Based on the penalty dual decomposition method, the transformed system energy efficiency optimization problem is decomposed into sub-problems for solution.
2. The energy efficiency optimization method of the D2D assisted wireless power supply mobile edge computing system according to claim 1 is characterized in that: Step S2 includes: S21, establishing a user D2D transmission offloading decision model according to the task transmission of the near user and the far user; S22, establishing a user computing model, including a local computing model, a remote user unloading model, and a local user unloading model; S23, establishing a user energy model based on a piecewise linear energy harvesting model; S24, based on tasks and energy, establish queue models respectively; S25. Define long-term network energy efficiency and establish a mathematical formula to describe the original optimization problem.
3. The energy efficiency optimization method of the D2D assisted wireless power supply mobile edge computing system according to claim 2 is characterized in that: In step S25, the original optimization problem takes the long-term network energy efficiency as the objective function, which is expressed as follows: ; In the formula, For system energy efficiency, for Energy consumption of time-slotted systems, For nearby users The amount of tasks processed in the time slot, , They are , The limit of the time average expectation value of The original optimization problem is: ; In the formula, is the number of single-antenna users, is the number of recent users, the subscript is Represents a remote user, with the subscript Indicates the nearest user, subscript Indicates user , superscript Indicates the maximum value, express The unloading decision of the time slot, express The transmission time of the time slot user sending information to the IRS, express The transmission power of the time slot user sending information to the IRS, Indicates user In time slot The local CPU frequency, represents the equivalent baseband channel from HAP to remote users, represents the equivalent baseband channel from HAP to remote users, Indicates Threshold of piecewise linear function, , are the transmission power and transmission time of the HAP broadcast wireless energy signal, represents the coefficient of the chip structure, Indicates user In time slot The task queue, Indicates user In time slot energy queue; among them, C1, C2 and C3 are discrete constraints generated by the user D2D transmission offloading strategy, C4 and C5 are the transmission time constraints that need to be met, C6 is the user CPU frequency constraint, C7 is the task and energy queue stability constraint, C8 is the user transmission power constraint, C9 and C10 are the power constraints generated by the piecewise linear energy harvesting model adopted by the model.
4. The energy efficiency optimization method of the D2D assisted wireless power supply mobile edge computing system according to claim 1, characterized in that: In step S3, the original optimization problem is transformed based on Lyapunov optimization, and the fractional form of the objective function is processed to obtain the converted optimization problem, which is expressed as follows: ; In the formula, is the weight parameter, Indicates user time slot The amount of tasks being processed.
5. The energy efficiency optimization method of the D2D assisted wireless power supply mobile edge computing system according to claim 4 is characterized in that: A set of discrete variables is introduced to linearize the piecewise linear function introduced in the nonlinear energy harvesting model. The transformed optimization problem expression is: ; In the formula, is the introduced discrete variable.
6. The energy efficiency optimization method of the D2D assisted wireless power supply mobile edge computing system according to claim 1, characterized in that: In step S4, the converted system energy efficiency optimization problem is solved, including: S41, using the augmented Lagrangian method to process the equality constraints in the optimization problem, i.e., the constraints brought by the piecewise linear function and the discrete constraints generated in the D2D offloading strategy, to obtain the augmented Lagrangian problem; S42, by using an effective block coordinate descent method, the augmented Lagrangian problem is divided into a D2D offloading strategy optimization subproblem, a transmission power and time optimization subproblem, and an auxiliary variable and local CPU frequency optimization subproblem, and they are solved in sequence; S43. An outer loop algorithm based on the penalty dual decomposition method updates the penalty factor and Lagrange multiplier to solve the optimal system energy efficiency.
7. The energy efficiency optimization method of the D2D assisted wireless power supply mobile edge computing system according to claim 6, characterized in that: In step S42, the D2D offloading strategy optimization sub-problem is as follows: ; In the formula, , , , and is the Lagrange multiplier, is the penalty factor.
8. The energy efficiency optimization method of the D2D assisted wireless power supply mobile edge computing system according to claim 6, characterized in that: In step S42, the transmission power and time optimization sub-problems are as follows: ; Among them, transmission power and transmission time are highly coupled variables. By introducing auxiliary variables and To solve.
9. The energy efficiency optimization method of the D2D assisted wireless power supply mobile edge computing system according to claim 6, characterized in that: In step S42, the auxiliary variables and local CPU frequency optimization sub-problems are as follows: ; In the formula, is an auxiliary variable, is the time slot length, For users In time slot energy consumption.
Citation Information
Patent Citations
D2D auxiliary cooperative computing unloading and resource allocation method in wireless energy supply edge computing scene
CN118102315A
Terminal multi-task parallel unloading method based on deep reinforcement learning
CN118215079A
Base station device and method for operating base station device
US20230269575A1
Task offloading and resource allocation method in uncertain network environment
US20240103915A1
Cited By
Distributed optimization-based wireless energy supply mobile edge computing system-level energy efficiency optimization method
CN121568144A
