A 6G Internet of Things multi-layer computing method and system enabled by IRS-BackCom

Abstract

The present invention discloses a 6G Internet of Things multi-layer computing method and system enabled by IRS-BackCom, establishes a multi-layer computing system model of the 6G Internet of Things enabled by IRS-BackCom; performs problem modeling, simplifies the communication problem into a three-layer logical communication model, calculates task scheduling, and obtains an overall communication model of a task allocation scheme for local computing and partial data unloading; performs system problem description and decomposition on the communication model to obtain a communication model, distributes local computing and migration computing problems to the user device layer, the access point layer, and the central processing unit layer for processing, schedules tasks in time, performs local computing on problems that meet local computing capabilities, performs partial data unloading computing on problems that exceed local computing capabilities, and performs decomposition and calculation on the remaining problems, thereby realizing the 6G Internet of Things multi-layer computing enabled by IRS-BackCom. The present invention makes full use of the optimal computing capabilities of user devices, access points, and central processing units to perform task computing, greatly improving the efficiency of the entire communication system.

Description

Technical Field

[0001] The present invention belongs to the technical fields of wireless communication, intelligent reflective surface backscattering and multi-layer computing, and specifically relates to a 6G Internet of Things multi-layer computing method and system enabled by IRS-BackCom. Background Art

[0002] With the development of the 6G Internet of Things (IoT) ecosystem, more and more convenient facilities such as smart transportation, healthcare, wearable devices, industrial automation and other applications are being implemented. Wireless networks need to support more and more user devices, and at the same time, network capabilities are also facing severe tests.

[0003] The emerging demands have given rise to various quality of service requirements, such as high capacity, low latency, high reliability, and low cost. The limited energy budget and computing power in IoT devices can hardly meet such demands. Computational data offloading technology and intelligent reflecting surface backscattering technology (Intelligent Reflecting Surface-BackCom) provide solutions to improve this problem. Mobile cloud computing (MCC) and multi-access edge computing (MEC) as two common computational data offloading technologies are often used to alleviate the conflict between resource-scarce demands and resource-constrained IoT devices.

[0004] The key idea of ​​mobile cloud computing (MCC) is to offload data bits from IoT devices to remote cloud data centers with rich computing power for processing. This approach has some disadvantages, such as high latency and huge round-trip consumption. Compared with MCC, MEC technology provides a more suitable way to effectively handle latency-sensitive service issues, such as real-time signal processing. This is mainly because in MEC technology, the computing unit is pushed to the edge of the network. Although the edge deployment of computing units brings the advantage of low latency, MEC technology also has limited computing power and cannot meet the requests for high-load task execution.

[0005] The multi-layer computing system integrates MCC and MEC, combining their respective advantages to improve the processing and feedback capabilities of the data offloading system. In the multi-layer computing system, computing units are deployed in a hierarchical manner. When the computing task load requested by the IoT device exceeds the data processing capacity of the lower layer, some computing tasks are migrated to a more powerful and higher layer. The multi-layer computing system is superior to the planar computing system in terms of latency. Although multi-layer computing technology can reduce the data processing load of IoT devices, the data offloading process greatly increases the overhead of the transmission power, thereby aggravating the energy consumption problem of IoT devices. The emerging smart reflective surface backscattering technology makes up for this deficiency.

[0006] Smart reflector backscattering technology incorporates backscattering into smart reflectors. It can achieve passive signal transmission without an active RF link. Smart reflectors are two-dimensional electromagnetic (EM) hyperplanes that can efficiently adjust the reflection coefficient of each element unit to change the reflection characteristics of the incident electromagnetic wave to enhance the received signal and suppress interference noise. Compared with active antennas, IRS consumes much less power. Summary of the invention

[0007] The technical problem to be solved by the present invention is to provide an IRS-BackCom-enabled 6G Internet of Things multi-layer computing method and system to address the deficiencies in the above-mentioned prior art, so as to solve the technical problems of low data transmission efficiency, unstable transmission and high power consumption between the user terminal, access point and central server.

[0008] The present invention adopts the following technical solutions:

[0009] A 6G Internet of Things multi-layer computing method enabled by IRS-BackCom, comprising the following steps:

[0010] S1. Establish a multi-layer computing system model of IRS-BackCom to enable 6G Internet of Things;

[0011] S2. Modeling the multi-layer computing system model of IRS-BackCom empowering 6G Internet of Things obtained in step S1, simplifying the communication problem into a three-layer logical communication model, obtaining the overall communication model of the computing task scheduling of the user device layer, access point layer, and central processor layer, and obtaining the task allocation scheme of local computing and partial data offloading;

[0012] S3. The communication model obtained in step S2 is described and decomposed into a three-layer communication model. Local computing and migration computing problems are assigned to the user device layer, access point layer, and central processing unit layer for processing. Tasks are scheduled in time, problems that meet local computing capabilities are calculated locally, problems that exceed local computing capabilities are partially calculated by data offloading, and the remaining problems are decomposed and calculated to achieve multi-layer computing of 6G IoT enabled by IRS-BackCom.

[0013] Specifically, in step S1, the multi-layer computing system model includes:

[0014] A power station PB, K intelligent reflector IRS-assisted user equipment UE on the T1 layer, M access points AP on the T2 layer connected to a MEC server, and a central server with rich resources on the T3 layer; the power station PB, each unified access point AP and the central server are equipped with N p ,N a and N cAntennas, N c ≥M×N a ,In a hierarchical network, each user equipment UE requests to perform a computing task, each task is bit-by-bit independent and ,split into multiple bit subsets, all channels follow quasi-static fading and the channel information is known.

[0015] Specifically, in step S2, a partial data offloading strategy is adopted, and all first-layer user equipment UE and second-layer access point AP perform calculation offloading and local calculation at the same time, and the calculation task of the user equipment UE is divided, a part of which is calculated locally on the user equipment UE, and the rest is offloaded to the access point AP. After the task transfer from the user equipment UE to the access point AP is completed, the calculation task bits received by each access point AP are divided into: processed locally on the access point AP, and migrated to the central server for calculation.

[0016] Furthermore, the local energy consumption E at the kth user equipment UE is calculated by the first layer local calculation loc,k for:

[0017]

[0018] Among them, t loc,k represents the execution time of local computation at the kth user equipment UE, t off represents the unloading time during which the energy station PB and all intelligent reflectors IRS are in the on state, μ represents the power consumption of an element unit of the intelligent reflector IRS, which is positively correlated with the phase resolution of the intelligent reflector IRS, and L represents the number of elements in the IRS;

[0019] The energy consumption E of the mth access point AP AP,m for:

[0020]

[0021] Among them, t AP,m and t mig,m denote the computation time and data migration time at the mth AP, respectively. AP,m represents the information transmission power of the mth AP, ε AP,m The energy consumption coefficient of the processor chip at m APs is closely related to the chip rack. is the CPU frequency of m APs.

[0022] Furthermore, the first-layer user equipment UE adopts IRS communication. When the energy station PB uses a directional antenna to radiate electromagnetic waves, the energy-carrying radio frequency signal reaching the intelligent reflecting surface IRS of each user equipment UE is used for backscatter communication. When the IRS performs backscatter communication, the incident signal as the signal carrier is remodulated, and the backscatter vector θk After modulation, it is converted into a signal x kr The passive beamforming vector θ kr .

[0023] Furthermore, the re-modulation is specifically:

[0024]

[0025] Among them, s represents the original data signal, x kr represents the modulated data signal of the Kth access point AP for the rth UE, w k is the beamforming vector on the kth group of antennas of PB.

[0026] Specifically, in step S3, the data that is unloaded to the access point AP in the first stage and is unloaded from the access point AP to the central server in the second stage is executed; by jointly optimizing the active beamforming at the energy station PB, the passive beamforming at the user equipment UE, the active beamforming at the access point AP, the bandwidth and power allocation between all user equipment UEs, and the local computing time maximization system calculation and bit expression optimization problem, specifically as follows:

[0027]

[0028]

[0029]

[0030]

[0031]

[0032]

[0033]

[0034]

[0035] C8:t off ≤T 1

[0036]

[0037] C10:t AP,m R AP,m +t mig,m R mig,m ≥S m

[0038]

[0039] C12:T 1 +T 2 =T

[0040] Among them, S m represents the calculation and bits received by the mth AP, T 1 and T 2 Represent the duration of the first and second phases, respectively. Respectively represented as w k ,θ kr ,p k ,t loc,k and v m C1 and C2 represent the active and passive beamforming constraints of PB and all UEs respectively. C3, C4, and C5 represent the power allocation or bandwidth allocation constraints between UEs. P represents the total power of PB. C5 aims to ensure fairness between UEs by setting the minimum and maximum bandwidth size limits. C6 is the energy constraint of each UE. represents the energy threshold of all UEs, C7 and C8 are time constraints, C9 represents the beamforming constraint at the AP; C10 is the number of data bits received by each AP that is lower than its local calculation and offloaded data processing capacity; C11 is the time constraint;

[0041] In the first stage, the calculation and bit maximization are used to obtain problem 1. In the second stage, the delay is minimized by problem transformation and alternative parameter optimization to obtain problem 2. Problems 1 and 2 are solved to obtain the calculation and bit and delay minimization. The voice, video and live video calculation problems are decomposed and processed, and the time and calculation tasks are allocated to obtain the time scheduling of multi-layer calculations. The first layer obtains the allocation plan of local calculations and bits and partial data unloading from the first layer to the second and third layers.

[0042] Question 1 is as follows:

[0043]

[0044] stC1-C8

[0045] Question 2 is as follows:

[0046]

[0047] stC9-C11

[0048] Furthermore, question 1 can be equivalently expressed as:

[0049]

[0050]

[0051]

[0052]

[0053]

[0054] Among them, B k is the bandwidth of the channel, α m,k is the auxiliary variable of Lagrangian dual transformation, Ω m,k In order to facilitate matrix calculation, variables are introduced, β m,k is a positive definite auxiliary variable, T m,k is the joint channel gain from k to m, In order to introduce variables to facilitate matrix calculations, is the set of APs, Θ kr for, is the set of UEs, is a set of IRS element units of the UE, and l is the lth element unit.

[0055] Furthermore, question 2 can be equivalently expressed as:

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063] in,(·) H represents the conjugate transpose of a matrix or vector, X mig,m and Y mig,m,r is a variable, R lb for The lower bound of α mig,m is a positive definite auxiliary variable, V m and is a variable, is the set of APs, B is the total bandwidth, S m is the data bit sum, p AP,m is the transmit power of the mth AP.

[0064] In a second aspect, an embodiment of the present invention provides a 6G Internet of Things multi-layer computing system enabled by IRS-BackCom, including:

[0065] System module, establish a multi-layer computing system model of IRS-BackCom to enable 6G Internet of Things;

[0066] The problem module models the multi-layer computing system model of IRS-BackCom empowering 6G IoT obtained by the system module, simplifies the communication problem into a three-layer logical communication model, obtains the computing task scheduling of the user device layer, access point layer, and central processor layer, and obtains the overall communication model of the task allocation scheme for local computing and partial data offloading;

[0067] The computing module describes and decomposes the communication model obtained by the problem module to obtain a three-layer communication model. The local computing and migration computing problems are divided into the user equipment layer, the access point layer and the central processing unit layer for processing. The tasks are scheduled in time, the problems that meet the local computing capacity are calculated locally, the problems that exceed the local computing capacity are partially offloaded for calculation, and the remaining problems are decomposed for calculation, so as to realize the multi-layer computing of 6G Internet of Things enabled by IRS-BackCom.

[0068] Compared with the prior art, the present invention has at least the following beneficial effects:

[0069] An IRS-BackCom-enabled 6G Internet of Things multi-layer computing method is disclosed, which establishes an IRS-BackCom-enabled 6G Internet of Things multi-layer computing system model; the obtained IRS-BackCom-enabled 6G Internet of Things multi-layer computing system model is used for problem modeling to simplify the original complex communication problem into a three-layer logical communication model, and obtain a communication model of time scheduling, local computing and partial data unloading of a user device layer, an access point layer and a central processing unit layer; the original complex task allocation problem is transformed into two computing problems of maximizing computing and minimizing bits and delays; and the system problem is expressed and decomposed to obtain an optimized computing result.

[0070] Furthermore, the originally complex local computing and migration computing problems can be divided into three layers for processing. The tasks are allocated and processed at the user device layer, access point layer, and central processing unit layer, and the tasks are reasonably scheduled at a certain time. Simple problems are calculated locally, and problems that exceed the local computing capacity are partially unloaded for computing. Complex computing problems are decomposed and efficiently calculated. This can improve the overall computing efficiency of the system, reduce the burden of computing tasks at each layer, and thus improve the overall computing efficiency.

[0071] Furthermore, a partial data offloading strategy is adopted, and all user equipment UE and access point AP perform calculation offloading and local calculation at the same time, and the calculation tasks of the user equipment UE are divided, a part of which is calculated locally in the user equipment UE, and the rest is offloaded to the access point AP. After the task transfer from the user equipment UE to the access point AP is completed, the calculation task bits received by each access point AP are divided into: processed locally in the access point AP, and migrated to the central server for calculation, thereby reasonably using the local computing power and the upper layer computing power. When the computing task throughput is relatively large, the system has stronger computing efficiency and computing power, and can adapt to high-concurrency, high-throughput computing tasks. Through partial data offloading and local calculation, computing tasks and computing resources can be fully utilized.

[0072] Furthermore, according to the establishment of the above model, the local energy consumption E at the kth user equipment UE is calculated respectively. loc,k and the energy consumption E of the mth access point AP AP,m The maximum power consumption limit of each part is then calculated, so as to obtain reasonable time scheduling and task allocation for local computing and task offloading.

[0073] Furthermore, when the energy station PB radiates electromagnetic waves using a directional antenna, the energy-carrying RF signal reaching the intelligent reflective surface IRS of each user device UE is used for backscatter communication. When the IRS performs backscatter communication, the incident signal as a signal carrier is remodulated, and the backscatter vector θ k After modulation, it is converted into a signal x kr The passive beamforming vector θ kr The intelligent reflective surface IRS can collect energy and resend information through modulation, thereby achieving low-power communication capabilities.

[0074] Furthermore, the IRS intelligent reflector remodulates and sends the communication signal, which can reduce the power consumption of each information transmission and improve the communication efficiency of the system itself.

[0075] Furthermore, the first stage is offloaded to the access point AP, and the data offloaded from the access point AP to the central server is executed in the second stage; by jointly optimizing the active beamforming at the energy station PB, the passive beamforming at the user equipment UE, the active beamforming at the access point AP, the bandwidth and power allocation between all user equipment UE, and the local computing time maximization system calculation and bit expression optimization problem, in the first stage, the calculation and bit maximization are obtained to obtain problem 1, and in the second stage, the delay minimization is calculated by problem transformation and alternative parameter optimization method to obtain problem 2. The original complex problem is transformed to facilitate the expression and calculation of the problem.

[0076] Furthermore, the calculation and bit maximization of problem 1 are equivalently expressed, and variables are introduced using Lagrange duality for secondary transformation. Thus, the original complex problem is solved using methods such as parameter substitution, partial derivative, and Gaussian randomization to obtain a solution with lower complexity.

[0077] Furthermore, the delay minimization of problem 2 is expressed equivalently, and the quadratic optimization is used to replace the optimization method for parameter optimization to obtain the optimal solution.

[0078] It can be understood that the beneficial effects of the second to third aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0079] In summary, the present invention makes the originally complex computing task problem more efficient and reasonable in time scheduling and partial task offloading, and fully utilizes the optimal computing capabilities of user equipment, access points, and central processing units for task computing, greatly improving the efficiency of the entire communication system.

[0080] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 It is a schematic diagram of the IRS-BackCom multi-layer computing system;

[0082] Figure 2 It is a time scheduling diagram for the two-stage process;

[0083] Figure 3 Schematic diagram of optimization of IRS-BackCom multi-layer calculation, where (a) is the offloading rate from UE to AP, and (b) is the offloading rate from AP to the central server;

[0084] Figure 4 Schematic diagram of the impact of the number of IRS elements calculated for the IRS-BackCom multi-layer on the sum of bits calculated by the user equipment, where (a) is the offloading rate from the UE to the AP, and (b) is the sum of bits calculated;

[0085] Figure 5 Schematic diagram of the impact of the total transmit power change of the energy station calculated for the IRS-BackCom multi-layer on the offloading rate of the user equipment and the access point rate system calculation and bits, where (a) is the offloading rate from the UE to the AP, and (b) is the calculation and bits;

[0086] Figure 6Schematic diagram of the impact of the average distance from the user end to the access point on the offloading rate and the calculation bits and bits of the system for IRS-BackCom multi-layer calculation, where (a) is the offloading rate from UE to AP, and (b) is the calculation and bits;

[0087] Figure 7 Schematic diagram of the impact of the number of access points and the number of antennas on the total number of system bits for IRS-BackCom multi-layer calculation, where (a) is the number of access points and (b) is the number of access point antennas;

[0088] Figure 8 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0089] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0090] In the description of the present invention, it should be understood that the terms “include” and “comprises” indicate the presence of described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0091] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0092] It should be further understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects are in an "or" relationship.

[0093] It should be understood that, although the terms first, second, third, etc. may be used to describe preset ranges, etc. in the embodiments of the present invention, these preset ranges should not be limited to these terms. These terms are only used to distinguish preset ranges from each other. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0094] The word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0095] Various structural schematic diagrams of the embodiments disclosed in the present invention are shown in the accompanying drawings. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0096] The present invention provides a 6G Internet of Things multi-layer computing method enabled by IRS-BackCom. In a hierarchical network, data of a computing task requested by each user equipment UE (User Equipment) is divided into three parts, which are respectively calculated on the UE of the T1 layer, the access point AP (Access Point) of the T2 layer, and the central server of the T3 layer; compared with the traditional active antenna transmission method, the UE utilizes a passive intelligent reflecting surface IRS (Intelligent Reflecting Surface) to unload data bits to the access point AP; completely different from the traditional active antenna method; based on an established network framework; the present invention aims to maximize the optimization problem of system computing and bits; within the considered time block, the present invention maximizes the system computing and bits by jointly optimizing active beamforming at an energy station, passive beamforming at a UE, active beamforming at an AP, bandwidth and power allocation between all UEs, and local computing time.

[0097] The present invention decomposes the optimization problem into two sub-problems, namely, the problem of maximizing the computation and bits in time stage 1 and the problem of minimizing the delay in time stage 2. The solutions of the two sub-problems are obtained by using the objective function conversion and the alternating optimization method. The present invention performs a large number of simulations to verify the feasibility of the system and shows that the system can achieve high-performance computation and bits in processing computation bits.

[0098] See also Figure 8 The present invention provides an IRS-BackCom-enabled 6G Internet of Things multi-layer computing method, comprising the following steps:

[0099] S1. Establish a multi-layer computing system model of 6G IoT enabled by intelligent reflective surface backscatter (IRS-BackCom);

[0100] See also Figure 1 The multi-layer computing system model based on IRS-BackCom to enable 6G IoT includes a power station PB (Power Beacon), K intelligent reflector IRS-assisted user equipment UE on the T1 layer, M access points AP on the T2 layer connected to a MEC server, and a central server with rich resources on the T3 layer. PB, each unified AP and central server are equipped with N p ,N a and N c Antennas, N c ≥M×N a .

[0101] make denote the set of UE, IRS element unit of k-th UE and AP respectively. In such a hierarchical network, each UE is requesting to perform a computational task. It is assumed that each task is bit-independent and can be divided into multiple bit subsets. All control links used for information exchange between all entities are fully smooth. And all channels follow quasi-static fading and the channel information is perfectly known.

[0102] The system adopts a partial data offloading strategy. For all UEs and APs, their computing units and offloading units are separated in circuit structure, thus supporting both computing offloading and local computing. The computing power of the UE itself is very weak, and its data processing speed is quite limited. In order to achieve efficient computing and bit and low latency, the UE's computing tasks are divided into multiple parts, one part is calculated locally at the UE, and the rest is offloaded to the AP. Once the task transfer from the UE to the AP is completed, the computing task bits received by each AP are further divided into two parts, one part is processed locally at the AP, and the other part is migrated to the central server for calculation.

[0103] S2, based on the IRS-BackCom 6G IoT-enabled multi-layer computing system model obtained in step S1, problem modeling is performed;

[0104] Based on the intelligent reflective surface backscattering-enabled 6G IoT multi-layer computing system model established in step S1, a partial data offloading strategy is adopted. For all UEs and APs, their computing units and offloading units are separated in the circuit structure, so computing offloading and local computing are supported simultaneously.

[0105] Due to the low computing power, the data processing speed of each UE is quite limited. In order to achieve efficient computing and low latency, the computing task of the UE is divided into multiple parts. One part is calculated locally at the UE, and the rest is offloaded to the AP. Once the task transfer from UE to AP is completed, the computing task bits received by each AP are further divided into two parts. One part is processed locally at the AP, and the other part is migrated to the central server for calculation.

[0106] S201, establish calculation and energy model;

[0107] The calculation formula for the data calculation time and data migration time is the time and energy consumed in transmitting at different computing layers.

[0108] When a UE requests to process a computing task, due to the low computing power, only a small part of the task bits are calculated locally; then after the task is offloaded from the UE to the AP, the data received by the AP is divided into two parts. One part is calculated locally, while the other part is migrated to the central server. Let f loc,k and C loc,k They represent the CPU frequency and the number of cycles required to calculate a unit bit at the kth UE. loc,k It means that the energy consumption coefficient of the processor chip at the kth UE is closely related to the chip rack. Let f AP,m ,C AP,m , ε AP,m Indicates the corresponding content of the mth AP.

[0109] The calculation rates performed at the kth UE and the mth AP are expressed as:

[0110]

[0111]

[0112] Because the central server has powerful computing power and the execution delay is negligible, the local energy consumption at the kth UE consists of data calculation and IRS operation, which can be expressed as:

[0113]

[0114] Among them, t loc,k represents the execution time of local computation at the kth UE, t off represents the unloading time (during the unloading period, PB and all IRSs are turned on), μ represents the power consumption of an element unit of a single IRS, and is positively correlated with the phase resolution of the IRS.

[0115] As the number of IRS elements increases, the energy consumed by IRS also increases. The energy consumption at the mth AP is given by:

[0116]

[0117] Among them, t AP,m and t mig,m denote the computation time and data migration time at the mth AP, respectively. AP,m Indicates the information transmission power of the mth AP.

[0118] S202. Establishing a communication model

[0119] The scale of the computation results of this system is generally much smaller than the computation task, so downlink communication will not be considered; and the return delay can also be reasonably ignored. For uplink task offloading, consider a time block T during which all channel gains remain unchanged. When performing task offloading, the time block T is divided into two phases T 1 and T 2 Two time periods, such as Figure 2 As shown. In the first order T 1 Offload data tasks from UE to AP; in the second stage T 2 , part of the task bits received by the AP are migrated to the central server; note that these two stages share the same spectrum resource B.

[0120] The electromagnetic waves emitted by the energy station PB deployed in the first phase are used as backscattered carrier signals at the IRS at each UE. Each UE occupies different spectrum resources to transmit some of its data bits to the AP; specifically, the total spectrum resource bandwidth B is divided into K resource blocks Among them, B k is allocated to the kth UE. The antennas at the PB are divided into k groups.

[0121] Among them, N p,k represents the number of antennas in the kth group, The kth group of antennas points to the kth UE and shares the same spectrum resource B with the kth UE. k .

[0122] make and They represent the channel gain matrices from the kth antenna group at the PB to the kth UE, from the kth antenna group at the PB to the mth AP, and from the kth UE to the mth AP, respectively.

[0123] The signal transmission from PB to the central server is negligible. Definition Θ k and θ k are the IRS k-th backscattering matrix and vector, Θ k =diag{θ k}.

[0124] When the PB transmits radiated electromagnetic waves using directional antennas, the energy-carrying RF signal reaching the IRS of each UE is used for backscatter communication.

[0125] When an IRS performs backscatter communication, the incident signal as the signal carrier is remodulated. This process is mathematically described as:

[0126]

[0127] Among them, s represents the original data signal, x kr represents the data signal modulated by the Kth AP for the rth UE, Symbol w k Refers to the beamforming vector on the kth antenna of PB. Backscattering vector θ k After modulation, it is converted into a signal x kr The passive beamforming vector θ kr ,and [·] l,l Represents the lth diagonal element of the matrix.

[0128] A partial data offloading strategy is used to reasonably distribute computing tasks among UE, AP and central server. For efficient computing and low latency, the computing tasks of UE are divided into multiple parts. One part is calculated locally at UE, and the rest is offloaded to AP. Once the task transfer from UE to AP is completed, the computing task bits received by each AP are further divided into two parts. One part is processed locally at AP, and the other part is migrated to the central server for calculation.

[0129] S3. In the multi-layer computing system model of IRS-BackCom enabling 6G Internet of Things established in step S1, problem modeling is performed based on step S2 to express and decompose the system problem.

[0130] The method of jointly optimizing active beamforming at PB, passive beamforming at UE, active beamforming at AP, bandwidth and power allocation among all UEs, and local computation time respectively makes it difficult to directly solve the system's computation and bit maximization problem and the execution time of local computation as well as the time allocation between the two stages. Therefore, it is proposed that the time block is divided into two interrelated continuous stages.

[0131] In the first phase, some computational bits are offloaded to the AP. In the second phase, the data offloaded from the AP to the central server is executed. Corresponding to these two phases, the problem can be divided into two problems, namely, the maximization of computation and bits in phase 1 and the minimization of delay in phase 2. The problem is solved by problem transformation and alternative parameter optimization methods.

[0132] S301. Formulate the problem

[0133] Considering the migration of computing tasks and data through two processes, when the migration of computing tasks from AP to central server is completed, the computing results can be obtained and returned immediately due to the powerful computing power of central server. In order to process as many computing bits as possible in the time block, the optimization problem is deduced by jointly optimizing active beamforming at PB, passive beamforming at UE, active beamforming at AP, bandwidth and power allocation among all UEs, and local computing time to maximize computing and bits;

[0134] In the first stage, signals from different UEs are received at the mth AP, and only the transmission from the kth UE is considered. The received signal at the mth AP is:

[0135]

[0136] in, is a complex Gaussian random vector with power spectral density N m ,and

[0137] Since the antenna at the PB points to each UE, the AP is rarely interfered by the PB. k The signal-to-noise ratio (SINR) at the mth AP on :

[0138]

[0139] Therefore, the sum rate from the kth UE to all APs is:

[0140]

[0141] In the second stage, all APs occupy the same spectrum resource B and send part of the received task bits to the central server through the wireless return link. represents the channel gain matrix from AP to the central server. The signal received at the central server is:

[0142]

[0143] Among them, n~(0,σ 2 I) is a complex Gaussian random vector with power spectral density N and σ 2 =BN. The transmission rate from the mth AP to the central server according to the received signal is:

[0144] R mig,m =B log 2 (1+γ mig,m )

[0145] in,

[0146] Due to the powerful computing power of the central server, when the computing task migration from the AP to the central server is completed, the computing result can be obtained and returned immediately. In order to process as many computing bits as possible in the considered time block, the present invention maximizes the computing and bits of the system by jointly optimizing the active beamforming at the PB, the passive beamforming at the UE, the active beamforming at the AP, the bandwidth and power allocation between all UEs, and the local computing time.

[0147] The optimization problem is formulated as:

[0148]

[0149]

[0150]

[0151]

[0152]

[0153]

[0154]

[0155] C8:t off ≤T 1 ,

[0156]

[0157] C10:t AP,m R AP,m +t mig,m R mig,m ≥S m,

[0158]

[0159] C12:T 1 +T 2 =T,

[0160] in, represents the number of bits received by the mth AP. 1 and T 2 Represent the duration of the first stage and the second stage respectively. Respectively represented as w k ,θ kr ,p k ,t loc,k and v m C1 and C2 represent the active and passive beamforming constraints of PB and all UEs respectively. C3, C4, and C5 represent the power allocation or bandwidth allocation constraints between UEs, where P represents the total power of PB. C5 aims to ensure fairness between UEs by setting the minimum and maximum bandwidth upper and lower bounds. C6 is the energy constraint of each UE. represents the energy threshold of all UEs; C7 and C8 are time constraints; C9 represents the beamforming constraint at the AP; C10 is established because the number of data bits received by each AP does not exceed its local calculation and offloaded data processing capacity; C11 is a time constraint.

[0161] Furthermore, the energy budget of the AP is not an important limiting factor and is therefore not considered as a constraint.

[0162] S302, decomposing the problem after stating the problem in step S301;

[0163] On the basis of establishing a multi-layer computing model, the original complex and difficult three-layer communication model's local computing tasks and partial data offloading task allocation scheme are transformed into two easily solvable sub-problems; namely, the problem of maximizing computing and bits and minimizing latency; this ensures that the overall communication capacity of the system is fully utilized to cope with high throughput; and the communication efficiency of the system is improved during the peak communication period.

[0164] The non-convex optimization problem (P0) is difficult to solve directly due to the presence of multiple coupled variables. These coupled variables are the active and passive beamforming vectors, power and bandwidth, local computation execution time, and the time allocation between the two phases. In the system considered, the time block is divided into two interrelated consecutive phases. In the first phase, some computation bits are offloaded to the AP and the corresponding constraints are C1-C8. In the second phase, the data offloaded from the AP to the central server is executed, and the corresponding constraints are C9-C11. Corresponding to these two phases, the problem (p0) is divided into the following two problems:

[0165] That is, the calculation and bit maximization of stage 1 and the delay minimization of stage 2.

[0166] For a given time allocation between two stages, the two optimization problems Problem 1 and Problem 2 can be solved by bisection to obtain T 1 and T 2 The optimal time allocation between . The solution to problem P1 can be obtained by solving problems (P1) and (P2) and using the bisection method.

[0167] Problem 1: Phase 1 computation and bit maximization:

[0168]

[0169] stC1-C8

[0170] Problem 2: Minimizing the latency of Phase 2:

[0171]

[0172] stC9-C11.

[0173] For a given time allocation, the solution process of the above two optimization problems is given respectively; then, the optimal time allocation between T1 and T2 can be obtained by bisection. In other words, by solving problems P1 and P2, the solution of the original problem P0 can be obtained by using bisection.

[0174] From question (P1), we can see that when given When , problem (P1) is simplified to a linear programming problem, and the objective function is only subject to constraints C6-C8. Considering that a linear programming problem is easy to solve, the solution method is omitted here. On the other hand, Depends on They only concern constraints C1-C5. So focus on the following questions.

[0175]

[0176] stC1-C5

[0177] In problem (P3), active beamforming, passive beamforming, power and bandwidth are coupled. In addition, the objective function is the sum of logarithmic functions. In view of these factors, this problem (P3) is still difficult to solve directly. In order to make problem (P3) feasible, an effective solution is proposed. The present invention first converts the sum of logarithmic functions into a form that is easier to handle, and then proposes an alternating optimization method to optimize each variable.

[0178] S3021, based on step S302, the objective function is converted by using a secondary conversion method for the first stage calculation and the bit maximization problem;

[0179] In order to transform the logarithmic function into a more tractable form, the Lagrange dual transformation is used to transform the objective function. The auxiliary variable α is introduced m,k and auxiliary variable vector β m,k , and then using the quadratic transformation, the sum of the logarithmic functions is rewritten as:

[0180]

[0181] in

[0182]

[0183] in

[0184]

[0185] A m,k =T m,k θ km

[0186]

[0187] Based on the new objective function, the problem (p3) is newly formulated as:

[0188]

[0189] stC1-C5

[0190]

[0191] S3022, based on step S302, alternately optimizing the problem parameters using methods such as derivative quadratic programming (QCQP) semidefinite relaxation (SDR) for the first stage calculation and bit maximization problem one;

[0192] Optimizing the variable α through a loop m,k ,β m,k , Solving the problem (P4) includes the following steps:

[0193] Step 1: Optimize α m,k and β m,k ;

[0194] Given In the case of m,k and β m,k Derivation to obtain the optimal and make

[0195] Then derive the optimal They are:

[0196]

[0197] Step 2: Optimization

[0198] When α is given m,k ,β m,k , When , problem (P4) can be reformulated as:

[0199]

[0200] stC4,C5

[0201] Obviously, (P5) is a quadratic programming (QCQP) problem and is therefore easy to solve.

[0202] Step 3: Optimization and

[0203] Given α m,k ,β m,k , After that, the objective function of the problem (P4) can be simplified as:

[0204]

[0205] Y k x k They are:

[0206]

[0207]

[0208] Then, question (P4) is reformulated as:

[0209]

[0210] stC1,C3

[0211] Use semidefinite relaxation (SDR) to lift it to a higher dimension, and Then the problem is rewritten equivalently as:

[0212]

[0213]

[0214]

[0215]

[0216]

[0217]

[0218] By removing the rank-1 constraint C16, (P6) is relaxed to a semidefinite programming (SDP) problem, which can be easily solved using the existing CVX solver. Then, the rank-1 solution can be recovered by singular value decomposition (SVD) or Gaussian randomization methods.

[0219] Step 4: Optimization

[0220] Given α m,k ,β m,k , The objective function of problem (P4) is simplified to:

[0221]

[0222] Then the problem (P4) is reformulated as:

[0223]

[0224] stC2

[0225] Derived:

[0226]

[0227] in,

[0228] So problem (P4) can be restated as:

[0229]

[0230]

[0231]

[0232]

[0233]

[0234] Ignoring the rank 1 constraint C20, this problem is is convex and easy to solve. According to its optimal solution The rank 1 solution can be restored using singular value decomposition (SVD) or Gaussian randomization. Based on this, we can get θ kr .

[0235] Problem 2: Minimizing the latency of Phase 2:

[0236]

[0237] stC9-C11

[0238] S3023, based on step S302, transform the delay minimization problem of stage 2;

[0239] Problem (P2) is equivalently rewritten as:

[0240]

[0241] stC9,C10

[0242] From this optimization problem, it can be easily deduced that t AP,m =t mig,m is a necessary condition for the optimal solution. To prove this, let t AP,m ≤t mig,m ; According to constraint C10, When t AP,m When the value is large, t mig,m Therefore, at t AP,m =t mig,m hour, is minimized.

[0243] The optimization problem (P2) can be further expressed as:

[0244]

[0245]

[0246]

[0247]

[0248] For a more concise problem, the problem can be expressed as:

[0249]

[0250]

[0251] Consider having The problem is equivalently transformed into:

[0252]

[0253]

[0254]

[0255] According to the left side of the quadratic transformation constraint C23, it is expressed as:

[0256]

[0257] in, Then the problem (P2) is transformed into:

[0258]

[0259]

[0260]

[0261] S3024. Based on step S302, the delay minimization problem of stage 2 is optimized by using derivation and semi-positive definite relaxation.

[0262] By optimizing the variable α mig,m β mig,m and Solving the problem (P8), this process is divided into the following steps:

[0263] Step 1: Optimize α mig,m β mig,m ;

[0264] Given For α mig,m β mig,m Derivation to find the optimal Then derive the optimal:

[0265]

[0266]

[0267] Step 2: Optimization

[0268] Given α mig,m β mig,m , constraint C24 is equivalent to:

[0269]

[0270] make Constraint C24 is transformed into:

[0271]

[0272] Then problem (P8) is simplified to:

[0273]

[0274] stC9,C25

[0275] By performing a semidefinite relaxation (SDR), this problem is equivalently transformed to:

[0276]

[0277]

[0278]

[0279]

[0280]

[0281]

[0282]

[0283] in,

[0284] Ignoring the rank 1 constraint C30, (P9) becomes a convex problem. According to its optimal solution The rank 1 solution can be recovered by using SVD or Gaussian randomization method; on this basis, we get

[0285] In another embodiment of the present invention, an IRS-BackCom-enabled 6G Internet of Things multi-layer computing system is provided, which can be used to implement the above-mentioned IRS-BackCom-enabled 6G Internet of Things multi-layer computing method. Specifically, the IRS-BackCom-enabled 6G Internet of Things multi-layer computing system includes a system module, a problem module and a computing module.

[0286] Among them, the system module establishes a multi-layer computing system model of IRS-BackCom to enable 6G Internet of Things;

[0287] The problem module models the multi-layer computing system model of IRS-BackCom empowering 6G IoT obtained by the system module, simplifies the communication problem into a three-layer logical communication model, obtains the computing task scheduling of the user device layer, access point layer, and central processor layer, and obtains the overall communication model of the task allocation scheme for local computing and partial data offloading;

[0288] The computing module describes and decomposes the communication model obtained by the problem module to obtain a three-layer communication model. The local computing and migration computing problems are divided into the user equipment layer, the access point layer and the central processing unit layer for processing. The tasks are scheduled in time, the problems that meet the local computing capacity are calculated locally, the problems that exceed the local computing capacity are partially offloaded for calculation, and the remaining problems are decomposed for calculation, so as to realize the multi-layer computing of 6G Internet of Things enabled by IRS-BackCom.

[0289] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention described and shown in the drawings here can usually be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0290] The system of the present invention considers the problem of maximizing the calculation and bits in the time block and minimizing the delay. To solve this problem, the present invention jointly optimizes the active beamforming at the PB, the passive beamforming of the UE, the active beamforming at the AP, the bandwidth and power allocation between all UEs, and the time of local calculation.

[0291] In the considered multi-layer computing system, the offloading rate from UE to AP and the calculation and bits of the system can be expressed as the communication capacity of the proposed IRS backscattering strategy and the task processing capacity of the system, respectively. Therefore, the present invention studies how these two performance indicators depend on several important parameters, including the number of elements of the IRS, the total transmit power of the PB, the average distance from the UE to the AP, the number of APs, and the number of antennas of the AP. In addition to the proposed optimization scheme, several simplified optimization schemes are also given for comparison. The achievable communication performance of the backscatter multi-layer computing optimization is evaluated by numerical simulation below.

[0292] Joint: The figure represents the optimization scheme for the considered IRS backscatter-supported multi-layer computing system, where active beamforming at the PB, passive beamforming at the UE, active beamforming at the AP, bandwidth and power allocation among all UEs, and time for local computation are all jointly optimized.

[0293] Active: This figure represents a simplified optimization scheme for the considered IRS backscatter-supported multi-layer computing system, where passive beamforming is randomly generated at the UE and other variables including active beamforming at the PB, active beamforming at the AP, bandwidth and power allocation among all UEs, and local computation time are jointly optimized.

[0294] Passive: In the considered IRS backscatter supported multi-layer computing system, the PB may not be a dedicated deployed BS. Instead, existing surrounding signal stations can also be used as PBs. In this case, the active beamforming at the PB is randomly generated. This figure represents this case, where all variables are jointly optimized except the active beamforming at the PB.

[0295] Random Time: This illustration represents another simplified optimization scheme for the considered IRS backscatter-supported multi-layer computing system, where the time allocation between the two stages is not optimized. Instead, the time is randomly divided, which is the only difference from the joint scheme.

[0296] Algorithm 1 summarizes the overall solution process of problem (P0), where t represents the tth iteration and ε represents a small positive precision limit or iteration precision. Algorithm 1 is convergent. It can be clearly seen from Algorithm 1 that there are three repeat-end loops. In the first repeat-end loop, the computational complexity mainly depends on the solution process of (P7). Because A m,k ,B m,k ,α m,k β m,k The solution can be obtained quickly without complex optimization process. In addition, the number of elements in IRS far exceeds the number of antennas at AP. By using the interior point method (IPM), the complexity of problem (P7) is given as:

[0297]

[0298] in, The nested repeat-end loop (P9) dominates the computational complexity. Using IPM, the complexity of problem (P9) is:

[0299]

[0300] in,

[0301] After solving (p7) and (p9), the solution is generally a matrix of rank 1, so the complexity is small and the vector of rank 1 can be ignored. Therefore, the total complexity of the problem (p0) is approximately:

[0302] C P0 =t loop1 C P7 +t loop2 n 3 C P9

[0303]

[0304]

[0305] Among them, t loop1 t loop2 Respectively represent the number of loops of the first and third repeat-end loops, Represents the complexity of the equal division method in the second loop.

[0306] In the numerical simulation, all channel data are randomly generated under the Rice distribution with factor κ. For all channels, the path loss is expressed as PL = PL 0 -25lg(d / d 0 )dB,PL 0 Indicates the reference distance d 0 = d represents the path loss at , and d represents the transmission distance; the number of antenna groups on all PBs is the same, and the antenna gain of each antenna group is represented by η. Only a small part of the electromagnetic power emitted from the PB reaches each AP. In addition, the number of elements of all IRSs is the same. Some important simulation parameter settings are listed in Table 1. The distances from PB to UE, from UE to AP, from PB to AP to AP, and from AP to the central server are spaced in the interval [d pu0 -10m,d pu0 +10m], [d ua0 -20m,d ua0 +20m], [d pa0 -20m,d pa0 +20m] and [d ac0 -20m,d ac0 +20m] are generated uniformly at random.

[0307] Table 1. Simulation parameter values

[0308]

[0309]

[0310] See also Figure 3 , showing the convergence behavior of all optimization solutions under random observations. Figure 3 (a) and Figure 3 (b) It can be clearly observed that all optimization solutions converge very quickly. Figure 3 In (a), the “joint” optimization scheme has the highest offloading rate from UE to AP. Figure 3 As can be seen in (b), when data is offloaded from AP to the central server, the proposed optimization scheme does not achieve the best performance.

[0311] See also Figure 4 , describes the relationship between the offload rate from UE to AP and the calculation of the system and the number of bits and elements of the IRS on the UE. Figure 4 (a) and 4(b), it can be found that as the number of IRS elements increases, the system offloading rate and the number of computation bits also increase. This result shows that the increase in the number of IRS elements helps to improve the system performance. In addition, the performance of the joint scheme is better than that of the active and passive schemes. Figure 4 In (b), the system computation and bits obtained by the random time scheme are smaller than the joint state, indicating that the time allocation between t1 and t2 plays a crucial role in improving the performance of the entire system.

[0312] See also Figure 5 and Figure 6 , respectively, shows how the total transmit power of the PB and the average distance from the UE to the AP affect the offload rate and the computation and bits from the UE to the AP. It can be seen from the figure that the increase in the total transmit power at the PB and the decrease in the average distance from the UE to the AP help improve the offload rate and system computation and bits from the UE to the AP.

[0313] See also Figure 7 , showing how the number of APs and their number of antennas affects computation and bits. Figure 7 It is not difficult to see in (a) that the offloading rate from UE to AP and the system's computation and bits increase as more APs are deployed.

[0314] In summary, the present invention provides an IRS-BackCom-enabled 6G Internet of Things multi-layer computing method and system, which divides the computing process into the user end, the access point, and the central server, and improves the computing efficiency by optimizing each process, thereby achieving the purpose of low power consumption; solves the problem of maximizing the calculation and in the time block; optimizes the active beamforming of the energy station, the passive beamforming at the user end, the active beamforming at the access point, the bandwidth and power allocation between all user devices, and the time of local calculation. Simulation shows that increasing the number of IRS elements, the total transmit power, the number of access points, and the number of antennas at the access point can promote computing and bits. On the contrary, the performance will decrease when the average distance from the user device to the access point increases; the results show that the IRS-BackCom multi-layer computing system is efficient and feasible, which can greatly reduce the communication pressure of communication equipment and provide a reliable, efficient, and low-power communication network.

Claims

1. A 6G Internet of Things multi-layer computing method enabled by IRS-BackCom, characterized in that: The following steps are involved: S1. Establish a multi-layer computing system model of IRS-BackCom to enable 6G Internet of Things; S2. Modeling the multi-layer computing system model of IRS-BackCom empowering 6G Internet of Things obtained in step S1, simplifying the communication problem into a three-layer logical communication model, obtaining the overall communication model of the computing task scheduling of the user device layer, access point layer, and central processor layer, and obtaining the task allocation scheme of local computing and partial data offloading; The first layer of local calculation calculates the local energy consumption at the kth user equipment UE for: in, represents the execution time of local computation at the kth user equipment UE, Indicates the unloading time. During the unloading period, the energy station PB and all intelligent reflective surfaces IRS are turned on. It means that the power consumption of an element unit of the smart reflector IRS is positively correlated with the phase resolution of the smart reflector IRS. is the number of elements in IRS, is the CPU frequency, The energy consumption coefficient of the processor chip at the kth UE is closely related to the chip rack; No. The energy consumption of an access point AP for: in, and Respectively expressed in The computation time and data migration time at each AP, Indicates The information transmission power of each AP is The energy consumption coefficient of the processor chip at m APs is closely related to the chip rack. is the CPU frequency of m APs; S3. The communication model obtained in step S2 is described and decomposed into a three-layer communication model. Local computing and migration computing problems are assigned to the user device layer, access point layer, and central processing unit layer for processing. Tasks are scheduled in time, problems that meet local computing capabilities are calculated locally, problems that exceed local computing capabilities are partially calculated by data offloading, and the remaining problems are decomposed and calculated to achieve multi-layer computing of 6G IoT enabled by IRS-BackCom.

2. The IRS-BackCom-enabled 6G IoT multi-layer computing method according to claim 1, characterized in that: In step S1, the multi-layer computing system model includes: A power station PB, K user equipment UEs assisted by intelligent reflectors IRS on the T1 layer, M access points APs connected to a MEC server on the T2 layer, and a central server with rich resources on the T3 layer; the power station PB, each unified access point AP and the central server are equipped with Antenna, ,In a hierarchical network, each user equipment UE requests to perform a computing task, each task is bit-by-bit independent and ,split into multiple bit subsets, all channels follow quasi-static fading and the channel information is known.

3. The IRS-BackCom-enabled 6G IoT multi-layer computing method according to claim 1, characterized in that: In step S2, a partial data offloading strategy is adopted, and all first-layer user equipment UE and second-layer access point AP perform calculation offloading and local calculation at the same time, and the calculation task of the user equipment UE is divided, a part of which is calculated locally on the user equipment UE, and the rest is offloaded to the access point AP. After the task transfer from the user equipment UE to the access point AP is completed, the calculation task bits received by each access point AP are divided into: being processed locally on the access point AP, and being migrated to the central server for calculation.

4. The IRS-BackCom-enabled 6G Internet of Things multi-layer computing method according to claim 1, characterized in that: The first layer of user equipment UE uses IRS communication. When the energy station PB uses a directional antenna to radiate electromagnetic waves, the energy-carrying radio frequency signal reaching the intelligent reflective surface IRS of each user equipment UE is used for backscatter communication. When the IRS performs backscatter communication, the incident signal as the signal carrier is remodulated, and the backscatter vector After modulation, it is converted into a signal The passive beamforming vector .

5. The IRS-BackCom-enabled 6G Internet of Things multi-layer computing method according to claim 4 is characterized in that: The re-modulation is as follows: in, and denote the channel gain matrices from the k-th antenna group at the PB to the k-th UE and from the k-th UE to the m-th AP, respectively. is the kth backscattering matrix of IRS, represents the original data signal, , represents the modulated data signal of the Kth access point AP for the rth UE, , is the beamforming vector on the kth group of antennas of PB.

6. A 6G IoT multi-layer computing system enabled by IRS-BackCom, characterized in that: include: System module, establish a multi-layer computing system model of IRS-BackCom to enable 6G Internet of Things; The problem module models the multi-layer computing system model of IRS-BackCom empowering 6G IoT obtained by the system module, simplifies the communication problem into a three-layer logical communication model, obtains the computing task scheduling of the user device layer, access point layer, and central processor layer, and obtains the overall communication model of the task allocation scheme for local computing and partial data offloading; The first layer of local calculation calculates the local energy consumption at the kth user equipment UE for: in, represents the execution time of local computation at the kth user equipment UE, Indicates the unloading time. During the unloading period, the energy station PB and all intelligent reflective surfaces IRS are turned on. It means that the power consumption of an element unit of the smart reflector IRS is positively correlated with the phase resolution of the smart reflector IRS. is the number of elements in IRS, is the CPU frequency, The energy consumption coefficient of the processor chip at the kth UE is closely related to the chip rack; No. The energy consumption of an access point AP for: in, and Respectively expressed in The computation time and data migration time at each AP, Indicates The information transmission power of each AP is The energy consumption coefficient of the processor chip at m APs is closely related to the chip rack. is the CPU frequency of m APs; The computing module describes and decomposes the communication model obtained by the problem module to obtain a three-layer communication model. The local computing and migration computing problems are divided into the user equipment layer, the access point layer and the central processing unit layer for processing. The tasks are scheduled in time, the problems that meet the local computing capacity are calculated locally, the problems that exceed the local computing capacity are partially offloaded for calculation, and the remaining problems are decomposed for calculation, so as to realize the multi-layer computing of 6G Internet of Things enabled by IRS-BackCom.

Images