Delayed unloading method and system for calculation task

Abstract

The invention discloses a delay unloading method and system for a calculation task, and the method comprises the steps: obtaining a task calculation request initiated by a mobile terminal in a first region; constructing a multi-target optimization problem from four dimensions of time delay, energy consumption, benefit and cost according to the calculation task request, and solving the multi-target optimization problem to obtain an optimal MEC server or an optimal MEC server group of a first region and an optimal MEC server or an optimal MEC server group of a second region; whether the connection time of the mobile terminal and the first area is smaller than the task unloading time or not is judged, and if not, the task is unloaded to the optimal MEC server or the optimal MEC server group in the first area; if so, entering the next step; starting delayed unloading, and waiting for the mobile terminal to move to the second area; and unloading the task to the optimal MEC server or the optimal MEC server group in the second area.

Description

Technical Field

[0001] The present invention relates to the field of mobile communication technologies, and particularly to a method and system for delaying the offloading of computing tasks. Background Art

[0002] A mobile terminal generates a new computing task W within a certain period of time. Due to its limited computing power, in addition to local task processing, the mobile terminal will select a multi-access edge computing node (Multi-Access Edge Computing, abbreviated as MEC) for task offloading based on an optimization objective (assumed to be MEC i )

[0003] As Figure 1 shown, considering terminal mobility, MEC i does not complete the processing of task W within the connection time between the terminal and MEC i , and the terminal moves out of the service range of MEC i . Then the uncompleted task will be transmitted from the current MEC i to the next available MEC server (assumed to be MEC j ) for continued processing, and this process is task migration.

[0004] Due to the high mobility of terminal users in mobile scenarios, although task migration can optimize resource utilization and improve task processing performance, it will also bring some negative impacts. First, data retransmission is required during the migration process, which will increase the energy consumption, cost of task processing, and greatly increase the network load. Second, data loss or inconsistency may also occur during the process of transferring a task from one node to another, further increasing the complexity of task processing. The existing processing method is to jointly optimize task offloading and task migration using an optimization algorithm, and the specific algorithm is as follows:

[0005] Multi-edge collaborative offloading and energy consumption threshold strategy: This method analyzes and calculates the task execution energy consumption and delay costs of local terminals, edge servers, and central clouds, and improves the energy consumption, service completion time, and data transmission energy consumption of mobile servers based on an energy consumption threshold-based task migration strategy.

[0006] Distributed multi-agent reinforcement learning: By extending the multi-agent deep reinforcement learning (Multi-Agent Deep Reinforcement Learning, MADRL) method, the distributed task migration problem can be optimized. This method reduces the task completion time and migration energy consumption through the cooperation of cooperative agents.

[0007] Energy Efficiency, Cost-Effectiveness, and QoS-Aware Method: ECQ is a heuristic method aiming to optimize energy efficiency, cost-effectiveness, and Quality of Service (QoS) in a dynamic and time-sensitive MEC environment. This method provides an efficient task migration solution by comprehensively considering energy consumption, cost, and QoS.

[0008] In mobile edge computing, although joint optimization algorithms can reduce latency and resource consumption, their complexity requires more computing resources. At the same time, these optimization algorithms search for the optimal offloading and migration solutions within the existing model framework, and cannot avoid the energy consumption, cost, and excessive retransmission data volume caused by migration between MECs in the case of a large task volume, which may lead to network congestion and even possible mistransmission. Summary of the Invention

[0009] To solve the deficiencies of the prior art, the present invention provides a method and system for delaying the offloading of computing tasks; the present application proposal aims to solve the deficiencies in the existing solutions, especially the challenges faced when migrating tasks between MECs. By judging whether the connection time between the terminal and the MEC meets the time required for the terminal task offloading, the offloading strategy is reasonably selected to ensure that task calculation can still be efficiently performed in the case of a large task volume. This can not only avoid the negative impacts brought by migration, such as high energy consumption, increased cost, and data transmission errors, but also has a simple judgment without occupying a large amount of computing resources, and can also improve the overall performance of task processing, thereby providing a better service experience for users.

[0010] On the one hand, a method for delaying the offloading of computing tasks is provided, including:

[0011] Obtain the task calculation request initiated by the mobile terminal when in the first area;

[0012] According to the computing task request, construct a multi-objective optimization problem from four dimensions of time delay, energy consumption, benefit, and cost, solve the multi-objective optimization problem to obtain the optimal MEC server or optimal MEC server group in the first area, and also obtain the optimal MEC server or optimal MEC server group in the second area;

[0013] Judge whether the connection time between the mobile terminal and the first area is less than the task offloading time. If not, offload the task to the optimal MEC server or optimal MEC server group in the first area; if so, proceed to the next step;

[0014] Initiate delayed offloading and wait for the mobile terminal to move to the second area;

[0015] Offload the task to the optimal MEC server or optimal MEC server group in the second area.

[0016] On the other hand, a latency offloading system for computing tasks is provided, including:

[0017] An acquisition module, configured to: acquire a task computing request initiated when the mobile terminal is in the first area;

[0018] A solution module, configured to: construct a multi-objective optimization problem from four dimensions of latency, energy consumption, benefit, and cost according to the computing task request, solve the multi-objective optimization problem, obtain the optimal MEC server or the optimal MEC server group in the first area, and also obtain the optimal MEC server or the optimal MEC server group in the second area;

[0019] A judgment module, configured to: judge whether the connection time between the mobile terminal and the first area is less than the task offloading time. If not, offload the task to the optimal MEC server or the optimal MEC server group in the first area; if so, proceed to the next step;

[0020] A waiting module, configured to: initiate latency offloading and wait for the mobile terminal to move to the second area;

[0021] An offloading module, configured to: offload the task to the optimal MEC server or the optimal MEC server group in the second area.

[0022] On yet another aspect, an electronic device is further provided, including:

[0023] A memory for non-temporarily storing computer-readable instructions; and

[0024] A processor for running the computer-readable instructions,

[0025] wherein, when the computer-readable instructions are run by the processor, the method described in the first aspect above is executed.

[0026] On yet another aspect, a storage medium is further provided, which non-temporarily stores computer-readable instructions. When the non-temporary computer-readable instructions are executed by a computer, the method described in the first aspect is executed.

[0027] On yet another aspect, a computer program product is further provided, including a computer program, and the computer program is used to implement the method described in the first aspect above when running on one or more processors.

[0028] The above technical solutions have the following advantages or beneficial effects:

[0029] 1. The latency offloading strategy can significantly improve the task processing benefit. Through this strategy, the benefit has increased by 37.78%, which not only optimizes resource utilization but also improves the user experience.

[0030] 2. The delayed offloading strategy performs excellently in reducing energy consumption. Compared with the normal offloading strategy, the energy consumption is reduced by 7.32%. This strategy reduces the energy consumption of task processing by reducing unnecessary task migrations.

[0031] 3. The delayed offloading strategy also has significant advantages in reducing costs. Compared with the normal offloading strategy, the cost is reduced by 6.33%. It reduces the additional costs brought by task migrations. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings forming a part of this invention are used to provide a further understanding of the invention. The schematic embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0033] Figure 1 It is a schematic diagram of the task migration process for Embodiment 1;

[0034] Figure 2 It is a flowchart of the delayed offloading for Embodiment 1;

[0035] Figure 3 It is a flowchart of the delayed offloading strategy in a mobile scenario for Embodiment 1;

[0036] Figure 4 It is a comparison chart of task volume - benefits of delayed offloading and normal offloading for Embodiment 1;

[0037] Figure 5 It is a comparison chart of connection time - benefits of delayed offloading and normal offloading for Embodiment 1;

[0038] Figure 6 It is a comparison chart of multi - user - benefits of delayed offloading and normal offloading for Embodiment 1;

[0039] Figure 7 It is a comparison chart of multi - user - energy consumption of delayed offloading and normal offloading for Embodiment 1;

[0040] Figure 8 It is a comparison chart of multi - user - costs of delayed offloading and normal offloading for Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the invention belongs.

[0042] Embodiment 1

[0043] This embodiment provides a method for delayed offloading of computing tasks;

[0044] A method for delayed offloading of computing tasks, comprising:

[0045] S101: Obtain the task calculation request initiated when the mobile terminal is in the first area;

[0046] S102: According to the calculation task request, construct a multi-objective optimization problem from four dimensions of delay, energy consumption, benefit, and cost, solve the multi-objective optimization problem, obtain the optimal MEC server or the optimal MEC server group in the first area, and also obtain the optimal MEC server or the optimal MEC server group in the second area (i.e., the optimal MEC i or the corresponding MEC group and MEC j or the corresponding MEC group);

[0047] S103: Determine whether the connection time between the mobile terminal and the first area is less than the task offloading time. If not, offload the task to the optimal MEC server or the optimal MEC server group in the first area; if so, proceed to the next step S104;

[0048] S104: Start delayed offloading and wait for the mobile terminal to move to the second area;

[0049] S105: Offload the task to the optimal MEC server or the optimal MEC server group in the second area.

[0050] Furthermore, the S101: Obtain the task calculation request initiated when the mobile terminal is in the first area;

[0051] Among them, the mobile terminal refers to the user's terminal device;

[0052] Among them, the first area refers to the area covered by the service provided by the first base station; the second area refers to the area covered by the service provided by the second base station;

[0053] Among them, the task calculation request includes: training tasks of machine learning models, etc.

[0054] Furthermore, the S102: According to the calculation task request, construct a multi-objective optimization problem from four dimensions of delay, energy consumption, benefit, and cost. Among them, the multi-objective optimization problem includes:

[0055] Task processing delay is:

[0056]

[0057] s.t.

[0058]

[0059] 0 < x i < W, i ∈ N;

[0060] Among them, x i is the task volume that the user plans to offload to the MEC i , is the connection time between the mobile user and the MEC i ; is the transmission time when the task volume x i is offloaded to the MEC i ; is the computing time when the task volume x i is in the MCE i ; E j is the energy consumption of the end user; is the upper limit of the end user's energy consumption; W is the total task volume; D m is the maximum duration of task processing; is the task processing delay at

[0061] is the task processing delay at

[0062] is the task processing delay at;

[0063] The energy consumption of task processing is:

[0064]

[0065] s.t.

[0066]

[0067] 0 < x i < W, i ∈ N;

[0068] is the energy consumption of the end user for task processing; is the energy consumption of task transmission; is the energy consumption of task computing in the MEC i ; is the energy consumption of task computing in the MEC j ; is the energy consumption of task migration.

[0069] The task processing cost C i,j :

[0070]

[0071] s.t.

[0072]

[0073] 0 < x i <W, i ∈ N;

[0074] is the task transmission cost; is the task computing cost; is the storage cost; is the migration cost; is the energy consumption cost;

[0075] Task processing benefit

[0076]

[0077] s.t.

[0078]

[0079]

[0080] 0 < x i <W, i ∈ N.

[0081] Wherein, is the task processing delay benefit; η i represents the delay benefit coefficient; represents the task processing energy consumption benefit; η j is the energy consumption benefit coefficient.

[0082] Furthermore, the task processing delay is obtained based on the task communication delay, task computing delay, and task migration delay:

[0083]

[0084] Specifically:

[0085]

[0086] Wherein, is the communication delay required for task transmission between the mobile user and the MEC i T HO is the user handover delay, k is the MEC storage ratio, is the uplink transmission rate between the user and the MEC j ; is the MEC i computing delay, is the MEC j computing delay, ρ is the conversion coefficient between the calculation result and the original task, is the downlink transmission rate between the user and the MEC j T l comp (xi ) Calculate the latency for the user terminal, is the migration latency, is for the mobile user and the MEC j The communication latency required for task transmission between them.

[0087] Furthermore, the calculation process of the task communication latency is as follows:

[0088] (1) The communication latency required for task transmission between the mobile user and the MEC i is expressed as:

[0089]

[0090] In the formula: is the task volume x i unloaded to the MEC i The transmission time, k is the MEC storage ratio;

[0091] is the task volume x i in the MEC i The calculation time;

[0092] is the calculation result ρx i transmitted from the MEC i to the mobile user; is the MEC i processing the task volume x i The required time;

[0093] Among them, x i is the task volume that the user plans to unload to the MEC i ; is the connection time between the mobile user and the MEC i ; is the MEC i Uplink transmission rate; is the MEC i Downlink transmission rate; V Ci is the MEC i Calculation rate; ρ is the conversion coefficient between the calculation result and the original task.

[0094] According to the connection time between the mobile terminal and the MEC server, it is divided into four scenarios, as follows:

[0095] Scenario 1 is:

[0096]

[0097] If the mobile terminal and MEC i If the connection time is less than the time required for task offloading, the mobile terminal will be moved to the second area and the task will be offloaded to MEC. j , the task is carried out by MEC j Calculate and send the result back to the mobile terminal.

[0098] Scenario 2:

[0099]

[0100] The mobile terminal offloads all tasks to MEC i , in MEC j The mobile terminal leaves the MEC during calculation i Service scope, MEC i The calculation results need to be migrated to MEC j , by MEC j The calculation results are sent back to the mobile terminal.

[0101] Scenario three is:

[0102]

[0103] The mobile terminal offloads all tasks to MEC i , MEC i Complete the task calculation in MEC i When the result is sent back to the mobile terminal, the mobile terminal leaves the MEC i Service scope, MEC i The remaining calculation results need to be migrated to MEC j , by MEC j The remaining calculation results are sent back to the mobile terminal.

[0104] Scene 4 is:

[0105]

[0106] The mobile terminal offloads all tasks to MEC i , MEC i Complete the task calculation and send the task result back to the mobile terminal.

[0107] (2) Mobile terminals and MEC j The communication delay required for task transmission between It is expressed as:

[0108]

[0109] Where: For users and MEC j Uplink transmission rate of For the downlink transmission rate between the user and the MEC j .

[0110] Furthermore, the task computing delay includes:

[0111] (1) The computing delay of the MEC i is expressed as: Denoted as:

[0112]

[0113] (2) The computing delay of the MEC j is expressed as: Denoted as:

[0114]

[0115] where V Cj is the computing rate of the MEC j .

[0116] (3) The computing delay T of the user terminal l comp (x i ) is expressed as:

[0117]

[0118] where W is the total amount of tasks, and V Cl is the computing rate of the user terminal

[0119] Furthermore, the task migration delay includes:

[0120] Since the connection time between the mobile terminal and the MEC i is random, when the connection time is insufficient, that is, when the MEC cannot complete the task transmission or task computing within the connection time, task migration will occur. Then the migration delay i is expressed as: Denoted as:

[0121]

[0122] where R i,j is the transmission rate between the MEC i and the MEC j , and T HO is the user handover delay

[0123] Furthermore, for the energy consumption of task processing, only the energy consumed by the user terminal and MEC server in computing is considered, and the energy consumed by the MEC server in transmission is ignored. Therefore, the task processing energy consumption is obtained based on the local computing energy consumption, task transmission energy consumption, and MEC computing energy consumption. The calculation process of task processing energy consumption specifically includes the following steps:

[0124] Calculate the local computing energy consumption. The local computing energy consumption is expressed as:

[0125]

[0126] where: P Cl is the local computing energy consumption per bit.

[0127] The task transmission energy consumption is expressed as:

[0128]

[0129] where, P Rl is the local transmission / reception energy consumption per bit.

[0130] The energy consumption on the terminal side is obtained based on the local computing energy consumption and task transmission energy consumption as:

[0131]

[0132] The task migration energy consumption is expressed as:

[0133]

[0134] where, P Re is the energy consumption per bit for transmission between edge servers.

[0135] MEC i Computing energy consumption is expressed as:

[0136]

[0137] MEC j Computing energy consumption is expressed as:

[0138]

[0139] where, P ci is the MEC i energy consumption per unit time; P Cj is the MEC j energy consumption per unit time.

[0140] The task processing energy consumption is obtained based on the local computing energy consumption, task transmission energy consumption, task migration energy consumption, and multi-access edge computing node computing energy consumption; the task processing energy consumption is expressed as:

[0141]

[0142] Furthermore, the task processing cost includes: communication cost, computing cost, storage cost, migration cost, and energy consumption cost.

[0143] Communication cost is:

[0144]

[0145] where α is the communication cost per unit time.

[0146] Computing cost is:

[0147]

[0148] where β is the computing cost per unit task.

[0149] Storage cost is:

[0150]

[0151] where σ is the storage cost per unit quantity.

[0152] Migration cost is:

[0153]

[0154] where γ is the migration cost per unit quantity.

[0155] Energy consumption cost is:

[0156]

[0157] where ε is the energy consumption cost per unit energy consumption.

[0158] The task processing cost is obtained based on the communication cost, computing cost, storage cost, migration cost, and energy consumption cost. The task processing cost is expressed as:

[0159]

[0160] Furthermore, the task processing benefit is defined as the benefit improved when the task is offloaded to the MEC server compared to the mobile UE independently processing the task; the task processing benefit includes: latency benefit, energy consumption benefit;

[0161] Time delay benefit of task processing It is expressed as:

[0162]

[0163] Energy consumption benefit of task processing It is expressed as:

[0164]

[0165] Task processing benefit is the weighted sum of time delay benefit and energy consumption benefit:

[0166]

[0167] where η i and η j both represent weights.

[0168] The problem under consideration is formulated as a mixed-integer non-linear programming (MINLP) that involves the joint optimization of task offloading decisions, initial offloading to MEC i and migration to MEC j Due to the combinatorial nature of this problem, it is difficult and impractical to solve for the optimal solution. To overcome this shortcoming, we introduce the sparrow optimization algorithm to solve it.

[0169] Furthermore, the step S102: Solve the multi-objective optimization problem to obtain the optimal MEC server or the optimal group of MEC servers in the first region, and also obtain the optimal MEC server or the optimal group of MEC servers in the second region, includes:

[0170] Perform task allocation through the sparrow optimization algorithm;

[0171] Considering that the task offloading volume is between 0 - W, the value of a single sparrow is defined as a one-dimensional vector.

[0172] The value of each sparrow represents a randomly distributed task offloading volume offloaded to MEC.

[0173] (1) Set the initial offloading to MEC as M = {M1, M2,..., M m} and the migration to MEC as N = {N1, N2,..., N m}, initialize the proportions of discoverers, followers, and vigilant ones, and the population size P;

[0174] (2) Define the best position of the population sparrows as P best , at the iteration times k = 1, 2,... k maxAmong them, the sparrows in the population calculate their current fitness value F through the fitness function; the fitness function is used to evaluate the optimization goal of each sparrow (i.e., the benefit of the task offloading scheme);

[0175] (3) Compare the fitness value F of the current sparrow with the fitness values of all sparrows in the population, select the sparrow with the optimal fitness value, and set its corresponding position as the new P best ;

[0176] (4) Compare the current optimal fitness value with the historical optimal fitness value. If the current optimal fitness value is better than the historical optimal fitness value, update the historical optimal fitness value, and set the position and value of the current sparrow as the historical optimal position;

[0177] (5) Repeat steps (3) and (4) until the number of iterations reaches k = k max , by evaluating the optimal fitness values of all historical sparrows, select the optimal sparrow position corresponding to the optimal fitness value. Complete the task offloading decision and find the best offloading scheme that maximizes the task processing benefit.

[0178] It should be understood that the selection of the optimal fitness value as the maximum or minimum is confirmed according to the optimization goal. For example, when the optimization goal is the minimum delay, the minimum delay is the optimal fitness value; when the optimization goal is the maximum benefit, the maximum benefit is the optimal fitness value.

[0179] Furthermore, in S103: Determine whether the connection time between the mobile terminal and the first area is less than the task offloading time; where the connection time refers to the time when the mobile terminal can perform data transmission with the optimal MEC server or the optimal MEC server group in the first area.

[0180] The task offloading time is calculated as follows:

[0181]

[0182] Furthermore, in S104: If the user connection time is less than the time for offloading the task to the MEC i , start delayed offloading. Determine the current cell or base station where the mobile terminal is located according to the CELL ID reported by the mobile terminal, and then determine the area where the mobile terminal has moved based on the above cell or base station. After waiting for the mobile terminal to move to the second area, offload the task to the MEC group or MEC server in the current area; if the user connection time is greater than or equal to the time for offloading the task to the MEC i , then offload the task in the first area.

[0183] It should be understood that the second area is the area covered by the service provided by the second base station.

[0184] The delayed offloading strategy is asFigure 2 As shown in the figure. First, the mobile terminal initiates a task request in the first area. Then, considering four dimensions of time delay, energy consumption, benefit, and cost, the optimal initial offloading MEC server or MEC server group (assumed to be MEC i ) and the optimal migrating MEC or MEC group in the second area (assumed to be MEC j ) are selected. Subsequently, it is determined whether the connection time between the terminal and the initial offloading MEC i is less than the time required for task offloading. If the connection time is less than the task offloading time, then delayed offloading is performed. When the terminal moves to the second area, the task is offloaded to the migrating MEC j for task processing; if the connection time is greater than the task offloading time, then the task is offloaded to the initial offloading MEC i for task processing.

[0185] In the mobile scenario, the delayed offloading strategy, as Figure 3 shown, when the task offloading time is greater than the connection time between the terminal and the MEC j , by delaying the task offloading to a set time point, preferably, waiting until connecting to the next MEC j and then performing task offloading processing, it avoids the increase in energy consumption, cost, and network load brought about by frequent task migrations.

[0186] As Figure 4 shown, it shows the change of the system benefit of the delayed offloading and normal offloading strategies with the increase of the task volume. In the case of a small task volume, the task processing benefits of the two strategies are the same because there is no task migration or a small amount of migrated tasks between the multi-access edge computing (MEC) at this time, so there is no significant difference in task processing benefits. However, when the task volume is large, the task processing benefit of the delayed offloading is higher than that of the normal offloading. In Scenario 1, it increases by 36.18% on average. This is because the delayed offloading strategy avoids a large number of task migrations caused by insufficient connection time between MECs, thereby reducing energy consumption and cost. Therefore, the delayed offloading strategy improves the task processing benefit by reducing the energy consumption and cost brought about by task migrations in the case of a high task volume.

[0187] As Figure 5 shown, it shows the change of the system benefit of the delayed offloading and normal offloading with the change of the connection time. When the connection time is short, the task processing benefit of the delayed offloading strategy is higher than that of the normal offloading strategy, increasing by 25.32% on average in Scenario 1. This is because in the case of insufficient connection time, the terminal tasks cannot be fully uploaded, resulting in frequent task migrations between MECs, and thus the energy consumption and cost increase greatly. The delayed offloading strategy reduces the energy consumption and cost by reducing the occurrence of task migrations, thereby making its task processing benefit higher than that of the normal offloading strategy.

[0188] As Figure 6 , Figure 7 and Figure 8 shown, in a multi-user environment, the delay offloading strategy is superior to the normal offloading strategy in terms of benefits, energy consumption, and cost. In a multi-user environment, the task volume and connection time of the terminal are random. The delay offloading strategy can provide better service in this case. This is because the delay offloading strategy can effectively reduce task migration caused by insufficient connection time or excessive task volume, thereby reducing energy consumption and cost and improving task processing benefits. Regardless of the task volume or connection time, the delay offloading strategy can better adapt to changes and provide more stable and efficient services for users. In a multi-user scenario, compared with the normal offloading strategy, the delay offloading strategy improves the benefits by an average of 37.78%, reduces the energy consumption by an average of 7.32%, and reduces the cost by an average of 6.33%.

[0189] A delay offloading strategy is proposed for task processing. That is: when the connection time between the terminal and the initial offloading MEC is short, task offloading is not performed temporarily, and task offloading processing is carried out after moving to the migration MEC range. This method can reduce energy consumption, cost, and network load, improve task processing benefits, thereby reducing the risk of data loss and transmission errors, and improving the success rate and efficiency of task offloading.

[0190] For the proposed delay offloading strategy, the present invention proposes a task processing method for delay offloading. According to the connection time between the mobile terminal and the initial offloading MEC i , the task offloading is divided into four scenarios. Based on the above 4 scenarios, considering communication, computing, storage and other resources as a whole, the optimization problem of the dynamic task offloading and migration strategy is proposed, and the sparrow optimization algorithm is used to select the task offloading node and allocate resources. Finally, the task offloading decision, the initial offloading MEC i and the migration MEC j are output.

[0191] Based on different scenario requirements, a delay offloading task processing time delay, energy consumption, cost, and benefit model is constructed. The task processing cost model includes communication cost, computing cost, storage cost, migration cost, and energy consumption cost. Among them, the communication cost includes the communication cost between the mobile terminal and the MEC i , the communication cost between the mobile terminal and the MEC j and the communication cost of the mobile terminal; the computing cost includes the computing cost of the mobile terminal, the MEC i , the MEC j ; the storage cost is the storage cost for the MEC to store tasks; the migration cost is due to the insufficient connection time between the mobile terminal and the MEC i , which is not enough to process the task on the MEC i , and the remaining tasks need to be migrated to the MEC jThe generated costs above. The energy consumption cost includes the mobile terminal, MEC i , MEC j The computing energy consumption cost of, as well as the energy consumption cost of the mobile terminal uploading the task volume and the task migration between MECs.

[0192] For the optimization problem with energy consumption, cost, benefit, and delay as the optimization objectives, the Sparrow Search Algorithm is introduced. The application of the Sparrow Search Algorithm (SSA) in task offloading can effectively find the optimal offloading scheme and avoid falling into local optima. By simulating the foraging behavior of sparrows, the Sparrow Search Algorithm can search globally, dynamically adjust the task offloading strategy, thereby improving the overall performance of task processing and resource utilization. In the mobile edge computing environment, the Sparrow Search Algorithm can intelligently decide which tasks should be offloaded to the edge server according to the real-time network conditions and computing resources, ensuring that the tasks are completed efficiently while reducing energy consumption and costs. In this way, the Sparrow Search Algorithm can not only optimize the task offloading process but also improve the reliability and stability of task processing.

[0193] Embodiment 2

[0194] This embodiment provides a delay offloading system for computing tasks, including:

[0195] An acquisition module, which is configured to: acquire the task computing request initiated by the mobile terminal when it is in the first area;

[0196] A solution module, which is configured to: construct a multi-objective optimization problem from four dimensions of delay, energy consumption, benefit, and cost according to the computing task request, solve the multi-objective optimization problem, and obtain the optimal MEC server or the optimal group of MEC servers in the first area, and also obtain the optimal MEC server or the optimal group of MEC servers in the second area;

[0197] A judgment module, which is configured to: judge whether the connection time between the mobile terminal and the first area is less than the task offloading time. If not, offload the task to the optimal MEC server or the optimal group of MEC servers in the first area; if so, proceed to the next step;

[0198] A waiting module, which is configured to: initiate delay offloading and wait for the mobile terminal to move to the second area;

[0199] An offloading module, which is configured to: offload the task to the optimal MEC server or the optimal group of MEC servers in the second area.

[0200] It should be noted here that the above-mentioned acquisition module, solution module, judgment module, waiting module, and unloading module correspond to steps S101 to S105 in the first embodiment. The examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the content disclosed in the first embodiment. It should be noted that the above modules, as part of the system, can be executed in a computer system such as a set of computer-executable instructions.

[0201] In the above embodiments, the descriptions of each embodiment have their own emphases. For parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0202] The proposed system can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the above-mentioned module division is only a logical function division. In actual implementation, there can be other division methods. For example, multiple modules can be combined or integrated into another system, or some features can be ignored or not executed.

[0203] Embodiment Three

[0204] This embodiment also provides an electronic device, including: one or more processors, one or more memories, and one or more computer programs; wherein, the processor is connected to the memory, and the above one or more computer programs are stored in the memory. When the electronic device runs, the processor executes the one or more computer programs stored in the memory so that the electronic device executes the method described in the first embodiment above.

[0205] It should be understood that in this embodiment, the processor can be a central processing unit CPU, and the processor can also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0206] The memory can include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory can also include a non-volatile random memory. For example, the memory can also store information about the device type.

[0207] In the implementation process, each step of the above method can be completed by the integrated logic circuit in the hardware of the processor or the instructions in the form of software.

[0208] The method in the first embodiment can be directly implemented by a hardware processor, or by a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, registers, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0209] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with this embodiment can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

[0210] Embodiment 4

[0211] This embodiment also provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the method described in the first embodiment is completed.

[0212] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A delayed unloading method for a computing task, characterized in that: include: Obtaining a task calculation request initiated by the mobile terminal in the first area; According to the computing task request, a multi-objective optimization problem is constructed from four dimensions of latency, energy consumption, benefit and cost, and the multi-objective optimization problem is solved to obtain an optimal MEC server or an optimal MEC server group in the first area, and also to obtain an optimal MEC server or an optimal MEC server group in the second area; Determine whether the connection time between the mobile terminal and the first area is less than the task offloading time, and if not, offload the task to the optimal MEC server or optimal MEC server group in the first area; If yes, go to the next step; Initiate delayed unloading and wait for the mobile terminal to move to the second area; Offload the task to the optimal MEC server or optimal MEC server group in the second region.

2. A delayed unloading method for computing tasks as claimed in claim 1, characterized in that: According to the computing task request, a multi-objective optimization problem is constructed from four dimensions: latency, energy consumption, benefit and cost. The multi-objective optimization problem includes: Task processing delay for: Among them, x i Plan to offload to MEC for users i The amount of tasks, For mobile users and MEC i Connection time; The amount of task x i Offloading to MEC i The transmission time; The amount of task x i In MEC i The calculation time of E l Energy consumption for end users; is the upper limit of energy consumption of the end user; W is the total task volume; D m Set the maximum duration for the task. The processing task delay is for The processing task delay is for The processing task delay is Task processing energy consumption for: Energy consumption for processing tasks for end users; Transmit energy consumption for tasks; For MEC i Task calculation energy consumption; For MEC j Task calculation energy consumption; Migrate energy consumption for tasks; Task processing cost C i,j : Transmit costs for tasks; Calculate costs for tasks; For storage costs; For migration costs; For energy consumption cost; Task processing benefits in, is the task processing delay benefit; η i represents the delay benefit coefficient; represents the energy efficiency of task processing; η j is the energy efficiency coefficient.

3. A delayed unloading method for computing tasks as claimed in claim 2, characterized in that: The task processing delay is obtained based on the task communication delay, task calculation delay and task migration delay: Specifically: in, For mobile users and MEC i The communication delay required for task transmission between HO is the user switching delay, k is the MEC storage ratio, For users and MEC j The uplink transmission rate is For MEC i Calculate the delay, For MEC j Calculation delay, ρ is the conversion coefficient between the calculation result and the original task, For users and MEC j The downlink transmission rate is Calculate the delay for the user terminal, is the migration delay, For mobile users and MEC j The communication delay required for task transmission between them.

4. A delayed unloading method for computing tasks as claimed in claim 3, characterized in that: The calculation process of task communication delay is as follows: (1) Mobile Users and MEC i The communication delay required for task transmission between two nodes is expressed as: Where: The amount of task x i Offloading to MEC i The transmission time is k, and k is the MEC storage ratio; The amount of task x i In MEC i The calculation time of The calculation result is ρx i From MEC i Time of transmission to mobile users; For MEC i Processing task volume x i Time required; Among them, x j Plan to offload to MEC for users j The amount of tasks; For mobile users with MCE j Connection time; For MEC i Uplink transmission rate; For MEC i Downlink transmission rate; V Ci For MEC i The calculation rate; ρ is the conversion coefficient between the calculation result and the original task; Mobile Terminal and MCE j The communication delay required for task transmission between It is expressed as: Where: For users with MCE j Uplink transmission rate of For users with MCE j downlink transmission rate.

5. A delayed unloading method for computing tasks as claimed in claim 3, characterized in that: Task calculation latency, including: (1)MCE i Calculating latency It is expressed as: (2)MCE j Calculating latency It is expressed as: Among them, V Cj For MCE j The calculation rate of (3) User terminal calculation delay It is expressed as: Among them, W is the total amount of tasks, V Cl Calculate the rate for the user terminal.

6. A delayed unloading method for computing tasks as claimed in claim 3, characterized in that: Task migration delay It is expressed as: Among them, R i,j For MEC i With MEC j The transmission rate between HO Switching delay for users.

7. A delayed unloading method for computing tasks as claimed in claim 2, characterized in that: The calculation process of task processing energy consumption specifically includes the following steps: Calculate local computing energy consumption, local computing energy consumption It is expressed as: Where: P Cl Calculate the energy consumption per bit locally; Task transmission energy consumption It is expressed as: Among them, P Rl The energy consumption per bit of local transmission / reception; The energy consumption on the terminal side is obtained based on the local computing energy consumption and task transmission energy consumption for: Task Migration Energy Consumption It is expressed as: Among them, P Re The energy consumption of transmitting a single bit between edge servers; MEC i Calculating energy consumption It is expressed as: MEC j Calculating energy consumption It is expressed as: Among them, P Ci For MEC i Energy consumption per unit time; P Cj For MEC j Energy consumption per unit time; The task processing energy consumption is obtained based on the local computing energy consumption, task transmission energy consumption, task migration energy consumption, and multi-access edge computing node computing energy consumption; the task processing energy consumption is expressed as: The task processing cost includes: communication cost, computing cost, storage cost, migration cost and energy consumption cost; communication cost for: Where α is the communication cost per unit time; Calculate costs for: Where β is the computational cost of a unit task; Storage costs for: Where σ is the storage cost per unit quantity; Migration costs for: Where γ is the migration cost per unit quantity; Energy cost for: Among them, ε is the energy cost per unit of energy consumption; The task processing cost is obtained based on the communication cost, computing cost, storage cost, migration cost and energy cost; the task processing cost is expressed as: Task processing benefit is defined as the benefit that is improved when the task is offloaded to the MEC server compared to the mobile UE independently processing the task. Task processing benefit includes: latency benefit and energy consumption benefit. Latency Benefits of Task Processing It is expressed as: Energy efficiency of task processing It is expressed as: Task processing benefits It is the weighted sum of delay benefit and energy consumption benefit: Among them, η i and η j Both represent weights.

8. A delayed unloading system for computing tasks, characterized in that: include: An acquisition module is configured to: acquire a task calculation request initiated by the mobile terminal when the mobile terminal is in the first area; A solution module is configured to: construct a multi-objective optimization problem from four dimensions of latency, energy consumption, benefit and cost according to the computing task request, solve the multi-objective optimization problem, obtain an optimal MEC server or an optimal MEC server group in the first area, and obtain an optimal MEC server or an optimal MEC server group in the second area; A judgment module, configured to: judge whether the connection time between the mobile terminal and the first area is less than the task offloading time, and if not, offloading the task to the optimal MEC server or optimal MEC server group in the first area; If yes, go to the next step; A waiting module is configured to: start delayed unloading and wait for the mobile terminal to move to the second area; An offloading module is configured to offload the task to the optimal MEC server or the optimal MEC server group in the second area.

9. An electronic device, comprising: a memory for non-transitory storage of computer readable instructions; as well as a processor for executing the computer readable instructions, When the computer-readable instructions are executed by the processor, the method described in any one of claims 1 to 7 is executed.

10. A storage medium, characterized in that: The computer-readable instructions are non-transitory stored, wherein when the non-transitory computer-readable instructions are executed by a computer, the method according to any one of claims 1 to 7 is performed.

Images