A multi-user security offloading method based on backscatter assisted edge computing

Abstract

The application provides a multi-user security offloading method based on backscatter-assisted edge computing, which comprises the following steps: obtaining the task offloading parameters of each user, the current user grouping, the randomized continuous wave parameters, the channel parameters of the backscatter communication stage, the channel parameters of the active transmission stage, and the eavesdropper channel parameters; determining the secrecy rates of each user in the backscatter communication stage and the active transmission stage according to the above parameters; determining the total task processing delay time, the total energy consumption of the user, and the energy collected by the user in the backscatter communication stage according to the secrecy rate of the backscatter communication stage and the secrecy rate of the active transmission stage; and minimizing the total task processing delay time by taking the total energy consumption of the user and the energy collected by the user in the backscatter communication stage as constraints, and jointly optimizing the randomized continuous wave parameters, the task offloading parameters, and the user grouping. Through the implementation of the application, the security performance and the task processing efficiency in the multi-user scenario are enhanced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of task offloading, and particularly relates to a multi-user security offloading method based on backscattering assisted edge computing. BACKGROUND

[0002] The rapid development of the Internet of Things promotes the emergence of various innovative applications, such as virtual reality (VR), Internet of Vehicles (IoV), smart home, smart city, etc. These new applications lead to a substantial increase in computationally intensive tasks. Fortunately, mobile edge computing (MEC) can provide additional computing resources for Internet of Things wireless devices (users). However, the energy and signal processing capabilities of wireless devices are usually limited by limited manufacturing costs, which cannot smoothly offload tasks. In order to cope with this challenge, wireless-powered mobile edge computing (WPMEC) has received extensive attention. WPMEC combines wireless power transmission (WPT) with mobile edge computing, allowing wireless devices to harvest energy from the radio frequency (RF) signals emitted by a specific energy source (such as a power beacon (base station)) and use the harvested energy for local computing and active transmission (active transmission), or reflect the incident signal by modulating the antenna impedance, which is also known as backscattering communication (backscattering communication).

[0003] Existing research primarily focuses on backscatter-assisted WPMEC from the perspectives of minimizing task processing latency, maximizing system energy efficiency (EE), and maximizing system computational bits. Some literature improves fairness by jointly optimizing the offloading parameters of IoT nodes, their local computational parameters, transmit power, and UAV trajectories to maximize the computational efficiency of the worst-performing IoT node. However, these studies only consider Time Division Multiple Access (TDMA) protocols, meaning only a single user is allowed to offload tasks within a time slot. Considering issues such as low spectrum resource utilization and short device battery life, other literature proposes a backscatter-assisted MEC network system based on Non-Orthogonal Multiple Access (NOMA) communication to maximize computational efficiency. Furthermore, UAV-assisted backscatter MEC systems have been considered to address the challenges of dense computing demands in hotspot areas. While ensuring the stability of hotspot energy queues, the long-term utility of all hotspots is maximized through joint optimization of data offloading decisions and contract design. However, since the desired signal of wireless devices is typically a finite codebook input rather than a Gaussian input in practice, the achievable rates described in the above works are not particularly accurate for backscatter-assisted WPMEC systems. Other literature has proposed an innovative user cooperation scheme integrating backscatter communication and active transmission in a cooperative WPMEC system composed of source nodes, auxiliary nodes, and hybrid access points to improve system efficiency. Considering the limited computing power of the MEC server, the Quality of Service (QoS) constraints of IoT nodes, and energy consumption constraints, two resource allocation schemes are proposed to maximize the total computing bits of all IoT nodes and the system's computational efficiency, respectively. However, existing research has not considered the information leakage problem caused by the broadcast characteristics of wireless channels; therefore, the system's security performance requires further discussion.

[0004] Due to the limited signal processing capabilities of IoT devices, Physical Layer Security (PLS) technology, which utilizes the characteristics of wireless channels to achieve secure transmission, is more suitable for MEC systems than traditional encryption techniques. Some literature has investigated the secure offloading of NOMA-assisted vehicular edge computing networks in the presence of multiple malicious eavesdropping vehicles to ensure secure wireless offloading from user vehicles to MEC servers. Under computational latency constraints, system energy consumption is minimized by jointly optimizing transmit power, computational resource allocation, and the selection of interfering vehicles in each NOMA cluster. Unmanned ground vehicles (UGVs) have been used to handle intensive computational tasks of UAVs in base station-free areas. Other literature, while ensuring communication security, maximizes the average utility of UAV-Ground Vehicle collaboration by jointly optimizing UAV trajectories, transmission power, and CPU frequency. However, security issues in backscatter-assisted WPMEC systems remain to be addressed. Therefore, the urgent task is to design a backscatter-assisted multi-user secure offloading system to ensure secure wireless offloading for multiple devices simultaneously. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a multi-user secure offloading method and apparatus based on backscatter-assisted edge computing, so as to meet the need to provide secure wireless offloading for multiple devices at the same time.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] This invention provides a multi-user secure offloading method based on backscatter-assisted edge computing, applied to a system consisting of a full-duplex power base station equipped with an edge computing server, multiple users, and an eavesdropper. The base station transmits a multi-carrier random continuous wave as a radio frequency source. Multiple users offload their tasks to the edge computing server through a hybrid backscatter communication and active transmission. The method includes: acquiring task offloading parameters for each user, the current user group, randomized continuous wave parameters, and channel parameters during the backscatter communication phase, the active transmission phase, and the eavesdropper's channel parameters; and then, based on the current user group, the eavesdropper's channel parameters, the channel parameters during the backscatter communication phase, the active transmission phase, and the randomized continuous wave parameters, the method proceeds securely. The continuous wave parameters are used to determine the confidentiality rate for each user during the backscatter communication phase and the active transmission phase. During the backscatter communication phase, multiple users within the same user group simultaneously offload their tasks, and the edge computing server decodes the target user's signal and treats other user signals as noise. Based on the task offloading parameters of each user, the confidentiality rate during the backscatter communication phase, and the confidentiality rate during the active transmission phase, the total task processing delay time, the total user energy consumption, and the energy collected by the user during the backscatter communication phase are determined. With the total user energy consumption and the energy collected by the user during the backscatter communication phase as constraints, the total task processing delay time is minimized, and the randomized continuous wave parameters, task offloading parameters, and user grouping are jointly optimized.

[0008] This invention proposes a multi-user secure offloading method based on backscatter-assisted edge computing. By constructing a system consisting of a full-duplex power base station, multiple users, and an eavesdropper, and utilizing the base station to transmit multi-carrier random continuous waves as a radio frequency source, users can offload tasks through hybrid backscatter communication and active transmission. This allows the acquisition of multiple parameters, which determine the confidentiality rate at each stage, the total task processing delay time, the total user energy consumption, and the acquisition energy. Then, with energy consumption and acquisition energy as constraints, multiple parameters are jointly optimized. This effectively solves the information leakage problem caused by the failure of existing technologies to consider the broadcast characteristics of wireless channels. It enables multiple users to securely offload tasks simultaneously, improves spectrum resource utilization, and reduces device energy consumption. By jointly optimizing randomized continuous wave parameters, task offloading parameters, and user grouping, the total system task processing delay time is significantly reduced, enhancing the system's security performance and task processing efficiency in multi-user scenarios. This provides a secure and efficient task offloading solution for the intensive computing needs of IoT devices.

[0009] Other advantages, objectives, and features of the invention will be set forth in the following description and will be apparent to those skilled in the art in some respects, or may be learned by practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0010] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following figures are provided for illustration:

[0011] Figure 1 This invention relates to a backscatter-assisted multi-user secure offloading system for wirelessly powered mobile edge computing.

[0012] Figure 2 This is a flowchart illustrating a specific example of a multi-user secure offloading method based on backscatter-assisted edge computing according to the present invention.

[0013] Figure 3 This is a time structure diagram of the backscatter-assisted multi-user secure offloading system in wirelessly powered mobile edge computing in this invention;

[0014] Figure 4 This is a schematic diagram of the joint optimization process for the three sub-objectives in this invention;

[0015] Figure 5 This is a diagram showing the relationship between the number of devices and the total system task processing latency in this invention;

[0016] Figure 6 This is a diagram showing the relationship between the amount of task data and the total system task processing latency in this invention.

[0017] Figure 7 This is a graph showing the relationship between the transmission power of the power beacon and the total system task processing delay in this invention. Detailed Implementation

[0018] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can also refer to the internal connection of two components; and they can refer to a wireless connection or a wired connection. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0020] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0021] This embodiment provides a multi-user secure offloading method based on backscatter-assisted edge computing, applicable to, for example... Figure 1 The system shown consists of a full-duplex power base station equipped with an edge computing server, M users, and an eavesdropper. Users are wireless devices with single antennas, and the set of users can be represented as M. The base station transmits multi-carrier random continuous waves as its radio frequency source. Multiple users offload tasks to the edge computing server through a hybrid backscatter communication and active transmission method, as follows: Figure 2 As shown, it includes:

[0022] S101, obtain the task offloading parameters, current user group, randomized continuous wave parameters, and channel parameters of the backscatter communication stage, the channel parameters of the active transmission stage, and the eavesdropper channel parameters for each user.

[0023] S102, based on the current user group, eavesdropper channel parameters, channel parameters in the backscatter communication stage, channel parameters in the active transmission stage, and randomized continuous wave parameters, determine the confidentiality rate of each user in the backscatter communication stage and the active transmission stage. In the backscatter communication stage, multiple users in the same user group simultaneously offload their tasks, and the edge computing server decodes the target user signal and treats other user signals as noise.

[0024] S103, based on the task offloading parameters of each user, the confidentiality rate of the backscatter communication phase, and the confidentiality rate of the active transmission phase, determine the total task processing delay time, the total energy consumption of the user, and the energy collected by the user in the backscatter communication phase.

[0025] S104 minimizes the total task processing delay time by taking the total user energy consumption and the energy collected by the user during the backscatter communication phase as constraints, and jointly optimizes the randomized continuous wave parameters, task offloading parameters, and user grouping.

[0026] For example, in order to illustrate the above process in detail, the following is the construction of various models of the above system:

[0027] Multicarrier randomized continuous wave signals can be mathematically formulated as follows:

[0028]

[0029] Where, x u,k,i ω represents the weight of the i-th subcarrier for the u-th user group in the k-th time slot. i =2πf i and f i Let represent the angular frequency and center frequency of the i-th subcarrier, respectively. Assume... The frequency spacing between adjacent carriers is the same, denoted by f. Δ This indicates that the carrier frequency f of the transmitted signal is... C It can be represented as:

[0030]

[0031] Among them, f1 and f I These represent the carrier frequency of the first subcarrier and the carrier frequency of the I-th subcarrier, respectively, where I represents the total number of subcarriers.

[0032] In existing research, multi-user backscatter communication mainly uses a constant value x. u,k,i Continuous waves, however, can be used to enhance security by randomizing continuous waves. Assume the source signal x... u,k It is random and follows a Gaussian distribution, that is: Let Φ represent the mean vector of the u-th user signal group. u Let E||x represent the variance vector of the u-th user signal group. u,k || 2 = P, where P represents the base station's transmit power. The channel response vector from the base station to user m can be given by the following formula:

[0033] h PW,m =[h PW,m,1 ,h PW,m,2 ,...,h PW,m,I (3)

[0034] Among them, h PW,m,i The complex channel response from the base station to user m at the i-th carrier frequency can be modeled as follows:

[0035]

[0036] Where, d PW,mLet m represent the distance from the base station to the user, γ represent the attenuation coefficient, v represent the propagation speed, and g0 represent the channel power gain when the reference distance is 1 meter, where g0 = (λ / 4π). 2 λ represents the wavelength of the carrier frequency, λ = v / f C In the k-th time slot, the radio frequency signal received by user m in the u-th group is represented as:

[0037] y m,u,k =X u,k h PW,m (5)

[0038] Where X u,k =Λ(x u,k In practice, modulation based on backscattering uses a finite alphabet, namely D.

[0039] Due to the full-duplex nature of the base station, it is assumed that the channel reciprocity between the BP and the user holds. The backscattered communication signal received by the base station from the u-th group in the k-th time slot can be given by the following formula:

[0040]

[0041] Among them, U u Let c represent the set of users contained in the u-th group. m,u,k ∈D represents the k-th data symbol from user m in group u. PB,u,k This represents the additive noise of the backscatter reader. I represents the thermal noise power of the reader at the base station. I Let represent an I×I identity matrix.

[0042] The channel response matrix between the base station and users from group u can be simplified to H u Then formula (6) can be transformed into:

[0043] y PB,u,k =X u,k H u c u,k +n PB,u,k (7)

[0044] Among them, c u,k This represents the symbol sequence transmitted by the user in group u in the k-th time slot.

[0045] The channel response vector between the base station and the eavesdropper is:

[0046] H PE =[h PE,1 ,h PE,2 ,...,h PE,I ] T (8)

[0047] Among them, h PE,i The complex channel response at the i-th carrier frequency between the base station and the eavesdropper can be expressed as:

[0048]

[0049] Where, d PE This represents the distance from the base station to the eavesdropper. The channel vector from user m to the eavesdropper can be represented as:

[0050] g WE,m =[g WE,m,1 ,g WE,m,2 ,...,g WE,m,I ] T (10)

[0051] Because communication between ground nodes is easily affected by obstacles and scatterers, the channel response from user m to the eavesdropper is simulated as a Rayleigh fading channel. Therefore, g WE,m,i It can be modeled as:

[0052]

[0053] Where, d WE,m ξ represents the distance from user m to the eavesdropper. WE,m,i ~CN(0,1). Therefore, the signal received by the eavesdropper during the backscatter communication phase is:

[0054]

[0055] Where, n EB,u,k This represents zero-mean additive Gaussian noise. The channel response matrix between the user from group u and the eavesdropper can be simplified to G. u Then formula (12) can be transformed into:

[0056] y EB,u,k =Λ(G u c u,k +H PE )x u,k +n EB,u,k (13)

[0057] Based on the above modeling and derivation, a task offloading model and the confidentiality rate for each user can be constructed, specifically:

[0058] Users offload tasks at an achievable level of security to ensure the base station can successfully decode information while preventing eavesdroppers from extracting information from the received signals. When multiple users in the same group offload tasks simultaneously, interference can occur between them. In the system of the invention, the edge computing server decodes the target user's signal while treating the signals of other users as noise. Therefore, the security level for an individual user within the group will be derived below. In this embodiment, for performance evaluation, the security level during the backscatter communication phase is expressed in bits used per channel (BPCU), and the security level during the active transmission phase is expressed in bits per second. For ease of description, the subscript u representing the packet order and the subscript k representing the time slot are omitted.

[0059] The process of determining the achievable rate of each user's legitimate link and the achievable eavesdropping link rate based on the current user group, eavesdropper channel parameters, channel parameters during the backscatter communication phase, and randomized continuous wave parameters is as follows:

[0060] The achievable rate of the legitimate link for the nth user in each group can be derived from the following formula:

[0061]

[0062] Where D represents a finite alphabet, A represents the number of symbol combinations that a single user can transmit, A = |D|, and L represents the number of symbol combinations that each group of users can transmit, L = |D|. I |, L′ represents the number of possible symbol combinations that other users in each group, excluding user n, can transmit, L′=|D I-1 |, This represents the average number of symbols transmitted to all users. This represents the average number of symbols transmitted to all users except user n. This can be considered as satisfying variables, This can be considered as satisfying variables, X = Λ(x),

[0063] The Gaussian randomness of multicarrier randomized continuous wave signals makes them artificial noise that interferes with eavesdropping links. The achievable eavesdropping link rate for user n can be expressed as:

[0064]

[0065] in, Represents the mean vector of the user signal. This represents the average number of symbols transmitted to all users. This can be considered as satisfying variables, in It can be deduced in This represents the average number of symbols transmitted to all users except user n. This can be considered as satisfying variables, in

[0066] After expressing the achievable rates of the legitimate link and the eavesdropping link as formulas (14) and (15) respectively, the confidentiality rate of user n can be expressed as:

[0067] R SB,n =max(R) PB,n -R EB,n ,0)(16)

[0068] However, due to R PB,n and R EB,n The complexity of integrals makes it difficult to obtain a closed-form expression for the confidentiality rate. Jansen's inequality is used to apply this to R... PB,n and R EB,n An approximate calculation is performed. The approximate value of a legitimate link can be expressed as:

[0069]

[0070] in,

[0071] Φ represents the variance vector of the user signal.

[0072] Similarly, the approximate value of the eavesdropping link can be expressed as:

[0073]

[0074] in,

[0075] Therefore, the confidentiality rate R SB It is easy to approximate as:

[0076]

[0077] The process of determining the user's active transmission rate and the eavesdropper's data rate when the user is offloading the channel, based on the channel parameters during the active transmission phase, the eavesdropper's channel parameters, and the randomized continuous wave parameters, is as follows:

[0078] This embodiment assumes that traditional radio frequency wireless communication technology is used during the active transmission phase. The active transmission rate of user m can be obtained by the following formula:

[0079]

[0080] Where B represents the channel bandwidth, p m Let N0 represent the transmit power of user m, N0 represent the power spectral density of additive white Gaussian noise (AWGN), and h represent the transmit power of user m. PW,m This represents the channel response vector from the base station to user m.

[0081] The base station enhances security during the active transmission phase by generating artificial noise to interfere with eavesdroppers. Therefore, the eavesdropper's data rate when user m offloads the channel can be expressed as:

[0082]

[0083] Among them, P J h represents the power of the artificial noise generated by the base station. WE,m h represents the channel response vector from user m to the eavesdropper. PE This represents the channel response vector between the base station and the eavesdropper.

[0084] Therefore, the achievable confidentiality rate for user m during the active transmission phase can be expressed as:

[0085] R SA,m =max(R) WA,m -R EA,m ,0) (22)

[0086] Next, the proposed system considers a partial offloading scheme, where computational tasks can be intentionally divided into different subtasks. Each subtask can be offloaded to an edge computing server or computed locally. Computationally intensive tasks generated by user m are denoted as Tasks. m =<α m ,β m D m C m |m∈M>, where α m β represents the offloading rate of user m during the backscatter communication phase. m D represents the offloading rate of user m during the active transmission phase. m C represents the size of the task data for user m. m In this embodiment, α represents computing resources. m β m C m D m As a parameter for task unloading, the data size unloaded by user m can be expressed as (α) m +β m )D m The total number of CPU cycles required by user m can be determined by C. m D m Given. In this embodiment, α m β m Dm C m As a parameter for task unloading.

[0087] The proposed system's time structure, such as Figure 3 As shown, each user performs local computation on their tasks within time block T. Based on the task offloading parameters for each user, the local processing latency for user m is determined as follows:

[0088]

[0089] Among them, f m This represents the CPU operating frequency of user m.

[0090] During the backscatter communication phase, users in group u can simultaneously offload tasks. Therefore, the task offloading delay of group u can be uniformly set to the maximum offloading delay of that group. Based on the task offloading parameters of each user and the security level of the backscatter communication phase, the maximum offloading delay time for each user group during the backscatter phase can be determined as follows:

[0091]

[0092] During the active transmission phase, each user offloads their tasks to the edge computing server. Based on each user's task offloading parameters and the confidentiality rate during the active transmission phase, the task offloading delay time for each user during this phase is determined. The task offloading delay for user m during this phase can then be expressed as:

[0093]

[0094] Edge computing servers will allocate their computing resources f E And calculation duration T E To process the received tasks.

[0095] To ensure service quality requirements, the edge computing server should at least complete all tasks that the user has unloaded, namely:

[0096]

[0097] Therefore, the system's task processing latency is the maximum value between each user's local computing latency and the total task processing latency of all users, and its calculation formula is:

[0098]

[0099] Among them, T PB Indicates the duration of the energy signal broadcast by the base station.

[0100] Then, the energy harvesting and consumption model of this system is analyzed, as follows:

[0101] This embodiment uses a nonlinear energy harvesting model to describe the energy harvesting circuit for each user. The energy harvested by user m during the backscatter communication phase can be expressed as:

[0102]

[0103] Among them, a m b m and e m Let P represent the nonlinear energy harvesting model parameters for user m, P represent the base station's transmit power, and I represent the total number of subcarriers. Each user's energy consumption consists of three parts: local computation energy consumption, backscatter communication stage energy consumption, and active transmission stage energy consumption. Specifically, user m's local computation energy consumption is determined based on the user's local processing latency, the user's computing frequency, and the energy efficiency coefficient, as follows:

[0104]

[0105] Where, ε m This represents the energy efficiency coefficient of user m.

[0106] The energy consumption of user m during the backscatter communication phase can be determined based on the maximum offloading delay time of each user group during the backscatter phase and the constant circuit power consumption during the backscatter communication phase, as given by the following formula:

[0107] E B,m =p B,m T B,m (30)

[0108] Where, p B,m This represents the constant circuit power consumption during the backscatter communication phase.

[0109] The energy consumption of user m during the active transmission phase can be determined based on the task offloading delay time of each user during the active transmission phase and the active transmission phase.

[0110] The constant circuit power consumption during the transmission phase is determined and expressed as:

[0111] E A,m =(p m +p A,m )T A,m (31)

[0112] Where, p A,m p represents the constant circuit power consumption during the active transmission phase. m This represents the transmit power of user m.

[0113] Finally, based on the above model construction and derivation, the optimization objective is described as follows: By jointly optimizing user grouping, the power of randomized continuous wave signal sources, the power of randomized continuous wave interference signals, the offloading rate of backscatter communication, the offloading rate of active transmission, the CPU computing frequency of users, the CPU computing frequency of edge computing servers, the duration of base station broadcast energy signals, the computing duration of edge computing servers, and the power of artificial noise generated by base stations, the task processing latency of the system is minimized. Therefore, the optimization problem can be formulated as:

[0114]

[0115] stC1:E L，m +E B,m +E A,m ≤E EH,m (32b)

[0116] C2:0≤f m ≤f max (32c)

[0117] C3:0≤f E ≤f Emax (32d)

[0118] C4: 0 ≤ P J ≤P Jmax (32e)

[0119] C5:

[0120] C6:

[0121] C7:α m ,β m α m +β m ∈[0,1] (32h)

[0122] C8:

[0123] C9:

[0124] C 10 :

[0125] C 11 :

[0126] in α=[α1,...,α M ], β=[β1,...,β M ], f = [f1,...,fM ], P max and P Jmax f represents the maximum transmit power and maximum interference power of PB, respectively. max and f Emax These represent the maximum computing frequency of the WD and edge servers, respectively.

[0127] In this problem, C1 represents the energy constraint, C2 and C3 constrain the maximum CPU operating frequency of the user and the edge computing server, respectively, C4 constrains the maximum power of the artificial noise generated by the base station, C5 represents the QoS constraint, C6 constrains the duration of the base station broadcasting the energy signal, C7 constrains the task offloading rate for each user, and C8 and C9 constrain the mean and variance of the randomized continuous wave signal. 10 and C 11 This indicates user grouping constraints.

[0128] To minimize the task processing latency of the proposed backscatter-assisted multi-user safe offloading system, joint optimization was performed on the randomized continuous wave settings, offloading parameter settings, and user grouping. Considering that the optimization problem is a non-convex mixed-integer programming problem with multiple coupled optimization variables in the objective function and constraints, the original optimization objective was decomposed into three sub-objectives using the block coordinate descent method: the first sub-objective for optimizing the randomized continuous wave parameters, the second sub-objective for optimizing the task offloading parameters, and the third sub-objective for optimizing user grouping.

[0129] To achieve joint optimization of the three sub-objectives, this embodiment treats the randomized continuous wave setting, the unloading parameter setting, and user grouping as three variable blocks. In the proposed algorithm, user grouping is performed first, and then the randomized continuous wave setting and the unloading parameter setting are alternately optimized in each iteration, such as... Figure 4 As shown, it specifically includes:

[0130] S1, based on the previous user grouping and task offloading parameters, aims to maximize the confidentiality rate of the backscatter communication phase. Through the SCA algorithm, randomized continuous wave parameters are obtained by repeated iterations.

[0131] S2, based on the randomized continuous wave parameters obtained from the first sub-target, with the goal of minimizing the total task processing delay time and the constraint that the total user energy consumption is less than or equal to the backscatter acquisition energy, the task unloading parameters are solved through convex optimization.

[0132] S3, determine the correlation between users based on the channel response parameters between users, and update the user grouping for this round based on the user correlation;

[0133] S4. Determine if the number of iterations meets the requirements. If it does not meet the requirements, divide the users in this round in step S3 into the user group of the previous round in step S1, and repeat steps S1-S4 until the requirements are met.

[0134] For the first sub-target, the process of optimizing the randomized continuous wave parameters is as follows:

[0135] The randomized continuous wave setup problem can be viewed as χ and The power distribution problem among elements. Due to χ and The changes only affect The value of can be expressed as subproblem P1a:

[0136]

[0137] stC8 and C9(33b)

[0138] Since the objective function of the optimization problem is related to χ and Since they are all non-convex, it is difficult to solve P1a directly. As an alternative, this embodiment will propose a solution based on the Continuous Convex Approximation (SCA). First, The gradient can be expressed as:

[0139]

[0140] Among them, V i,j W i,j V′ i,j,l and W′ i,j,l The definition is as follows:

[0141]

[0142] in, The gradient relative to χ can be set and The result is as follows:

[0143]

[0144] Substituting formula (36) into formula (34) yields the gradient. Therefore, in the t-th SCA iteration, The surrogate function for χ can be expressed as:

[0145]

[0146] Where χ (t) Let χ represent the optimized value after the t-th SCA iteration, where ν is a positive constant. Therefore, The proxy problem concerning χ can be expressed as:

[0147]

[0148] stC8 (38b)

[0149] C 12 :χ±0 (38c)

[0150] P1b is a strictly concave function associated with convex constraints, and the solution χ′ can be efficiently obtained using convex tools such as CVX. Similarly, in the t-th SCA iteration, about The proxy function can be represented as:

[0151]

[0152] therefore, about The proxy problem can be described as follows:

[0153]

[0154] stC8 (40b)

[0155] C 13 :

[0156] P1c is a strictly concave function associated with convex constraints, and the solution... It can be solved efficiently using convex tools such as CVX.

[0157] For the second sub-objective, the process of optimizing the task unloading parameters is as follows:

[0158] Given specific user groups and specific settings for randomized continuous waves, the system's task processing latency is further reduced by addressing the issue of offloading parameter settings, where subproblem P2a can be represented as:

[0159]

[0160] stC1-C7(41b)

[0161] Since the objective function of P2a contains a minimax function, an auxiliary variable μ = T is introduced. total Rewrite P2a as follows:

[0162]

[0163] C2-C7 (42c)

[0164] C 14 :TL，m” μ (42d)

[0165] C 15 :T E，m” μ (42e)

[0166] Among them, C 14 and C 15 It is obtained by introducing an auxiliary variable, C′1, by replacing T with μ in C1. L,m This was obtained. However, P2b remains complex. Through simple derivation, it can be seen that the minimum task processing latency of the system is achieved under the premise that the edge computing server executes the task at its maximum CPU computing frequency and the base station transmits artificial noise at its maximum power, i.e. Therefore, P2b can be rewritten as:

[0167]

[0168] stC2, 6, and C7 43b

[0169] C″1:

[0170] C′5:

[0171] C′ 14 (1-α m -β m C m D m” f m μ (43e)

[0172]

[0173] Among them, E′ A,m It is by passing P J =P Jmax Substituting into formula (21) yields the result. However, due to C″1 and C′... 14 Chinese f m Despite the coupling between μ and f, P2c remains non-convex because each user has a unique f. m However, since they share the same μ, they cannot be solved by introducing auxiliary variables. Nevertheless, with μ fixed, the optimization problem P2c can be simplified to P2d, expressed as:

[0174] P2d:A={α,β,f,T PB ,T E}(44a)

[0175] stC″1,C2,C′5,C6,C7,C′ 14 ,and C′15 (44b)

[0176] P2d is a concave function that can be efficiently solved using convex tools such as CVX.

[0177] For the third sub-objective, the process of optimizing user grouping is as follows:

[0178] Given specific settings for the randomized continuous wave and unloading parameters, the task processing latency of the system is further reduced by solving the user grouping problem, where subproblem P3a can be expressed as:

[0179]

[0180] stC 10 and C 11 (45b)

[0181] In practice, P3a can be solved using an exhaustive search and by replacing all grouping schemes. Unfortunately, this method is highly complex. This embodiment proposes a correlation-based user grouping solution that minimizes overall system latency by optimizing user grouping. In this method, user grouping is based on a correlation coefficient, and the correlation coefficient between user m and user n can be calculated as follows:

[0182]

[0183] Where θ m,n This represents the correlation between user m and user n. h PW,m The transpose of the matrix, h PW,m The conjugate of h PW,m Let m represent the channel response vector from the base station to user m. h PW,n The transpose of the matrix, h PW,n The conjugate of h PW,n Let represent the channel response vector from the base station to user n. The basic idea of ​​the proposed method is to improve the confidentiality of each user group during the backscatter communication phase by dividing highly correlated users into different groups.

[0184] This invention proposes a multi-user secure offloading method based on backscatter-assisted edge computing. By constructing a system consisting of a full-duplex power base station, multiple users, and an eavesdropper, and utilizing the base station to transmit multi-carrier random continuous waves as a radio frequency source, users can offload tasks through hybrid backscatter communication and active transmission. This allows the acquisition of multiple parameters, which determine the confidentiality rate at each stage, the total task processing delay time, the total user energy consumption, and the acquisition energy. Then, with energy consumption and acquisition energy as constraints, multiple parameters are jointly optimized. This effectively solves the information leakage problem caused by the failure of existing technologies to consider the broadcast characteristics of wireless channels. It enables multiple users to securely offload tasks simultaneously, improves spectrum resource utilization, and reduces device energy consumption. By jointly optimizing randomized continuous wave parameters, task offloading parameters, and user grouping, the total system task processing delay time is significantly reduced, enhancing the system's security performance and task processing efficiency in multi-user scenarios. This provides a secure and efficient task offloading solution for the intensive computing needs of IoT devices.

[0185] The following is a simulation example for this embodiment:

[0186] This invention optimizes the total task processing latency of a backscatter-assisted mobile edge computing system, effectively eliminating the risk of eavesdropping and improving system security, while also enabling multi-user task offloading. The simulation scenario is set within a 150m × 150m square area, containing 20 users, 1 eavesdropper, and 1 base station. It is assumed that users have BPSK codebook input, and the task data size generated by each user is set to 100 kbits. The number of CPU cycles required to compute each bit of data is set to 1000 revolutions per bit.

[0187] In the proposed algorithm, RCW settings, offload parameter settings, and user grouping have all been optimized. To illustrate the advantages of the multi-user secure offload scheme proposed in this invention, we considered the following benchmark schemes for performance comparison, including a no-eavesdropping scheme, a no-local-computation scheme, a no-backscatter communication offload scheme, and a no-active-transmission offload scheme.

[0188] The relationship between total system task processing latency and the number of users is as follows: Figure 5As shown in the figure, the total system latency of all task offloading schemes increases with the number of users. For the backscatter communication-free offloading scheme, an increase in the number of users means an increase in the total latency of all users offloading tasks during the active transmission phase. For the scheme proposed in this invention, the eavesdropper-free scheme, and the local computing-free scheme, an increase in the number of users not only affects the number of user groups but also increases the overall latency of all users offloading tasks during the active transmission phase. Since the energy required for backscatter communication is lower than that required for active transmission, the active transmission-free offloading scheme can efficiently complete energy harvesting and task offloading when the number of users is small. However, the task offloading speed of backscatter communication is lower than that of active transmission, resulting in a longer latency for energy harvesting and task offloading as the number of users increases. As can be seen from the figure, the overall system latency of the proposed scheme is very close to that of the eavesdropper-free scheme, which verifies that the proposed scheme effectively improves the security of users during task offloading. At the same time, the overall system latency of the scheme proposed in this invention is consistently lower than that of other schemes, demonstrating its feasibility and superiority.

[0189] Figure 6 The relationship between task data volume and total system processing latency was compared. As the task data volume increased from 60 kbits to 140 kbits, the relationship between task data volume and overall system latency for each scheme was approximately linear. Among all schemes, the scheme without active transmission offloading performed the worst, followed by the scheme without backscatter communication offloading. The scheme proposed in this invention and the scheme without local computation both achieved good results. As shown in the figure, the overall system latency of the proposed scheme is very close to the latency of the scheme without eavesdroppers, reflecting the superiority of the proposed physical layer security method.

[0190] Figure 7 The relationship between base station transmit power and total system task processing latency is illustrated in the figure. As can be seen from the figure, the overall system latency decreases with increasing transmit power for each offloading scheme. For each scheme, increasing transmit power not only affects the confidentiality rate of the backscatter communication phase but also increases the energy required for local computation and data transmission. Since backscatter communication requires less energy than active transmission, the offloading scheme without active transmission has lower task processing latency at low transmit power. Because the task offloading rate of active transmission is higher than that of backscatter communication, the offloading scheme without backscatter communication performs better at higher transmit power. RCW cannot sufficiently interfere with eavesdroppers at low transmit power, resulting in a slight performance gap between the proposed scheme and the eavesdropper-free scheme. However, the overall system latency of the proposed scheme is very close to that of the eavesdropper-free scheme and is consistently lower than other schemes even at higher base station transmit power, demonstrating the superiority of the proposed scheme.

[0191] Simulation results demonstrate the accuracy of the comparison between the proposed approximation and the precise value, and verify that the proposed scheme can significantly reduce the system's task processing latency while efficiently achieving multi-user task offloading.

[0192] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made to it in form and detail without departing from the scope defined by the claims of the present invention.

Claims

1. A multi-user secure offloading method based on backscatter-assisted edge computing, characterized in that, This system is applied to a full-duplex power base station equipped with an edge computing server, multiple users, and eavesdroppers. The base station transmits multi-carrier random continuous waves as a radio frequency source, and multiple users offload tasks to the edge computing server through hybrid backscatter communication and active transmission, including: Obtain the task unloading parameters, current user group, randomized continuous wave parameters, and channel parameters during the backscatter communication phase, the active transmission phase, and the eavesdropper channel parameters for each user. Based on the current user group, eavesdropper channel parameters, channel parameters in the backscatter communication phase, channel parameters in the active transmission phase, and randomized continuous wave parameters, the confidentiality rate of each user in the backscatter communication phase and the active transmission phase is determined. In the backscatter communication phase, multiple users in the same user group simultaneously offload their tasks, and the edge computing server decodes the target user signal and treats other user signals as noise. Based on the task offloading parameters of each user, the confidentiality rate of the backscatter communication phase, and the confidentiality rate of the active transmission phase, determine the total task processing delay time, the total energy consumption of the user, and the energy collected by the user during the backscatter communication phase. Constrained by the total user energy consumption and the energy collected by the user during the backscatter communication phase, the total task processing delay time is minimized by jointly optimizing the randomized continuous wave parameters, task offloading parameters, and user grouping. Based on the current user group, eavesdropper channel parameters, channel parameters during the backscatter communication phase, channel parameters during the active transmission phase, and randomized continuous wave parameters, determine the confidentiality rate for each user during the backscatter communication phase and the active transmission phase, including: Based on the current user group, the eavesdropper channel parameters, the channel parameters during the backscatter communication phase, and the randomized continuous wave parameters, determine the achievable rate of each user's legitimate link and the achievable eavesdropping link rate for each user. The confidentiality rate of a user during the backscatter communication phase is determined based on the achievable rate of each user's legitimate link and the achievable eavesdropping link rate of the user. Based on the channel parameters during the active transmission phase, the eavesdropper's channel parameters, and the randomized continuous wave parameters, determine the user's active transmission rate and the eavesdropper's data rate when the user is offloading the channel. The confidentiality rate of each user during the active transmission phase is determined based on the data rate during the user's active transmission phase and the data rate of the eavesdropper when the user offloads the channel. Constrained by the total user energy consumption and the energy collected by the user during the backscatter communication phase, the total task processing delay is minimized by jointly optimizing the randomized continuous wave parameters, task offloading parameters, and user grouping, including: Based on the block coordinate descent method, the total user energy consumption and the energy collected by the user during the backscatter communication phase are taken as constraints. The total task processing delay time is minimized, and the randomized continuous wave parameters, task offloading parameters and user grouping are jointly optimized. The results are decomposed into three sub-objectives: the first sub-objective is used to optimize the randomized continuous wave parameters, the second sub-objective is used to optimize the task offloading parameters, and the third sub-objective is used to optimize the user grouping. The three sub-objectives are optimized collaboratively, as follows: S1, based on the previous user grouping and task offloading parameters, aims to maximize the confidentiality rate of the backscatter communication phase. Through the SCA algorithm, randomized continuous wave parameters are obtained by repeated iterations. S2, based on the randomized continuous wave parameters obtained from the first sub-target, with the goal of minimizing the total task processing delay time and the constraint that the total user energy consumption is less than or equal to the backscatter acquisition energy, the task unloading parameters are solved through convex optimization. S3, determine the correlation between users based on the channel response parameters between users, and update the user grouping for this round based on the user correlation; S4. Determine if the number of iterations meets the requirements. If it does not meet the requirements, divide the users in this round in step S3 into the user group of the previous round in step S1, and repeat steps S1-S4 until the requirements are met.

2. The multi-user secure offloading method based on backscatter-assisted edge computing according to claim 1, characterized in that, Based on the task offloading parameters for each user, the confidentiality rate during the backscatter communication phase, and the confidentiality rate during the active transmission phase, the total task processing latency is determined, including: Determine the local processing delay time for each user based on their task unloading parameters; Based on the task offloading parameters of each user and the confidentiality rate of the backscatter communication phase, determine the maximum offloading delay time for each user group in the backscatter phase. Based on the task unloading parameters of each user and the confidentiality rate during the active transmission phase, determine the task unloading delay time for each user during the active transmission phase. Based on the task unloading parameters, constrain the time required for the edge server to complete all task unloading calculations; The duration of the base station broadcast energy signal is constrained based on the maximum offload delay time of each user group during the backscatter phase. The total task processing delay time is determined based on the user's local processing delay time, the duration of the base station broadcast energy signal, the task offloading delay time of each user during the active transmission phase, and the time it takes for the edge server to complete all task offloading calculations.

3. A multi-user secure offloading method based on backscatter-assisted edge computing according to claim 2, characterized in that, The user's total energy consumption consists of local computing energy consumption, backscatter phase energy consumption, and active transmission phase energy consumption. The methods for determining the user's total energy consumption include: Obtain the user's operating frequency, energy efficiency coefficient, constant circuit power consumption during the backscatter communication phase, and constant circuit power consumption during the active transmission phase. The user's local computing energy consumption is determined based on the user's local processing latency, the user's computing frequency, and the energy efficiency coefficient. The energy consumption of the backscatter phase is determined based on the maximum offload delay time of each user group in the backscatter phase and the constant circuit power consumption in the backscatter communication phase. The energy consumption of the active transmission phase is determined based on the task offloading delay time of each user during the active transmission phase and the constant circuit power consumption during the active transmission phase.

4. A multi-user secure offloading method based on backscatter-assisted edge computing according to claim 1, characterized in that, The confidentiality rate for users during the backscatter communication phase is: ； in, express Approximate value, , for Approximate value, Indicates the first The achievable rate for a user's legitimate link For users n The achievable eavesdropping link rate, for Approximate value; The user's confidentiality rate during the active transmission phase is : ； in, For users m Rate during the active transmission phase, To eavesdrop on users m The data rate of the eavesdropper when the channel is offloaded. , , Indicates channel bandwidth. Indicates user m The transmission power, This represents the power spectral density of additive white Gaussian noise. This represents the power of the artificial noise generated by the base station. Indicates the distance from the base station to the user The channel response vector, Indicates from user The channel response vector between the eavesdropper and the eavesdropper. This represents the channel response vector from the base station to the eavesdropper.

5. A multi-user secure offloading method based on backscatter-assisted edge computing according to claim 2, characterized in that, The total task processing delay time is: ； in, This indicates the local processing delay time for user m. ; Indicates user Offloading rate during the backscatter communication phase Indicates user Offload rate during the active transmission phase, This represents the CPU operating frequency of user m. , Representing users respectively The computing resources and task data size; Indicates the duration of the energy signal broadcast by the base station. This indicates the time it takes for the edge computing server to complete all tasks that the user has uninstalled. This represents the sum of the task unloading delay times for each user during the active transmission phase.

6. A multi-user secure offloading method based on backscatter-assisted edge computing according to claim 3, characterized in that, The user's local calculated energy consumption is: ； in, This represents the energy efficiency coefficient of user m. This represents the CPU operating frequency of user m. Local processing delay time; The energy consumption of the backscattering stage is: ； in, This represents the constant circuit power consumption of user m during the backscatter communication phase. This represents the maximum offload delay time for user m during the backscatter communication phase; The energy consumption during the active transmission phase is: ； in, This represents the constant circuit power consumption of user m during the active transmission phase. express, This represents the task unloading delay time for user m during the active transmission phase.

7. A multi-user secure offloading method based on backscatter-assisted edge computing according to claim 1, characterized in that, The energy collected by the user during the backscatter communication phase is determined by the following formula: ； Where I represents the total number of subcarriers, and P represents... , and Indicates user Parameters of the nonlinear energy harvesting model, Indicates the distance from the base station to the user The channel response vector, Indicates the duration of the energy signal broadcast by the base station. For the first Maximum uninstallation delay for group tasks.

8. A multi-user secure offloading method based on backscatter-assisted edge computing according to claim 1, characterized in that, Based on the channel response parameters between users, determine the correlation between users, including: ； in, Indicates user m and users n The correlation between them express The transpose of the matrix, express conjugate, Indicates the distance from the base station to the user The channel response vector, express The transpose of the matrix, express conjugate, Indicates the distance from the base station to the user n The channel response vector.

Images