Task pre-migration strategy in mobile edge computing environment

Abstract

The present invention discloses a task pre-migration strategy in a mobile edge computing environment, including the following steps: Step 1, establish a judgment direction model, establish a hexagonal cellular structure with the base station of the corresponding edge server as the center, and the coverage of the edge server forms the same as the coverage of the communication network, and obtain a user movement direction prediction model; Step 2, determine the relative position between the user and the base station of the current edge server, so as to predict the new edge server that may need to receive the task migration after the user moves; Step 3, perform relevant pre-migration operations on the new edge server determined in the previous step. The present invention is less affected by the dynamics of the user group, and the prediction model of the method is simple, the calculation amount is small, and the object of pre-migration can be quickly determined.

Description

Technical Field

[0001] The present invention belongs to the technical field of cloud computing, and in particular relates to a task pre-migration strategy in a mobile edge computing environment. Background Art

[0002] In the mobile edge computing mode, it is necessary to solve the task migration problem when users move between edge servers, especially for tasks with high data volume, such as unmanned driving and real-time monitoring. These tasks require real-time image processing and recognition, with huge data volume and high latency requirements. For example, for unmanned vehicles, the road conditions are complex and the safety requirements are high. It is necessary to ensure "uninterrupted" connection to ensure the safe driving of unmanned vehicles. If the task migration operation between servers is performed after the unmanned vehicle is disconnected from the edge server, it is impossible to achieve "seamless" connection in the task execution process, which affects the speed of the unmanned vehicle's judgment of road conditions and causes accidents. Therefore, seamless migration has become a hot topic in academia and industry.

[0003] The concept of seamless migration originates from the concept of seamless switching in cellular networks, that is, without affecting voice calls and data transmission, mobile users can move freely from one geographical location to another, and there can be multiple cellular base stations with limited coverage between the two locations. The difference is that when switching cellular networks, communication is switched, and the amount of data transmitted in this process is very small; while the focus of task migration is how to transfer programs, data and execution status from one edge cloud to another edge cloud, and the amount of data transmitted in this process is relatively large. The concept of task migration in mobile edge computing is somewhat similar to the concept of real-time migration of virtual machines in cloud data centers, but there is an essential difference between the two. Task migration needs to endure the uncertainty brought by bandwidth and computing power, and these uncertainties are often related to the dynamic nature of user groups, that is, the unpredictability of demand. Summary of the invention

[0004] The purpose of the present invention is to provide a task pre-migration strategy in a mobile edge computing environment, which is used to solve the technical problem in the prior art that when seamless switching between edge servers is performed, the amount of data transmitted is relatively large, and the dynamic nature of the user group, that is, the demand is difficult to predict, so the seamless migration is affected by the communication bandwidth and computing power, resulting in uncertainty.

[0005] The task pre-migration strategy in the mobile edge computing environment includes the following steps:

[0006] Step 1: Establish a direction judgment model. Establish a hexagonal honeycomb structure with the base station corresponding to the edge server as the center. The coverage of the edge server forms the same as the coverage of the communication network, and obtains a user movement direction prediction model; the base station corresponding to the current edge server is evenly divided into 6 triangular areas around the center. Only one side of the triangular area is connected to the adjacent new edge server, so there is a corresponding relationship between the triangular area and the new edge server.

[0007] Step 2: Determine the relative position between the user and the base station of the current edge server, so as to predict the new edge server that may need to receive task migration after the user moves; determine which triangular area the user is in based on the relative position, and the new edge server corresponding to the triangular area is the new edge server that may need to receive task migration.

[0008] Step 3: Perform relevant pre-migration operations on the new edge server determined in the previous step.

[0009] Preferably, in step 2, the vector User x，y and BS x，y Represents the plane position of the user and the base station, where User y Indicates the user's y coordinate, BS y represents the y coordinate of the base station, then User x Indicates the user's x coordinate, BS x represents the x-coordinate of the base station, calculates the vector slope k from the base station to the user, and locates the user in the direction determination model to verify whether the user is still within the coverage of the current edge server, where Since the shape of the triangular area is an equilateral triangle, the angle between the two sides of the triangular area is 60°. The triangular area is divided according to the slopes of the sides connecting the triangular area with the base station at the center to determine which triangular area the user is located in.

[0010] Preferably, the triangular area starts from the side with a slope of 0 and located on the positive side of the x-axis, and the triangular areas are numbered 1-6 in a counterclockwise direction, then: And User y ≥BS y , the user's position is 1; if And User y ≥BS y , then the user's position is 2. By analogy, the determination conditions for the triangular areas in the other four directions are obtained: if And User y ≥BS y , the user's location is 3; if And User y≤BS y , the user's position is 4; if And User y ≤BS y , the user's location is 5; if And User y ≤BS y , then the user's location is 6, and the above method is used to determine which triangle area the user's location is in.

[0011] Preferably, the relevant pre-migration operations are as follows: during the user uploading data stage, the current edge server will synchronize the received data to the new edge server; when the current edge server performs calculations, the result data will be sent to the new edge server to synchronously replace the data in the new edge server; when the current edge server returns the result data, it will delete the returned data and send a signal to the new edge server to synchronously delete the corresponding result data on the new edge server.

[0012] Preferably, the task pre-migration strategy also includes:

[0013] Step 4: When the user moves to the new triangular area, the user is located and the coverage of the edge server where the user is located is determined, so as to determine whether the task migration to the new edge server is completed.

[0014] Preferably, if the new triangular area is still within the coverage of the original edge server, data migration does not occur, and steps two to three are repeated to re-determine the new edge server that needs to receive task migration and perform task pre-migration on it; if the new triangular area is located in the coverage of the new edge server that previously accepted task pre-migration, the new edge server is exchanged with the edge server that previously performed data transmission and calculation, and calculation is performed on the new edge server.

[0015] The present invention has the following advantages: the scheme uses the honeycomb structure of the communication network to establish a corresponding judgment direction model, and uses the model and the user location to predict a new edge server where data migration may occur in a short period of time. When the communication interval is very short, the reliability of the prediction of the new edge server can be basically maintained, and the dynamics of the user group are less affected. In addition, the prediction model of this method is simple, the calculation amount is small, and the pre-migration object can be quickly determined.

[0016] The amount of data transmitted each time during the pre-migration process is limited to the corresponding operations and transmissions of a limited number of data. The uploaded data and result data will be synchronized to the new edge server over time, and the operation of returning the result data only sends an operation signal for the new edge server to delete it synchronously. In this way, the amount of data transmitted in a short period of time is greatly reduced, and the new edge server itself does not need to calculate the received data. Therefore, the pre-migration process of this task will basically not be limited by bandwidth and will not affect the computing power of the new edge server. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a judgment direction model for the task pre-migration strategy in the mobile edge computing environment of the present invention.

[0018] Figure 2 A schematic diagram of determining the relative position of a user in the present invention and corresponding to a triangular area.

[0019] Figure 3 It is a schematic diagram of the pre-migration operation performed in the data uploading phase of the present invention.

[0020] Figure 4 It is a schematic diagram of the pre-migration operation performed in the server computing phase in the present invention.

[0021] Figure 5 It is a schematic diagram of the pre-migration operation performed in the data return phase of the present invention.

[0022] Figure 6 It is a curve comparison diagram of the total energy consumption of the method of the present invention and other methods under normal circumstances.

[0023] Figure 7 This is a curve comparison diagram of the total delay of the method of the present invention and other methods under normal circumstances.

[0024] Figure 8 This is a curve comparison diagram of total energy consumption of the method of the present invention and other methods when the server service capacity is small.

[0025] Fig. 9 This is a curve comparison diagram of the total delay between the method of the present invention and other methods when the server service capacity is small.

[0026] Fig.10 This is a curve comparison diagram of total energy consumption between the method of the present invention and other methods when the local device computing power is large.

[0027] Fig.11 This is a curve comparison diagram of the total delay between the method of the present invention and other methods when the local device has a larger computing power. DETAILED DESCRIPTION

[0028] The specific implementation modes of the present invention will be further explained in detail below by describing the embodiments with reference to the accompanying drawings, so as to help those skilled in the art to have a more complete, accurate and in-depth understanding of the inventive concept and technical solution of the present invention.

[0029] like Figure 1-11 As shown, the present invention provides a task pre-migration strategy in a mobile edge computing environment, comprising the following steps:

[0030] Step 1: Establish a direction determination model.

[0031] In order to pre-migrate tasks, we first need to know the user's movement direction. We use a simple method to predict the user's movement direction. Referring to the cellular structure of the current communication network, a hexagonal cellular structure is established with the base station of the corresponding edge server as the center. The coverage of the edge server forms the same coverage as the communication network, and the user movement direction prediction model is obtained. In this model, the coverage of the edge server is divided into a finite number of directions. The division method of the corresponding hexagonal cellular structure is as follows: Figure 1 As shown in the figure, the coverage of edge server A is a hexagonal structure, which can be evenly divided into 6 triangular areas around the base station corresponding to edge server A. Figure 1 Take the area numbered 1 in the figure as an example. Its three sides represent three possible moving directions. That is, if a user in area numbered 1 moves away from the area, there are basically only three possible moving directions corresponding to the three sides of the triangular area. In two directions, the user will move to the adjacent triangular area within the coverage of edge server A. In these two cases, task migration is not required. In the third direction, the user will move to the coverage of the adjacent edge server B in the corresponding direction. Therefore, in this model, as long as the user is in which triangular area within the coverage of the edge server, the coverage of the next possible server that the user may enter after moving can be determined.

[0032] Step 2: Determine the relative position between the user and the base station of the current edge server, so as to predict the new edge server that may need to receive the task migration after the user moves.

[0033] The plane coordinates of the user are determined based on the user's positioning information. The plane coordinates of each base station are known. The relative positions of the two are determined based on the plane coordinates of the base station corresponding to the edge server with which the user currently has a data transmission relationship. Figure 2 As shown: using vector User x，y and BS x，y Represents the plane position of the user and the base station, where x and y represent the horizontal and vertical coordinates, then User y Indicates the user's y coordinate, BS yrepresents the y coordinate of the base station, then User x Indicates the user's x coordinate, BS x Represents the x-coordinate of the base station. Calculate the vector slope k from the base station to the user, and locate the user in the direction determination model to verify whether the user is still within the coverage of the current edge server. Since the shape of the triangular area is an equilateral triangle, the angle between the two sides of the triangular area is 60°. And User y ≥BS y , the user's position is 1; if ( or ) and User y ≥BS y , then the user's position is 2. By analogy, the determination conditions for the triangular areas in the other four directions are obtained: if And User y ≥BS y , the user's location is 3; if And User y ≤BS y , the user's location is 4; if ( or ) and User y ≤BS y , the user's location is 5; if And User y ≤BS y , then the user's location is 6. Figure 1 As shown, if it is confirmed that the relative position between the user and the base station of the corresponding edge server is located in the triangular area numbered 1, a new edge server, namely edge server B, for task migration may be required.

[0034] Step 3: Perform relevant pre-migration operations on the new edge server determined in the previous step.

[0035] After the relative position between the user and the base station of the current edge server is determined, and before the user enters the coverage area of ​​the new edge server, the relevant pre-migration operations are as follows. Figure 3 As shown in , during the user uploading data stage, edge server A will synchronize the received data to edge server B, where di (i=1, 2...) represents the i-th data uploaded by the user. Figure 4 As shown in FIG. 1 , when edge server A performs calculations, the result data will be sent to edge server B to synchronously replace the data in edge server B, where Ri (i=1, 2…) represents the result calculated by the edge server based on the i-th data, and Ri will replace di in edge server A and edge server B. Figure 5 As shown, when edge server A returns result data Ri, it will delete the returned data Ri, and send a signal to edge server B to synchronously delete the corresponding result data Ri on edge server B. The three stages corresponding to different data can be performed synchronously, that is, when edge server A obtains data di, it starts to calculate, and when it obtains the calculation result Ri, it starts to return the result data, which does not hinder the upload, calculation and return of other data.

[0036] The amount of data transmitted each time during the above pre-migration process is limited to the corresponding operations and transmissions of a limited number of data. The uploaded data and result data will be synchronized to the new edge server over time, and the operation of returning the result data only sends an operation signal for the new edge server to delete it synchronously. In this way, the amount of data transmitted in a short period of time is greatly reduced, and the new edge server itself does not need to calculate the received data. Therefore, the task pre-migration process will basically not be limited by bandwidth and will not affect the computing power of the new edge server.

[0037] Step 4: When the user moves to the new triangular area, the user is located and the coverage of the edge server where the user is located is determined, so as to determine whether the task migration to the new edge server is completed.

[0038] If the new triangular area is still within the coverage of the original edge server, no data migration occurs, and steps 2 to 3 are repeated to re-determine the new edge server that needs to receive the task migration and perform task pre-migration on it; if the new triangular area is located in the coverage of the new edge server that previously received task pre-migration, the new edge server is exchanged with the edge server that previously performed data transmission and calculation, and the calculation is performed on the new edge server, while the task pre-migration is performed on the edge server that previously performed calculation. This is because when the task pre-migration is implemented according to this method, the triangular area where the user is located after the movement must be adjacent to the triangular area where the user is located before the movement. Therefore, the identity of the edge server that originally performed calculation is swapped with the new edge server, and becomes the only edge server that can allow the user to immediately enter the coverage of the new triangular area before the next confirmation of the moving direction. Since the task pre-migration causes the data in the two edge servers that perform calculations successively to be synchronized, it is only necessary to hand over the calculation process to the new edge server to complete the seamless switching of the task between the two edge servers, and because it is not easily affected by bandwidth and computing power, the seamless switching is stable and reliable.

[0039] Figures 6 to 11Based on actual experiments, various cloud computing methods are compared in various aspects when moving between edge servers. The methods involved in the above figure include: the method disclosed in this application, that is, the method of pre-migration of tasks using a judgment direction model (Pre-migration Offloading, PMO); the method of ordinary computing offloading (WithoutOffloading, WO); the method of computing offloading without a migration mechanism (Without-migration Offloading, WOMO), and the method of computing offloading with a migration mechanism (With-migration Offloading, WMO). It can be seen from the figure that compared with other methods, after adopting the method provided by this application, the energy consumption and latency of the edge server in various situations are relatively good (only when the computing power of the local device is large, because most of the computing power comes from the local device, the method of ordinary computing offloading is better than other methods. This situation is not the main case of application of this method).

[0040] The present invention is described above by way of example in conjunction with the accompanying drawings. It is obvious that the specific implementation of the present invention is not limited to the above-mentioned method. As long as various non-substantial improvements are made using the inventive concept and technical solution of the present invention, or the inventive concept and technical solution are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

Claims

1. Task pre-migration strategy in mobile edge computing environment, characterized by: The following steps are involved: Step 1: Establish a judgment direction model, establish a hexagonal cellular structure with the base station corresponding to the edge server as the center, and the coverage of the edge server forms the same as the coverage of the communication network, and obtain a user movement direction prediction model; The base station corresponding to the current edge server is evenly divided into 6 triangular areas in the circumference, and only one side of the triangular area is connected to the adjacent new edge server, so there is a corresponding relationship between the triangular area and the new edge server; Step 2: Determine the relative position between the user and the base station of the current edge server, so as to predict the new edge server that may need to receive the task migration after the user moves; Determine which triangular area the user is in according to the relative position, and the new edge server corresponding to the triangular area may be the new edge server that needs to receive task migration; Step 3: Perform relevant pre-migration operations on the new edge server determined in the previous step; In the step 2, the vector User x,y and BS x,y Represents the plane position of the user and the base station, where User y Indicates the user's y coordinate, BS y represents the y coordinate of the base station, then User x Indicates the user's x coordinate, BS x represents the x-coordinate of the base station, calculates the vector slope k from the base station to the user, and locates the user in the direction determination model to verify whether the user is still within the coverage of the current edge server, where Since the shape of the triangular area is an equilateral triangle, the angle between the two sides of the triangular area is 60°. The triangular area is divided according to the slopes of the sides connecting the triangular area and the base station at the center to determine which triangular area the user is located in. The triangular area starts from the side with a slope of 0 and located on the positive side of the x-axis, and the triangular areas are numbered 1-6 in a counterclockwise direction. Then: If And User y ≥BS y , the user's position is 1; if or At the same time User y ≥BS y , the user's position is 2; and so on, the determination conditions for the triangular areas in the other four directions are obtained: if And User y ≥BS y , the user's location is 3; if And User y ≤BS y , the user's location is 4; if or At the same time User y ≤BS y , the user's location is 5; if And user y ≤BS y , then the user's location is 6, and the above method is used to determine which triangle area the user's location is in; The relevant pre-migration operations are as follows: during the user data upload stage, the current edge server will synchronize the received data to the new edge server; when the current edge server performs calculations, the result data will be sent to the new edge server to synchronously replace the data in the new edge server; when the current edge server returns the result data, it will delete the returned data and send a signal to the new edge server to synchronously delete the corresponding result data on the new edge server.

2. The task pre-migration strategy in the mobile edge computing environment according to claim 1 is characterized in that: Also includes: Step 4: When the user moves to the new triangular area, the user is located and the coverage of the edge server where the user is located is determined, so as to determine whether the task migration to the new edge server is completed.

3. The task pre-migration strategy in the mobile edge computing environment according to claim 2 is characterized in that: If the new triangular area is still within the coverage of the original edge server, data migration will not occur, and steps 2 to 3 will be repeated to re-determine the new edge server that needs to receive task migration and perform task pre-migration on it; if the new triangular area is located in the coverage of the new edge server that previously accepted task pre-migration, the new edge server will be exchanged with the edge server that previously performed data transmission and calculation, and calculation will be performed on the new edge server.

Images