Edge computing task unloading optimization method and device immune to data noise

Abstract

The invention provides an edge computing task unloading optimization method and device immune to data noise. The method comprises the following steps: acquiring environment monitoring data at a previous moment; inputting the environmental monitoring data at the previous moment into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain environmental monitoring data after noise filtering; and on the basis of the noise-filtered environment monitoring data, a heuristic greedy calculation unloading algorithm is utilized to calculate to-be-unloaded tasks of all mobile devices at the current moment, and an optimal task unloading scheme is obtained. According to the particle Kalman cascade filter provided by the invention, the data volatility is effectively reduced, and the original data distribution is recovered. And the task completion delay can be obviously reduced.

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology, and in particular to an edge computing task offloading optimization method and device that are immune to data noise. Background Art

[0002] The Industrial Internet of Things (IIoT) plays a vital role in industrial data collection, mobile device management, and dynamic resource allocation. The existing IIoT has introduced the Passive Optical Network (PON) architecture, which enables long-distance, high-speed data transmission. This is crucial for improving the efficiency of industrial automation and intelligent manufacturing systems, especially in data-intensive application scenarios.

[0003] Edge computing is commonly used in the Industrial Internet of Things (IIoT). Edge computing refers to providing localized services for industrial production data informatization, leveraging nearby computing, network, and storage resources close to the actual work environment or data source. This ensures real-time performance and data reliability requirements in industrial production. In IIoT edge computing systems, a large number of mobile devices offload computing tasks to MEC (Mobile Edge Computing) servers or the cloud. However, facing challenges such as processing latency, determining where to offload computing tasks to the cloud or MEC servers is a key technical issue currently requiring attention. Existing technologies typically rely on collecting environmental monitoring data to develop a series of task offloading strategies optimized for low latency. However, in the IIoT, where numerous devices are interconnected, especially in industrial PON environments, environmental monitoring data collected can be subject to noise from multiple sources, reducing the effectiveness of these offloading strategies. Summary of the Invention

[0004] The present invention provides an edge computing task offloading optimization method and device that are immune to data noise, which is particularly suitable for scenarios such as intelligent manufacturing and remote equipment monitoring in the industrial Internet of Things that are sensitive to delay and have severe noise interference. It is used to solve the defects of the existing technology that the environmental monitoring data collected in the industrial PON environment is subject to noise interference, and improve the effectiveness of the computing task offloading strategy.

[0005] The present invention provides an edge computing task offloading optimization method that is immune to data noise, comprising the following steps.

[0006] Obtain environmental monitoring data from the previous moment; the environmental monitoring data from the previous moment includes environmental monitoring data from all mobile edge computing nodes, all cloud nodes, and all mobile devices in the industrial Internet of Things; Inputting the environmental monitoring data at the previous moment into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain noise-filtered environmental monitoring data; Based on the noise-filtered environmental monitoring data, a heuristic greedy calculation offloading algorithm is used to calculate the tasks to be offloaded of all mobile devices at the current moment to obtain an optimal task offloading solution.

[0007] According to an edge computing task offloading optimization method that is immune to data noise provided by the present invention, before obtaining the environmental monitoring data at the previous moment, the method includes: A comprehensive noise model is established for the noise in the environmental monitoring data at the previous moment; the comprehensive noise model generates mixed noise by weighted superposition of Gaussian noise, impulse noise, etc., and the state prediction of the particle filter is adjusted based on the model (such as Formula 1).

[0008] According to an edge computing task offloading optimization method that is immune to data noise provided by the present invention, the environmental monitoring data at the previous moment is input into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain the noise-filtered environmental monitoring data, including: Taking each data in the environmental monitoring data at the previous moment as a particle state, randomly sampling from the environmental monitoring data at the previous moment to obtain particle states of a plurality of initial particles, and assigning an initial weight to each initial particle; Based on the comprehensive noise model, the states of the multiple initial particles are predicted to obtain the particle states at the current moment; Calculating the weight of the initial particle at the current moment according to the particle state at the current moment and the initial weight of the initial particle; Resampling from the multiple initial particles according to the weight of the initial particle at the current moment to obtain a particle set; The particle set is used as the input of the Kalman filter to obtain the final state estimation of the particles output by the Kalman filter at the current moment.

[0009] According to an edge computing task offloading optimization method that is immune to data noise provided by the present invention, the particle set is used as the input of the Kalman filter to obtain the final state estimate of the particles output by the Kalman filter at the current moment, including: Calculate the Kalman gain; Update the state of the particle at the current moment according to the observation noise covariance matrix; According to the Kalman gain, the final state estimate of the particle at the current moment is updated.

[0010] According to the present invention, a method for optimizing edge computing task offloading that is immune to data noise is provided. Based on the noise-filtered environmental monitoring data, a heuristic greedy calculation offloading algorithm is used to calculate the tasks to be offloaded of all mobile devices at the current moment to obtain an optimal task offloading solution, including: Get the tasks to be uninstalled for each mobile device at the current moment; Calculate the average delay of all tasks to be offloaded at the current moment; The noise-filtered environmental monitoring data is iteratively calculated using a heuristic greedy calculation offloading algorithm to minimize the average delay and obtain an optimal task offloading solution for the mobile device at the current moment.

[0011] According to an edge computing task offloading optimization method that is immune to data noise provided by the present invention, the environmental monitoring data includes any one of the load queue length, transmission rate, computing power, and mobile user location of the mobile edge computing server (i.e., MEC server).

[0012] The present invention also provides an edge computing task offloading optimization device that is immune to data noise, comprising the following modules: A data acquisition module is used to obtain environmental monitoring data at the previous moment; the environmental monitoring data at the previous moment includes environmental monitoring data of all mobile edge computing nodes, all cloud nodes, and all mobile devices in the industrial Internet of Things; a data denoising module, configured to input the environmental monitoring data at the previous moment into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain noise-filtered environmental monitoring data; The optimal task offloading solution output module is used to calculate the tasks to be offloaded of all mobile devices at the current moment based on the noise-filtered environmental monitoring data using a heuristic greedy calculation offloading algorithm to obtain the optimal task offloading solution.

[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, it implements any of the above-described edge computing task offloading optimization methods that are immune to data noise.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described edge computing task offloading optimization methods that are immune to data noise.

[0015] The present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-described edge computing task offloading optimization methods that are immune to data noise.

[0016] The present invention provides a data-noise-immune edge computing task offloading optimization method and device. This method obtains environmental monitoring data from the previous moment, including environmental monitoring data from all mobile edge computing nodes, all cloud nodes, and all mobile devices in the Industrial Internet of Things (IIoT). This environmental monitoring data is then input into a particle Kalman cascade filter, which then filters the noise from the previous moment to obtain noise-free environmental monitoring data. Based on the noise-free environmental monitoring data, a heuristic greedy computation offloading algorithm is then used to calculate the tasks to be offloaded for all mobile devices at the current moment, resulting in an optimal task offloading solution. This application demonstrates excellent performance in mitigating the impact of data noise on task offloading in industrial PON networks. The proposed particle Kalman cascade filter (PKCF) effectively reduces data volatility and restores the original data distribution. It captures complex noise distributions through non-parametric sampling using a particle filter and corrects state errors through linear optimal estimation using a Kalman filter. Compared to a single filter, this method reduces mean squared error by 50% in impulse noise scenarios. In addition, these findings emphasize the importance of data filtering (denoising) technology in ensuring the reliability of the task offloading process in industrial PON, providing an application basis for subsequent theoretical research. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 This is one of the flow charts of the edge computing task offloading optimization method that is immune to data noise provided by the present invention.

[0019] Figure 2 This is a schematic diagram of the network structure of the industrial Internet of Things provided by the present invention.

[0020] Figure 3 It is a schematic diagram of the data distribution state of the environmental monitoring information provided by the present invention before and after denoising.

[0021] Figure 4 It is a schematic diagram of the execution flow of the particle Kalman cascade filter (PKCF) provided by the present invention.

[0022] Figure 5 This is the second flow chart of the edge computing task offloading optimization method that is immune to data noise provided by the present invention.

[0023] Figure 6 This is a structural diagram of the edge computing task offloading optimization device that is immune to data noise provided by the present invention.

[0024] Figure 7 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0025] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0026] The following combination Figure 1-Figure 7 Specific embodiments of the present invention are described.

[0027] Figure 1 This is one of the flow charts of the edge computing task offloading optimization method that is immune to data noise provided by the present invention, such as Figure 1 As shown, the method includes the following: Step 101: Obtain the pre-processed environmental monitoring data of the previous moment (i.e. , representing the i-th environmental monitoring data at the previous moment (i.e., moment t-1); the preprocessed environmental monitoring data at the previous moment includes the environmental monitoring data of all mobile edge computing nodes, all cloud nodes, and all mobile devices in the industrial Internet of Things.

[0028] Among them, the network structure diagram of the industrial Internet of Things is as follows Figure 2 As shown, Figure 2 The Industrial Internet of Things (IIoT) consists of cloud servers, PON network layer, MEC (Mobile Edge Computing) server layer, and terminal layer. The PON network layer is a point-to-multipoint network structure consisting of the following three parts: 1. Optical Line Terminal (OLT)

[0029] 2. Optical Distribution Network (ODN): ODN is a passive device that connects OLT and ONU. It is composed entirely of passive components such as optical splitters and does not require an external power supply.

[0030] 3. Optical Network Unit (ONU), located on the user side, enables users to access the PON network and enjoy various services.

[0031] The Industrial Internet of Things (IIoT) needs to meet low-latency requirements, and the hierarchical control structure of Mobile Edge Computing (MEC) is particularly critical. This structure is a network architecture that pushes computing and storage resources to the edge of the network, close to user devices, to improve user experience and network efficiency. Furthermore, the IIoT in this application introduces a passive optical network structure, which not only provides significant bandwidth advantages and anti-interference capabilities, enabling long-distance, high-speed data transmission, but also offers cost and maintenance advantages by using passive components to build an optical distribution network (ODN).

[0032] To offload user-side tasks to the most appropriate MEC server node or to the cloud, it is necessary to collect environmental monitoring data from the entire Industrial Internet. This environmental monitoring data includes any of the following: MEC server load queue length, transmission rates, computing power, and mobile user positions. First, obtain the raw environmental monitoring data from the Industrial Internet at the previous moment.

[0033] As the number of network layers increases, there are more noise sources, such as communication resource limitations, equipment aging, sensor failures, etc. Therefore, the data collected in the network must be denoised first.

[0034] Step 102: The environmental monitoring data at the last moment ( ) is input into a particle Kalman cascade filter (PKCF), so that the particle Kalman cascade filter (PKCF) performs noise filtering on the environmental monitoring data at the previous moment to obtain noise-filtered environmental monitoring data.

[0035] The particle Kalman cascade filter (PKCF) combines the advantages of particle filtering (PF) and Kalman filtering (KF) to address state estimation problems in nonlinear and non-Gaussian systems. Its basic principle is to approximate the probability distribution of the system state using a collection of randomly sampled particles. Each particle represents a possible state of the system, and its weight reflects the likelihood of that state occurring. As new observation data arrives, the particle weights are updated based on the observation model and state transition model. Through operations such as resampling, particles with low weights are removed and particles with high weights are retained, thereby achieving dynamic estimation of the system state.

[0036] Specifically, the environmental monitoring data at the previous moment is input into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain the noise-filtered environmental monitoring data.

[0037] Step 103 : Based on the noise-filtered environmental monitoring data, a heuristic greedy calculation offloading algorithm is used to calculate the tasks to be offloaded of all mobile devices at the current moment to obtain an optimal task offloading solution.

[0038] Among them, the Heuristic Greedy Computing Offloading Algorithm (HGCO) is a task offloading strategy used in edge computing environments. It aims to optimize task offloading decisions to reduce energy consumption and latency while improving the overall performance of the system.

[0039] Specifically, the environmental monitoring data at the previous moment after noise filtering is input into the heuristic greedy computation offloading algorithm to obtain the optimal task offloading solution.

[0040] The above embodiment obtains environmental monitoring data from the previous moment; this environmental monitoring data includes environmental monitoring data from all mobile edge computing nodes, all cloud nodes, and all mobile devices in the Industrial Internet of Things. This environmental monitoring data is then input into a particle Kalman cascade filter, which then filters the noise from the environmental monitoring data to obtain noise-filtered environmental monitoring data. Based on the noise-filtered environmental monitoring data, a heuristic greedy computation offloading algorithm is used to calculate the tasks to be offloaded for all mobile devices at the current moment, resulting in an optimal task offloading solution. This application demonstrates excellent performance in mitigating the impact of data noise on task offloading in industrial PON networks. The proposed particle Kalman cascade filter (PKCF) effectively reduces data volatility and restores the original data distribution. Furthermore, it significantly reduces task completion latency, particularly in environments with high user density and severe impulse noise. These findings highlight the importance of data filtering (denoising) technology in ensuring the reliability of task offloading in industrial PONs, providing an applied foundation for subsequent theoretical research.

[0041] In one embodiment, the above step 101 includes: establishing a comprehensive noise model for the noise in the environmental monitoring data at the previous moment; the comprehensive noise model includes at least one of a Gaussian noise model, an impulse noise model, a sinusoidal noise model and a constant noise model.

[0042] Specifically, before performing noise filtering, a comprehensive noise model needs to be established to accurately reflect the multi-type noise in the industrial PON environment. This model is crucial for accurately reflecting the typical system data type and needs to include noise information such as the load queue length, transmission rate, computing power, and mobile user location of MEC (mobile edge computing). These parameters are of great significance in guiding computing offloading decisions and directly affect the results of offloading tasks and the allocation of network resources, such as Figure 3 As shown, Figure 3 The distribution of environmental monitoring information such as the transmission rate, computing capacity, task queue length, and user position of the MEC (Mobile Edge Computing) server before and after considering the noise model in the simulation phase is shown. In each chart, the original data ( Figure 3 Original in the data shows relatively stable fluctuations within a certain range, and there are no obvious abnormal values that deviate from the average value, which indicates the stability of the original data. Figure 3 The original data in the simulation are all the simulation environment information obtained in the simulation stage, that is, accurate data without noise. Figure 3 As can be seen from the figure, the data distribution range is significantly widened after the noise model is added. In particular, under the influence of impulse noise, the error in the noisy data compared to the original data increases significantly. Taking the computing power of the MEC server as an example (see the second horizontal subfigure), the original data fluctuates between 500 and 800 megacycles. This fluctuation is mainly affected by algorithm efficiency and internal server factors, and exhibits a certain degree of stability. However, after the noise model is introduced, the fluctuation range of computing power increases significantly, from 100 to 1200 megacycles. The noise model includes Gaussian noise with a standard deviation of 16 megacycles, impulse noise with a peak value between 400 and 450 megacycles, constant noise, and sinusoidal nonlinear noise. The presence of impulse noise, in particular, significantly increases the deviation between the noisy data and the original data, with the maximum error reaching 400 megacycles. This phenomenon simulates the data inaccuracy caused by sensor accuracy and environmental factors in real-world scenarios.

[0043] Figure 3 The chart shown intuitively demonstrates the distribution of system data collected in a real industrial environment and analyzes the impact of noise on MEC server and user performance indicators.

[0044] The above embodiment provides a reliable data model for subsequent data denoising by establishing a comprehensive noise model.

[0045] In one embodiment, the above step 102 specifically includes the following steps: Figure 4 As shown, Figure 4 A schematic diagram of the execution flow of the particle Kalman cascade filter (PKCF) is shown.

[0046] Step 401: take each data in the environmental monitoring data of the last moment as the particle state of a particle, randomly sample a particle set from the environmental monitoring data of the last moment, and calculate the particle state of each particle in the particle set. Assign weight ; in, Represents the i-th environmental monitoring data at the previous moment (i.e., moment t-1); Represents the initial weight of the i-th environmental monitoring data at the previous moment (i.e., moment t-1).

[0047] Step 402: Based on the comprehensive noise model, the particle state of each particle in the particle set is predicted to obtain the particle state of each particle at the current moment. The expression of state prediction is as follows: ; (1) in, is the i-th environmental monitoring data at the current moment (i.e., time t) (i.e., the particle state of the i-th particle at the current moment); represents the particle state of the i-th particle (i-th environmental monitoring data) at the previous moment (i.e., moment t-1), and noise is the process noise, which is used to describe the uncertainty in the system state transition model.

[0048] Step 403: Based on the particle state of each particle at the current moment and the weight of each particle at the previous moment , calculate the weight of each particle at the current moment (i.e., time t) .

[0049] Specifically, the state update expression is as follows: ; (2) in, represents the weight of the i-th environmental monitoring data at the previous moment (i.e., moment t-1); R is the covariance of the observation noise, which represents the uncertainty of the observation value or the size of the observation error, where the observation noise represents the error in the measurement link; is the particle state (i.e., environmental monitoring data) of the i-th particle at the current moment (i.e., time t), is the actual observation value at time t, that is, the noisy environmental monitoring data, that is, the environmental monitoring data containing comprehensive noise; Step 404: Based on the weight of each particle at the current moment , resampling is performed from the particle set to obtain a new particle set; Specifically, the resampling process includes: retaining particles with higher weights and discarding particles with lower weights with a certain probability, so that the new particle set can better represent the posterior probability distribution; Step 405 : performing weighted summation on the particle states of all particles in the new particle set according to the particle state of each particle in the new particle set and the weight of each particle to obtain a system state estimate.

[0050] Specifically, the process of system state estimation (System State Estimate) in particle filtering (Particle Filter) is shown as follows: ; (3) in, Represents the state estimate of the system at the current time t, that is, the system state estimate predicted by the particle filter, and serves as the input of the subsequent Kalman filter, that is, in formula (6) ; is the weight of the i-th particle at time t, is the particle state of the i-th particle at time t (i.e., the current time), and N is the total number of particles. It should be noted that the system refers to the mobile edge computing nodes and mobile devices in the industrial Internet of Things.

[0051] Step 406: The system state estimate As input to the Kalman filter, calculate the Kalman gain ; Among them, the system state estimate It is the system state estimate output by the particle filter; Specifically, the core of the Kalman filter is to estimate the state of a linear dynamic system through recursion, in two steps: prediction and update. The update step is based on a series of measurements observed over time, which in this model corresponds to the output of the particle filter. The basic process of the Kalman filter includes initializing the estimate, predicting state transitions, updating with measurement data, and correcting the estimation error.

[0052] First, the state prediction in Kalman filter: ; (4) in, is the covariance matrix of the prediction error at time t, which represents the uncertainty of the predicted state. The prediction error refers to the error between the predicted value of the system state and the actual state; is the covariance matrix of process noise, which represents the uncertainty of the system model, that is, the random disturbance in the system dynamic process. Q can be estimated through experimental data, historical data or system analysis.

[0053] State Update in Kalman Filter: ; (5) ; (6) ; (7) in, is the Kalman gain, R is the covariance matrix of the observation noise, which is used to characterize the error of the measurement equipment (such as sensor accuracy limitation, environmental interference); R represents the uncertainty of the actual observation value of the noisy environmental monitoring data, that is, the random disturbance in the observation process; is the actual observation value of the noisy environmental monitoring data at time t; is the state estimate at the tth moment output by the Kalman filter; The predicted state of the environmental monitoring data at time t is predicted based on the environmental monitoring data at time t-1, that is, the output of the particle filter at time t-1 ; The output of the particle filter As the input of Kalman filter, get the output of Kalman filter ; is the covariance matrix of the error of the updated system state estimation (specifically, environmental monitoring data), which represents the uncertainty of the updated state.

[0054] System State Estimate in Kalman Filter: ; (8) in, is the final state estimate at the tth moment output by the Kalman filter, that is, the state estimate after combining the prediction and observation information; is the state update for the tth time step; is the predicted state estimate; is the Kalman gain; is the observation matrix.

[0055] In this embodiment, a particle Kalman cascade filter (PKCF) is used to denoise the environmental monitoring data, which combines the advantages of particle filtering (PF) and Kalman filtering (KF) to improve the accuracy of system state estimation.

[0056] In one embodiment, the above-mentioned step 103 includes: obtaining the tasks to be offloaded of each mobile device at the current moment; calculating the average delay of all tasks to be offloaded at the current moment; wherein the average delay is calculated based on the distance between each mobile device and the MEC or the distance between each mobile device and the cloud server; using a heuristic greedy computing offloading algorithm (HGCO) to iteratively calculate the average delay to minimize the average delay, thereby obtaining an optimal task offloading solution for the mobile device at the current moment; and outputting the optimal task offloading solution.

[0057] Specifically, assume that there are M mobile devices and N MEC servers in the IIoT environment, where mobile device i generates a delay-sensitive and indivisible task in each time slot. ; The task is to use the first parameter group ( , ) represents; the first parameter group ( , ) including tasks Packet size ,Task Uninstall decision ,Task The number of CPU cycles required to complete ; ; (9) For tasks , calculate the total delay of unloading to each MEC server. The total delay consists of three parts: transmission delay, queuing delay and calculation delay. Total latency of offloading to the nth MEC server ( n 0-N), set the minimum value ; ; (10) in, Indicates a task The packet size; Representing mobile devices to MEC servers n Wireless transmission rate; Indicates the MEC server n The load queue length; Indicates the MEC server n The calculation rate of Indicates a task The number of CPU cycles required to complete.

[0058] Since the optimization goal is to minimize the average delay, it is necessary to calculate the average delay of M tasks corresponding to M mobile devices (that is, all mobile devices) at the current time t. , and normalize the average delay; ; (11) in, It represents the minimum total delay of completing the task generated by the i-th mobile device, where the total delay includes computation delay, queuing delay, and transmission delay.

[0059] For the nth MEC server among N MEC servers, it is necessary to ensure the total size of the data packets of all tasks offloaded to the MEC server. Lower than the storage capacity of the MEC server : ; (12) Among them, a binary variable is introduced To express the task Whether to offload to the nth server, when the task Uninstall decision =n, =1, at this time Indicates a task Unload to the nth MEC server; when the task Uninstall decision ≠n, =0, at this time Indicates a task Offload to a cloud server.

[0060] ; (13) Ensure that each MD (Mobile Device) task must and can only choose one offloading destination, whether it is the MEC server or the cloud.

[0061] The goal of this application is to fully utilize the centralized cloud and distributed MEC server resources to develop a collaborative computing offloading solution to minimize the average latency of all computing tasks, that is, to make the P1 Minimize. This embodiment adopts the heuristic greedy computation offloading algorithm (HGCO) to iteratively calculate the above-mentioned noise-filtered environmental monitoring data, which is the Pareto optimal solution that minimizes the average delay of formula (11) under the storage constraint of formula (12), and obtains the optimal task offloading solution for each mobile device at the current moment.

[0062] The above embodiment achieves the flexibility of task offloading through the heuristic greedy computation offloading algorithm (HGCO), which can balance the network load and prevent single-point overload of centralized cloud and multi-access edge computing.

[0063] In summary, the proposed strategy demonstrates excellent performance in mitigating the impact of data noise on task offloading in industrial PONs. The proposed PKCF significantly reduces task offloading latency, particularly in environments with high user density and severe impulse noise. These findings highlight the importance of data preprocessing and filtering techniques in ensuring the reliability of task offloading in industrial PONs.

[0064] The edge computing task offloading optimization device that is immune to data noise provided by the present invention is described below. The edge computing task offloading optimization device that is immune to data noise described below and the edge computing task offloading optimization method that is immune to data noise described above can be referenced to each other.

[0065] like Figure 6 As shown, the edge computing task offloading optimization device provided by the present invention, which is immune to data noise, includes the following modules: The data acquisition module 601 is used to obtain environmental monitoring data at the previous moment; the environmental monitoring data at the previous moment includes environmental monitoring data of all mobile edge computing nodes, all cloud nodes, and all mobile devices in the industrial Internet of Things; A data denoising module 602 is configured to input the environmental monitoring data at the previous moment into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain noise-filtered environmental monitoring data; The optimal task offloading solution output module 603 is configured to calculate the tasks to be offloaded of all mobile devices at the current moment using a heuristic greedy calculation offloading algorithm based on the noise-filtered environmental monitoring data to obtain an optimal task offloading solution.

[0066] In one embodiment, the device further includes a noise model building unit configured to: A comprehensive noise model is established for the noise in the environmental monitoring data at the previous moment; the comprehensive noise model includes at least one of a Gaussian noise model, an impulse noise model, a sinusoidal noise model and a constant noise model.

[0067] In one embodiment, the data denoising module 602 is further configured to: Taking each data in the environmental monitoring data at the previous moment as a particle state, randomly sampling from the environmental monitoring data at the previous moment to obtain particle states of a plurality of initial particles, and assigning an initial weight to each initial particle; Based on the comprehensive noise model, the states of the multiple initial particles are predicted to obtain the particle states at the current moment; Calculating the weight of the initial particle at the current moment according to the particle state at the current moment and the initial weight of the initial particle; Resampling from the multiple initial particles according to the weight of the initial particle at the current moment to obtain a particle set; The particle set is used as the input of the Kalman filter to obtain the final state estimation of the particles output by the Kalman filter at the current moment.

[0068] In one embodiment, the data denoising module 602 is further configured to: calculate a Kalman gain; update the state of the particle at the current moment according to the observation noise covariance matrix; and update the final state estimate of the particle at the current moment according to the Kalman gain.

[0069] In one embodiment, the optimal task offloading solution output module 603 is further configured to: The present invention obtains the tasks to be offloaded of each mobile device at the current moment; calculates the average delay of all the tasks to be offloaded at the current moment; and uses a heuristic greedy calculation offloading algorithm to iteratively calculate the noise-filtered environmental monitoring data to minimize the average delay, thereby obtaining the optimal task offloading solution for the mobile device at the current moment.

[0070] In one embodiment, the environmental monitoring data includes any one of the load queue length, transmission rate, computing power, and mobile user location of the MEC.

[0071] Figure 7 An example of a physical structure diagram of an electronic device is shown below. Figure 7 As shown, the electronic device may include: a processor 710, a communications interface 720, a memory 730, and a communications bus 740, wherein the processor 710, the communications interface 720, and the memory 730 communicate with each other via the communications bus 740. The processor 710 may call logic instructions in the memory 730 to execute an edge computing task offloading optimization method that is immune to data noise. The method includes: obtaining environmental monitoring data at a previous moment; the environmental monitoring data at a previous moment includes environmental monitoring data of all mobile edge computing nodes, all cloud nodes, and all mobile devices in the industrial Internet of Things; inputting the environmental monitoring data at a previous moment into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at a previous moment to obtain noise-filtered environmental monitoring data; based on the noise-filtered environmental monitoring data, using a heuristic greedy calculation offloading algorithm with the goal of minimizing average task delay, iteratively calculating the tasks to be offloaded for all mobile devices at the current moment, and generating an optimal task offloading solution that meets the storage capacity constraints of the MEC server.

[0072] Furthermore, the logic instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0073] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the edge computing task offloading optimization method that is immune to data noise provided by the above methods, and the method includes: obtaining environmental monitoring data at the previous moment; the environmental monitoring data at the previous moment includes the environmental monitoring data of all mobile edge computing nodes, all cloud nodes and all mobile devices in the industrial Internet of Things; inputting the environmental monitoring data at the previous moment into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain noise-filtered environmental monitoring data; based on the noise-filtered environmental monitoring data, using a heuristic greedy calculation offloading algorithm to calculate the tasks to be offloaded of all mobile devices at the current moment to obtain the optimal task offloading solution.

[0074] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the edge computing task offloading optimization method that is immune to data noise provided by the above-mentioned methods, the method comprising: obtaining environmental monitoring data at the previous moment; the environmental monitoring data at the previous moment includes the environmental monitoring data of all mobile edge computing nodes, all cloud nodes and all mobile devices in the industrial Internet of Things; inputting the environmental monitoring data at the previous moment into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain noise-filtered environmental monitoring data; based on the noise-filtered environmental monitoring data, using a heuristic greedy calculation offloading algorithm to calculate the tasks to be offloaded of all mobile devices at the current moment to obtain the optimal task offloading solution.

[0075] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0076] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An edge computing task offloading optimization method that is immune to data noise, characterized in that: include: Obtain environmental monitoring data from the previous moment; the environmental monitoring data from the previous moment includes environmental monitoring data from all mobile edge computing nodes, all cloud nodes, and all mobile devices in the industrial Internet of Things; Inputting the environmental monitoring data at the previous moment into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain noise-filtered environmental monitoring data; Based on the noise-filtered environmental monitoring data, a heuristic greedy calculation offloading algorithm is used to calculate the tasks to be offloaded of all mobile devices at the current moment to obtain an optimal task offloading solution.

2. The edge computing task offloading optimization method immune to data noise according to claim 1 is characterized in that: Before obtaining the environmental monitoring data at the previous moment, the following steps are included: A comprehensive noise model is established for the noise in the environmental monitoring data at the previous moment; the comprehensive noise model includes at least one of a Gaussian noise model, an impulse noise model, a sinusoidal noise model and a constant noise model.

3. The edge computing task offloading optimization method immune to data noise according to claim 2 is characterized in that: The step of inputting the environmental monitoring data at the previous moment into a particle Kalman cascade filter so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain the noise-filtered environmental monitoring data includes: Taking each data in the environmental monitoring data at the previous moment as the particle state of a particle, randomly sampling from the environmental monitoring data at the previous moment to obtain a particle set, and assigning a weight to the particle state of each particle in the particle set; Based on the comprehensive noise model, predict the particle state of each particle in the particle set and calculate the weight of each particle at the current moment; Calculate the weight of each particle at the current moment based on the particle state of each particle at the current moment and the weight of each particle at the previous moment; Resampling the particle set according to the weight of each particle at the current moment to obtain a new particle set; According to the particle state of each particle in the new particle set and the weight of each particle, a weighted sum is performed on the particle states of all particles in the new particle set to obtain a system state estimate.

4. The edge computing task offloading optimization method immune to data noise according to claim 3 is characterized in that: The particle set is used as the input of the Kalman filter to obtain the final state estimate of the particles output by the Kalman filter at the current moment, including: Calculate the Kalman gain; Update the state of the particle at the current moment according to the observation noise covariance matrix; According to the Kalman gain, the final state estimate of the particle at the current moment is updated.

5. The edge computing task offloading optimization method immune to data noise according to claim 1 is characterized in that: The method of calculating the tasks to be offloaded of all mobile devices at the current moment using a heuristic greedy calculation offloading algorithm based on the noise-filtered environmental monitoring data to obtain an optimal task offloading solution includes: Get the tasks to be uninstalled for each mobile device at the current moment; Calculate the average delay of all tasks to be offloaded at the current moment; The noise-filtered environmental monitoring data is iteratively calculated using a heuristic greedy calculation offloading algorithm to minimize the average delay and obtain an optimal task offloading solution for the mobile device at the current moment.

6. The edge computing task offloading optimization method immune to data noise according to claim 1 is characterized in that: The environmental monitoring data includes any one of the load queue length, transmission rate, computing power, and mobile user location of the mobile edge computing server.

7. An edge computing task offloading optimization device that is immune to data noise, characterized in that: include: A data acquisition module is used to obtain environmental monitoring data at the previous moment; the environmental monitoring data at the previous moment includes environmental monitoring data of all mobile edge computing nodes, all cloud nodes, and all mobile devices in the industrial Internet of Things; a data denoising module, configured to input the environmental monitoring data at the previous moment into a particle Kalman cascade filter, so that the particle Kalman cascade filter performs noise filtering on the environmental monitoring data at the previous moment to obtain noise-filtered environmental monitoring data; The optimal task offloading solution output module is used to calculate the tasks to be offloaded of all mobile devices at the current moment based on the noise-filtered environmental monitoring data using a heuristic greedy calculation offloading algorithm to obtain the optimal task offloading solution.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the edge computing task offloading optimization method that is immune to data noise as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the edge computing task offloading optimization method that is immune to data noise as described in any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the edge computing task offloading optimization method that is immune to data noise as described in any one of claims 1 to 6 is implemented.

Images