Energy efficiency optimization method of D2D_MEC system based on IRS assistance

Abstract

The present application provides an energy efficiency optimization method for a D2D_MEC system assisted by an IRS, which mainly includes the following steps: a task-type dual-antenna user TU searches for a nearby RU with low task computing intensity, establishes a D2D link through the IRS, and obtains the channel information of the D2D link of the current time slot and the buffer queue length of the RU at the beginning of each time slot, and offloads part of the tasks to the RU for calculation; establishes a cellular link with the base station in the system through the IRS, obtains the channel information of the cellular link and the edge cloud buffer queue length at the beginning of each time slot, and offloads part of the computing tasks to the edge cloud for calculation, and according to the respective buffer queue lengths in the edge cloud, TU and RU, by optimizing the division ratio of the computing tasks arrived at the TU, the distribution of the transmission power of the TU cellular link and the D2D link, the phase shift of the reflection unit of the IRS, and the distribution of the computing frequency of the edge cloud, TU and RU, the energy efficiency of the IRS-assisted D2D_MEC system on a long time scale is minimized.

Description

Technical Field

[0001] The present application relates to the technical field of D2D communication and mobile edge computing, and in particular to an energy efficiency optimization method for a D2D_MEC system based on IRS assistance. Background Art

[0002] A significant challenge in 5G wireless communications is the poor penetration of millimeter waves, the primary band, which are easily blocked by obstacles. Without additional support, wireless communication in complex scenarios, such as indoor environments, is difficult to achieve. Previously, a proposed approach addressed this issue: designing passive, reflective, reconfigurable, and low-cost intelligent surfaces (LIS / Large Intelligent Surface / Reconfigurable Meta-Surfaces / IRS / Intelligent Reflecting Surface, referred to as IRS) to assist in wireless communications. By appropriately placing and distributing these surfaces, signals can be controllably reflected to the desired direction at the IRS, effectively adding a detour channel to achieve effective wireless communication and improve the nearby wireless communication environment. This is particularly true when direct signals between two communicating parties are blocked. IRSs, consisting of an IRS controller and a large number of low-cost passive reflective elements, are considered an innovative, hardware-efficient technology that surpasses fifth-generation wireless systems. IRSs can dynamically adjust their reflection coefficients to alter signal propagation, thereby enhancing and suppressing desired and interfering signals, respectively. Therefore, by cleverly coordinating these reflection units to adjust the amplitude and phase of the reflected signal, IRS can create a favorable signal propagation environment to significantly improve the wireless communication coverage, throughput and energy efficiency.

[0003] An important parameter in the IRS is the reflection coefficient matrix. This matrix represents the phase shift of each element in the IRS and is a key parameter for improving signal strength and wireless communication coverage through IRS reflection. An inaccurately estimated reflection coefficient matrix will result in poor communication quality, impacting communication effectiveness. Due to its passive, low-cost, and reconfigurable nature, IRS holds great promise in mobile communications. However, due to the complex and diverse nature of real-world application scenarios, a solution for adjusting the IRS reflection coefficient matrix is ​​needed to enable IRS-assisted wireless communication systems to communicate more effectively in these complex and changing scenarios.

[0004] The rapid development of the Internet of Things (IoT) will enable tens of billions of resource-constrained mobile devices, such as mobile devices, sensors, and wearable computing devices, to connect to the internet via cellular networks. Limited battery life and limited mobile computing capabilities pose significant challenges to IoT design. Mobile edge computing, as an effective solution that provides cloud-like computing capabilities to end users at the network edge and base stations, has attracted widespread attention from academia and industry. Designing energy-efficient control strategies is particularly important for reducing system energy consumption.

[0005] Some computationally intensive devices can offload their computational tasks directly to nearby idle devices via D2D links, allowing wireless devices to share their unused computing resources, thereby improving overall computing performance and saving resources. Currently, work on D2D collaborative computing primarily focuses on one-time optimization of static computing tasks for mobile users.

[0006] Therefore, when the actual task arrival time varies, how to allocate the amount of D2D offloading tasks to achieve mutual benefit between users and reduce system energy consumption is a problem worth studying. Summary of the Invention

[0007] The exemplary embodiments of the present application provide an energy efficiency optimization method based on an IRS-assisted D2D_MEC system, so as to at least reduce the total delay and system energy consumption of the D2D_MEC system in processing data, and achieve the technical effect of parallel transmission and computing.

[0008] Each exemplary embodiment of the present application provides an energy efficiency optimization method based on an IRS-assisted D2D_MEC system, including the following steps:

[0009] Step 1: The task-type dual-antenna user TU optimizes the ratio of the task volume arriving in the current time slot allocated to each buffer of the TU based on the queue length of all buffers of the TU in the current time slot and the task volume arriving in the current time slot of the TU;

[0010] Step 2: Obtain channel information of the IRS-assisted D2D link, channel information of the cellular link, information of the D2D_MEC system, and the buffer queue length of the TU for offloading in the current time slot. The TU offloads the tasks in the buffer to the edge cloud and the resource-based single-antenna user RU through the IRS-assisted cellular link and the IRS-assisted D2D link, respectively, allocates the transmit power of the cellular offloading of the TU and the offloading of the RU, and optimizes the phase shift of the IRS-assisted reflection unit.

[0011] Step 3: Optimize the allocation of computing frequencies of the TU, RU, and edge cloud based on the buffer queue lengths of the TU, RU, and edge cloud for the task calculation in the current time slot, and obtain the optimal energy efficiency solution of the IRS-assisted D2D_MEC system in the current time slot;

[0012] Step 4: Update the queue lengths of all buffers in the D2D_MEC system for the next time slot and the average energy efficiency of the IRS-assisted D2D_MEC system from the first time slot to the current time slot according to the energy efficiency optimal solution;

[0013] Step 5: Repeat steps 1 to 4 until the number of time slots reaches a preset number of time slots, and obtain the optimal average energy efficiency value of the IRS-assisted D2D_MEC system.

[0014] By adopting the above technical solution, the task-based dual-antenna user TU of the present invention searches for nearby D2D devices with low task computation intensity (resource-based single-antenna user RU), establishes a D2D link through the IRS, obtains the channel information of the D2D link and the buffer queue length of the RU at the beginning of each time slot, and offloads part of the computation task to the RU. It also establishes a cellular link with the base station in the system through the IRS, obtains the channel information of the cellular link and the buffer queue length of the edge cloud at the beginning of each time slot, and offloads part of the computation task to the edge cloud for computation.

[0015] According to the queue lengths of the buffers in the edge cloud, TU, and RU, the energy efficiency of the IRS-assisted D2D_MEC system is minimized over long time scales by optimizing the division ratio of the computing tasks arriving at the TU, the distribution of the transmission power of the TU cellular link and the D2D link, the phase shift of the IRS reflection unit, and the distribution of the computing frequency of the edge cloud, TU, and RU.

[0016] The present application has the following beneficial effects: the present invention effectively reduces the energy efficiency of the system by assisting in offloading computing tasks of cellular links and D2D links through IRS; introduces Lyapunov optimization theory, uses dynamic strategies to process time-varying wireless channel information and the arrival of TU and RU random tasks, minimizes the system energy efficiency for each time slot to optimize the system performance over a long time scale, not only reduces the delay of edge offloading, but also reduces the impact of random channels on system performance; and through D2D communication-assisted computing, effectively reduces the total delay of system data processing on the basis of mobile edge computing, thereby improving system performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 This is a flow chart of energy efficiency optimization of the IRS-assisted D2D_MEC system in the present invention.

[0019] Figure 2 This is a structural diagram of each buffer for energy efficiency optimization of the IRS-assisted D2D_MEC system in the present invention.

[0020] Figure 3 This is a scene model diagram of the IRS-assisted D2D_MEC system in the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the preferred embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0022] like Figure 1 As shown, an energy efficiency optimization method based on an IRS-assisted D2D_MEC system includes the following steps:

[0023] Step 1: The task-type dual-antenna user TU optimizes the ratio of the task volume arriving in the current time slot allocated to each buffer of the TU based on the queue length of all buffers of the TU in the current time slot and the task volume arriving in the current time slot of the TU;

[0024] Step 2: Obtain channel information of the IRS-assisted D2D link, channel information of the cellular link, information of the D2D_MEC system, and the buffer queue length of the TU for offloading in the current time slot. The TU offloads the tasks in the buffer to the edge cloud and the resource-based single-antenna user RU through the IRS-assisted cellular link and the IRS-assisted D2D link, respectively, allocates the transmit power of the cellular offloading of the TU and the offloading of the RU, and optimizes the phase shift of the IRS-assisted reflection unit.

[0025] Step 3: Optimize the allocation of computing frequencies of the TU, RU, and edge cloud based on the buffer queue lengths of the TU, RU, and edge cloud for the task calculation in the current time slot, and obtain the optimal energy efficiency solution of the IRS-assisted D2D_MEC system in the current time slot;

[0026] Step 4: Update the queue lengths of all buffers in the D2D_MEC system for the next time slot and the average energy efficiency of the IRS-assisted D2D_MEC system from the first time slot to the current time slot according to the energy efficiency optimal solution;

[0027] Step 5: Repeat steps 1 to 4 until the number of time slots reaches a preset number of time slots, and obtain the optimal average energy efficiency value of the IRS-assisted D2D_MEC system.

[0028] In step 1, all buffers of the TU include: buffer Q for local calculation L , the current time slot queue length is Q L (t), buffer Q used to offload tasks to the edge cloud RC , the current time slot queue length is Q RC (t), buffer Q used to offload tasks to RU RD The current time slot queue length is Q RD (t), the TU task arrival amount is L(t), and the task complexity is γ D (t), the current time slot TU arrives at the task given to Q RC , Q RD , Q L The allocation ratios are α(t), (1-α(t))β(t), (1-α(t))(1-β(t)), and the length of one time slot is T.

[0029] In step 2, the information of the D2D link assisted by the IRS includes: the channel gain g from TU to IRS L (t), channel gain g from IRS to RU D (t), TU to RU channel gain h D (t), bandwidth B D , channel noise N0, and the transmit power P of the D2D link D (t); The IRS-assisted cellular link channel information includes: IRS to base station channel gain g C (t), TU to base station channel gain h C (t), bandwidth B C , channel noise N0, and the transmit power P of the cellular link D (t); The phase shift matrix of the IRS reflection unit is The maximum transmit power of TU is P max .

[0030] The task offloading amount of the current time slot of the cellular link is expressed as

[0031] The D2D link time slot task offloading amount is expressed as

[0032] Among them, the cellular link transmission energy consumption of the current time slot is P C T, the cellular link transmission energy consumption of the current time slot is P D T.

[0033] In step 3,

[0034] In the RU, the buffer used for computing tasks is Q D , the current time slot queue length is Q D (t), the amount of tasks arriving at the RU in the current time slot is D(t), and the task complexity is γ D (t);

[0035] In the edge cloud, the buffer used by the edge cloud for computing tasks is Q C , the current time slot queue length is Q C (t);

[0036] The computation frequency of the edge cloud computing task in the current time slot is f C (t), the maximum value of the edge cloud computing frequency is The calculation frequency of the RU in the current time slot is f L (t), the maximum value of the calculated frequency is The calculation frequency of the TU calculation task in the current time slot is f D (t), the maximum value of the calculated frequency is

[0037] In the current time slot, the sizes of TU, ​​RU, and edge cloud computing tasks are: The total computing task size of the system is: D A (t) = D C (t)+D L (t)+D D (t).

[0038] In step 4, all buffers Q in the next time slot RC , Q RD , Q L , Q D , Q C The queue lengths are:

[0039] Q RC (t+1)=[Q RC (t)-R C (t)] + +α(t)L(t),

[0040] QRD (t+1)=[Q RD (t)-R D (t)] + +(1-α(t))β(t)L(t),

[0041] Q L (t+1)=[Q L (t)-D L (t)] + +(1-α(t))(1-β(t))L(t),

[0042] Q D (t+1)=[Q D (t)-D D (t)] + +R D (t)+D(t),Q C (t+1)=[Q C (t)-D C (t)] + +R C (t).

[0043] The computing energy consumptions of the TU, the RU and the edge cloud of the current time slot are E L (t) = k L T(f L (t)) 3 、 E D (t) = k D T(f D (t)) 3 、E C (t) = k C T(f C (t)) 3 ;

[0044] The total energy consumption of the IRS-assisted D2D_MEC system in the current time slot is E A (t) = E L (t)+E D (t)+E C (t)+P D T+P C T;

[0045] In step 4, the average energy efficiency of the D2D_MEC system is the average of the system energy efficiencies of each time slot, and the system energy efficiency is defined as:

[0046]

[0047] The optimization in steps 1 to 3 includes solving the following problems:

[0048] P1:

[0049] st0≤α(t)≤1,t∈Ω,

[0050] 0≤β(t)≤1,t∈Ω,

[0051]

[0052]

[0053]

[0054] 0≤φ n (t)≤2π,n∈N,t∈Ω,

[0055] 0≤P D (t)≤P max ,t∈Ω,

[0056] 0≤P C (t)≤P max ,t∈Ω,

[0057] 0≤P D (t)+P C (t)≤P max ,t∈Ω,

[0058] Q RC (t),Q RD (t),Q C (t),Q D (t),Q L (t)are mean rate stable t∈Ω.

[0059] Among them, the average rate stability satisfies The optimization goal can be changed to

[0060]

[0061] At this time, the P1 problem is transformed into P1-1. Using Lyapunov optimization, P1-1 can be split into sub-problems P2, P3, P4, P5, and P6. V is the control parameter, which is a positive number:

[0062] P2:

[0063]

[0064] st0≤α(t)≤1,t∈Ω,

[0065] 0≤β(t)≤1,t∈Ω,

[0066] P3:

[0067]

[0068] st0≤α(t)≤1,t∈Ω,

[0069] 0≤φ n (t)≤2π,n∈N,t∈Ω,

[0070] 0≤P D (t)≤P max ,t∈Ω,

[0071] 0≤P C (t)≤P max ,t∈Ω,

[0072] 0≤P D (t)+P C (t)≤P max ,t∈Ω,

[0073] P4:

[0074]

[0075] P5:

[0076]

[0077] P6:

[0078]

[0079] In step 1, the workload arriving at the D2D_MEC system and the workload ratio allocated to each buffer of the TU are optimized to solve problem P2;

[0080] In step 2, allocating the transmit power of the cellular offloading of the TU and the RU offloading, and optimizing the phase shift of the IRS-assisted reflection unit is to solve problem P3;

[0081] In step 3, the distribution of computing frequencies of the TU, the RU, and the edge cloud is optimized to solve problems P4, P5, and P6, respectively.

[0082] like Figure 2 As shown, the following will explain steps 1 to 5 in detail:

[0083] Define the system queue vector: Θ(t) = (QRC (t),Q RD (t),Q C (t),Q L (t),Q D (t)), define the Lyapunov function The Lyapunov drift ΔΘ(t) is used to measure the queue stability: ΔΘ(t) = Ε{L(Θ(t+1))-L(Θ(t))|Θ(t)}. The system energy efficiency is minimized while ensuring the stability of the system queue. The objective function of the optimization problem P1-1 is used as the penalty function, and the optimization objective is reconstructed as The upper bound of :

[0084]

[0085] In step 1, from The number of items related to the TU arrival task split ratio is extracted from the upper bound of as the optimization objective of problem P2.

[0086] In step 2, from The number of terms related to TU transmit power and IRS phase shift is extracted from the upper bound of as the optimization objective of problem P3.

[0087] In step 3, from The number of items related to the computing frequency of TU, ​​RU, and edge cloud are extracted from the upper bound as the optimization objectives of problems P4, P5, and P6 respectively.

[0088] In step 4, the optimal solution for the current time slot is obtained according to the previous three steps, the system energy efficiency for the current time slot is updated, and the lengths of all buffer queues in the system for the next time slot are updated.

[0089] In step 5, steps 1 to 4 are repeated with the updated buffer queue length until the number of time slots is long enough or reaches a given number of time slots, and the optimal value of the average energy efficiency of the IRS-assisted D2D_MEC system is obtained.

[0090] The following combination Figure 3The logic of the present invention is briefly explained: The IRS-assisted D2D_MEC system is equipped with a base station equipped with an edge cloud server, two users located near the base station, and an IRS. Both users can establish a cellular link with the base station. The task-based user can offload computing tasks to the edge cloud server via the IRS, but the resource-based user cannot offload computing tasks. A D2D link can be established between the task-based user and the resource-based user. The task-based user can offload computing tasks to the resource-based user via the IRS, assuming that the resource-based user is willing to help the task-based user process computing tasks. The cellular link and the D2D link use different frequency bands to ensure that the two communication methods do not interfere with each other. The task-based user and the resource-based user generate new computing tasks in each time slot. We assume that computing tasks generated in the current time slot can only be processed in the current time slot and subsequent time slots. Both users and the edge cloud have cache queues. Computation tasks generated in the current time slot and previous time slots are first assigned and then enter the cache queue.

[0091] In summary, the energy efficiency optimization method for the IRS-assisted D2D_MEC system proposed in the present invention adopts a cache loading and unloading task allocation method, combined with D2D communication-assisted computing, IRS-assisted cellular communication and D2D communication, which effectively reduces the energy efficiency of data processing in the MEC system, improves the battery life of smart mobile devices, and extends the service life of the mobile edge network.

[0092] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications based on these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An energy efficiency optimization method based on IRS-assisted D2D_MEC system, characterized in that: The steps include: Step 1: The task-type dual-antenna user TU optimizes the ratio of the task volume arriving in the current time slot allocated to each buffer of the TU based on the queue length of all buffers of the TU in the current time slot and the task volume arriving in the current time slot of the TU; Step 2: Obtain channel information of the IRS-assisted D2D link, channel information of the cellular link, information of the D2D_MEC system, and the buffer queue length of the TU for offloading in the current time slot. The TU offloads the tasks in the buffer to the edge cloud and the resource-based single-antenna user RU through the IRS-assisted cellular link and the IRS-assisted D2D link, respectively, allocates the transmit power of the cellular offloading of the TU and the offloading of the RU, and optimizes the phase shift of the IRS-assisted reflection unit. Step 3: Optimize the allocation of computing frequencies of the TU, RU, and edge cloud based on the buffer queue lengths of the TU, RU, and edge cloud for the task calculation in the current time slot, and obtain the optimal energy efficiency solution of the IRS-assisted D2D_MEC system in the current time slot; Step 4: Update the queue lengths of all buffers in the D2D_MEC system for the next time slot and the average energy efficiency of the IRS-assisted D2D_MEC system from the first time slot to the current time slot according to the energy efficiency optimal solution; Step 5: Repeat steps 1 to 4 until the number of time slots reaches a preset number of time slots, and obtain the optimal average energy efficiency value of the IRS-assisted D2D_MEC system; In step 1, all buffers of the TU include: buffer Q for local calculation L , the current time slot queue length is Q L (t), buffer Q used to offload tasks to the edge cloud RC , the current time slot queue length is Q RC (t), buffer Q used to offload tasks to RU RD The current time slot queue length is Q RD (t), the TU task arrival amount is L(t), and the task complexity is γ D (t), the current time slot TU arrives at the task given to Q RC , Q RD , Q L The allocation ratios are ɑ(t), (1-ɑ(t))β(t), (1-ɑ(t))(1-β(t)), and the length of one time slot is T. In step 2, the information of the D2D link assisted by the IRS includes: the channel gain g from TU to IRS L (t), channel gain g from IRS to RU D (t), TU to RU channel gain h D (t), bandwidth B D , channel noise N0, and the transmit power P of the D2D link D (t); The IRS-assisted cellular link channel information includes: IRS to base station channel gain g C (t), TU to base station channel gain h C (t), bandwidth B C , channel noise N0, and the transmit power P of the cellular link C (t); The phase shift matrix of the IRS reflection unit is The maximum transmit power of TU is P max ; The task offloading amount of the current time slot of the cellular link is expressed as The D2D link time slot task offloading amount is expressed as Among them, the cellular link transmission energy consumption of the current time slot is P C T, the D2D link transmission energy consumption of the current time slot is P D T; In the step 3, In the RU, the buffer used for computing tasks is Q D , the current time slot queue length is Q D (t), the amount of tasks arriving at the RU in the current time slot is D(t), and the task complexity is γ D (t); In the edge cloud, the buffer used by the edge cloud for computing tasks is Q C , the current time slot queue length is Q C (t); The computation frequency of the edge cloud computing task in the current time slot is f C (t), the maximum value of the edge cloud computing frequency is The calculation frequency of the RU in the current time slot is f L (t), the maximum value of the calculated frequency is The calculation frequency of the TU calculation task in the current time slot is f D (t), the maximum value of the calculated frequency is In the current time slot, the sizes of TU, ​​RU, and edge cloud computing tasks are: The total computing task size of the system is: D A (t) = D C (t)+D L (t)+D D (t); In step 4, all buffers Q in the next time slot RC , Q RD , Q L , Q D , Q C The queue lengths are: Q RC (t+1)=[Q RC (t)-R C (t)] + +α(t)L(t), Q RD (t+1)=[Q RD (t)-R D (t)] + +(1-α(t))β(t)L(t), Q L (t+1)=[Q L (t)-D L (t)] + +(1-α(t))(1-β(t))L(t), Q D (t+1)=[Q D (t)-D D (t)] + +R D (t)+D(t), The computing energy consumptions of the TU, the RU and the edge cloud of the current time slot are E L (t) = k L T(f L (t)) 3 、E D (t) = k D T(f D (t)) 3 、E C (t) = k C T(f C (t)) 3 ; The total energy consumption of the IRS-assisted D2D_MEC system in the current time slot is E A (t) = E L (t)+E D (t)+E C (t)+P D T+P C T; In step 4, the average energy efficiency of the D2D_MEC system is the average of the system energy efficiencies of each time slot, and the system energy efficiency is defined as:

2. The energy efficiency optimization method based on the IRS-assisted D2D_MEC system according to claim 1 is characterized in that: The optimization in steps 1 to 3 includes solving the following problems: st0≤α(t)≤1,t∈Ω, 0≤β(t)≤1,t∈Ω, 0≤φ n (t)≤2π,n∈N,t∈Ω, 0≤P D (t)≤P max ,t∈Ω, 0≤P C (t)≤P max ,t∈Ω, 0≤P D (t)+P C (t)≤P max ,t∈Ω, Q RC (t),Q RD (t),Q C (t),Q D (t),Q L (t)are mean rate stable t∈Ω; Among them, the average rate stability satisfies The optimization goal can be changed to At this time, the P1 problem is transformed into P1-1, and P1-1 is split into sub-problems P2, P3, P4, P5, and P6. V is the control parameter, which is a positive number: st0≤α(t)≤1,t∈Ω, 0≤β(t)≤1,t∈Ω, st0≤α(t)≤1,t∈Ω, 0≤φ n (t)≤2π,n∈N,t∈Ω, 0≤P D (t)≤P max ,t∈Ω, 0≤P C (t)≤P max ,t∈Ω, 0≤P D (t)+P C (t)≤P max ,t∈Ω, 3. The energy efficiency optimization method of the IRS-assisted D2D_MEC system according to claim 2, characterized in that: In step 1, the workload arriving at the D2D_MEC system and the workload ratio allocated to each buffer of the TU are optimized to solve problem P2; In step 2, allocating the transmit power of the cellular offloading of the TU and the RU offloading, and optimizing the phase shift of the IRS-assisted reflection unit is to solve problem P3; In step 3, the distribution of computing frequencies of the TU, the RU, and the edge cloud is optimized to solve problems P4, P5, and P6, respectively.

Images