Position scheduling method for mobile computing resources in edge environment

Abstract

The invention provides a location scheduling method for mobile computing resources in an edge environment. The method comprises the following steps: firstly, constructing a communication system architecture consisting of a user side and a mobile edge server; modeling a communication system architecture into a communication model, establishing a variable-size packing problem with geographical constraints for location scheduling of mobile computing resources in an edge environment, and establishing constraint conditions and a target function; an online scheduling scheme is designed for the size-variable packing problem with geographical constraints, and the specific process is as follows: a mobile edge server actively synchronizes user setting information with other mobile servers, and determines the potential staying position of the mobile edge server; geographic constraints are considered, a quadruped robot stay position decision algorithm based on a greedy algorithm is designed, hot spot areas are combined, and tasks are distributed to an edge server; afterwards, the mobile edge server arrives at the staying position and starts to process the unloading task, and finally the staying position of deployment is determined.

Description

Technical Field

[0001] The present disclosure relates to the field of computers, and more particularly, to a location scheduling method for mobile computing resources in an edge environment. Background Art

[0002] In recent years, the Internet of Things (IoT) has flourished, providing real-time analytics services for projects such as intelligent transportation, smart cities, large-scale gaming, and video processing. This has led to exponential growth in the scale of edge devices and the demand for data processing. The number of connected devices worldwide has exceeded 10 billion, and this expansion has led to a dramatic increase in data generation. Currently, the amount of edge data generated globally exceeds 100 exabytes (EB) daily. It is estimated that by 2030, over 70% of this data will be processed at edge nodes outside of traditional cloud computing centers.

[0003] To support a large number of devices and process massive amounts of data, Mobile Edge Computing (MEC) enables latency-sensitive applications by deploying edge computing resources close to mobile user equipment (UE). In contrast, while traditional cloud computing architectures offer powerful computing and storage capabilities, their centralized nature leads to significant data transmission latency. This latency issue is particularly acute when faced with highly dynamic and mobile user demands. Consequently, edge computing, as a new computing paradigm, has garnered widespread attention and research.

[0004] Edge computing deploys computing resources at the edge of the network, closer to data sources and users, significantly reducing latency and improving system responsiveness and real-time performance. Many current research approaches, such as task offloading, caching, and resource allocation, assume that edge servers are pre-deployed. Existing edge server deployment schemes show that concentrating more edge servers in hotspots outperforms evenly distributed approaches. However, due to user mobility and dynamic demand, hotspots can experience a sudden drop in user density within a short period of time. To address population mobility, existing research has proposed methods that periodically map user clusters to fixed edge servers to reduce infrastructure costs. However, due to the uneven distribution of tasks, some fixed-location edge servers inevitably become overloaded, while servers in other locations experience low resource utilization. Therefore, techniques such as task migration are needed to balance workloads across edge servers. This, in turn, incurs additional communication and signaling overhead, increasing task latency as tasks need to be transferred between servers.

[0005] In terms of research content, most existing work focuses on trajectory planning for mobile edge servers. During the planning process, the user device's location remains fixed, while the server continuously moves between several fixed user devices. While this mobility approach is simple, it has limited effectiveness in improving server utilization and is not suitable for situations where personnel mobility is unstable. Regarding edge server deployment, most existing research focuses on deploying edge servers on drones, which have gained widespread recognition as a viable solution for assisting wireless communication networks. However, drone deployment still has many drawbacks, such as susceptibility to weather conditions, exposure to high-altitude signal interference, and poor adaptability in complex terrain (such as mountains and jungles). Existing work is limited by the following: first, the mobility of user devices leads to uneven dynamic distribution of tasks; second, the network's scalability is insufficient. When a large number of user devices offload tasks, the server's trajectory is affected by each user device, making adjustments time-consuming and costly. Given the inefficiency of fixed edge server deployment and the limitations of current mobile server trajectory planning, it is worthwhile to study how to schedule edge servers and appropriately locate them in time-varying hotspots, and how to associate mobile user devices with edge servers. Summary of the Invention

[0006] The purpose of the embodiments disclosed herein is to provide a method for location scheduling of mobile computing resources in an edge environment. This invention models edge server scheduling as an NP-hard problem of variable-size bin packing with geographic constraints and proposes an online mobile edge server scheduling scheme that sequentially determines the hovering locations of mobile edge servers. As the communication radius increases, hotspot areas are identified based on task intensity. This invention overcomes the problems of low resource utilization and uneven service distribution in current server deployment methods.

[0007] In a general aspect, a location scheduling method for mobile computing resources in an edge environment is provided. First, a communication system architecture consisting of a user terminal and a mobile edge server is constructed. The set of mobile edge servers is represented as N, the set of user tasks is represented as M, the mobile edge server is represented as n, and the user task is represented as m.

[0008] The communication system architecture is then modeled as a communication model, and a variable-size bin packing problem with geographical constraints is formulated for the location scheduling of mobile computing resources in edge environments, with constraints and objective functions established;

[0009] An online scheduling scheme is designed for the variable-size packing problem with geographical constraints. The specific process is as follows: the mobile edge server actively synchronizes user device information with other mobile servers and determines its potential resting location; considering the geographical constraints, a greedy algorithm is designed for the quadruped robot's resting location decision, which merges hot spots and assigns tasks to edge servers; then, the mobile edge server arrives at its resting location and starts processing the offloaded tasks, and finally determines the deployed residence location.

[0010] The mobile edge server is mounted on a quadruped robot.

[0011] The specific construction method of the communication model is: using (u m ,v m ) and (u n ,v n ) represent the coordinates of task m and mobile edge server n, respectively. The distance between the server and task m is as follows:

[0012]

[0013] The path loss of the channel is modeled as:

[0014]

[0015] in, It is the path loss index obtained by comprehensively considering the line-of-sight and non-line-of-sight channels, and PL(d0) is the path loss at the reference distance d0.

[0016] The data transfer rate can be derived from the following formula:

[0017]

[0018] Where B0 is the channel bandwidth, P TX is the transmission power of the user equipment, N0 is the noise power, h mn is the total channel gain, which includes the effects of path loss, shadowing, and multipath fast fading;

[0019] Let x mn Indicates whether task m is offloaded to edge server n, x mn =1 means yes, x mn =No; due to the limited computing power of each edge server, if task m is offloaded to edge server n, the task delay By communication time and calculation delay composition:

[0020]

[0021] in is the communication delay, B(d mn ) is the distance d mn Communication efficiency when

[0022] Further extended to the case where the task size is variable, the variable-size task can be evenly divided into small tasks; is the queuing delay of task offloading, where ω = P + P' is the workload P of task m plus the current workload P' of the assigned edge server n, Q E It is the CPU processing capacity of the edge server. Each edge server sets an upper limit on the number of pods. If the number of offloaded tasks exceeds the number of pods, the tasks will be queued and executed on a first-come, first-served basis.

[0023] The objective function is:

[0024]

[0025] The goal is to maximize in is the collection of all tasks, is the set of all mobile edge servers, x mn Indicates whether task m is assigned to server n.

[0026] The constraints are specifically:

[0027] x mn ∈{0,1}, That is, for any m in set M, any n in set N, x mn It can only take the value 0 or 1, which clarifies the binary nature of the assignment relationship between tasks and servers. That is, a task is either assigned to a server or not assigned.

[0028] r min ≤r n ≤r max , That is, for the set Each server n in the communication range r n Must be within the minimum communication range r min and the maximum communication range r max between;

[0029] j is for the set Each server n, S n Defined as the set of all tasks assigned to server n, when x mn = 1, task m belongs to the set S n ;

[0030] That is, when the number of tasks assigned to server n |S n When | is greater than 0, the function value is 1, otherwise it is 0. The entire formula means that the number of servers assigned tasks cannot exceed φ;

[0031] Used to constrain the delay of task processing, where represents the communication delay, A is the task data volume, B(d mn ) is the distance d between server n and task m mn The relevant data transmission speed, Indicates computation delay, |S n | is the number of tasks assigned to server n, P is the computational workload of a single task, Q E is the task processing capability of the server, is a predefined task deadline;

[0032] d mn ≤r n , That is, for the set Each server n and its assigned task set S n For each task m in the server, the distance d between task m and server n is mn Must be less than or equal to the communication range r of server n n ;

[0033] That is, for any task m in the set, the sum of the number of servers it is assigned to is at most 1. A task can only be processed by one server at a time, avoiding repeated or chaotic assignment of tasks.

[0034] The specific implementation of the quadruped robot's stop position decision algorithm is as follows: first, hotspot areas are merged, user devices are aggregated in sequence, and mobile edge servers are allocated. The service area is discretized into J grids with a discretization radius of j;

[0035] In each time slot, the task intensity of grid j∈J is expressed as follows according to the Euclidean distance:

[0036]

[0037] where ε mj ∈{0,1} indicates whether the distance between task m and the center of grid j is within radius r. When the capacity of a grid reaches its upper limit, a mobile edge server is deployed at the center of the grid, and the tasks that have been served are removed. If the upper limit has not been reached, the radius r is increased and the above verification process is repeated. The heat is updated each time a server is deployed or r changes. The estimated value of C(r) is:

[0038]

[0039] Where C(r) represents the number of tasks completed by a single mobile edge server within the communication range r. If r reaches the maximum communication range, servers will be deployed to these grids in the order of decreasing task intensity, regardless of whether the task intensity of the current grid exceeds the server capacity;

[0040] Repeat the above steps until the number of servers reaches the initial server budget or all tasks are processed. The y operation is terminated when the number of remaining tasks is lower than the threshold θ·C(r).

[0041] The technical effects to be achieved by the embodiments of the present invention are:

[0042] Achieve higher resource utilization of the server and maximize the number of service tasks under the condition of limited server resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The above and other objects and features of the present disclosure will become more apparent from the following description in conjunction with the accompanying drawings.

[0044] Figure 1 Schematic diagram showing edge server placement according to the background technology of the present disclosure

[0045] Figure 2 is a core method diagram illustrating a method for location scheduling of mobile computing resources in an edge environment according to an embodiment of the present disclosure;

[0046] Figure 3 is a schematic diagram illustrating expansion of server coverage radius according to an embodiment of the present disclosure;

[0047] Figure 4 1 is a diagram showing the number of service tasks in different deployment schemes according to an embodiment of the present disclosure;

[0048] Figure 5 Graphs show service task percentages for different deployment scenarios according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0049] The following detailed description is provided to help the reader gain a comprehensive understanding of the methods, devices and / or systems described herein. However, various changes, modifications and equivalents of the methods, devices and / or systems described herein will be clear after understanding the disclosure of the present application. For example, the order of operations described herein is merely an example and is not limited to those orders set forth herein, but can be changed as will be clear after understanding the disclosure of the present application, except for operations that must occur in a specific order. In addition, for greater clarity and conciseness, descriptions of features known in the art may be omitted.

[0050] The features described herein can be implemented in different forms and should not be construed as limited to the examples described herein. Rather, the examples described herein are provided to illustrate only some of the many possible ways to implement the methods, devices, and / or systems described herein, which will become clear after understanding the disclosure of this application.

[0051] As used herein, the term "and / or" includes any one of the associated listed items and any combination of any two or more.

[0052] Although terms such as "first," "second," and "third" may be used herein to describe various members, components, regions, layers, or portions, these members, components, regions, layers, or portions should not be limited by these terms. Instead, these terms are used solely to distinguish one member, component, region, layer, or portion from another member, component, region, layer, or portion. Thus, what is referred to as a first member, first component, first region, first layer, or first portion in the examples described herein may also be referred to as a second member, second component, second region, second layer, or second portion without departing from the teachings of the examples.

[0053] In the specification, when an element (such as a layer, region, or substrate) is described as being “on,” “connected to,” or “coupled to” another element, the element may be directly “on,” “connected to,” or “coupled to” the other element, or one or more other elements may be present therebetween. Conversely, when an element is described as being “directly on,” “directly connected to,” or “directly coupled to” another element, there may be no other elements present therebetween.

[0054] The terms used herein are intended only to describe various examples and are not intended to limit the disclosure. Unless the context clearly indicates otherwise, the singular is intended to include the plural. The terms "comprise," "include," and "have" indicate the presence of the recited features, quantities, operations, components, elements, and / or combinations thereof, but do not preclude the presence or addition of one or more other features, quantities, operations, components, elements, and / or combinations thereof.

[0055] Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which the present disclosure pertains after understanding the present disclosure. Unless expressly defined otherwise herein, terms (such as those defined in general dictionaries) should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and the present disclosure, and should not be interpreted in an idealized or overly formal manner.

[0056] Furthermore, in describing the examples, when it is deemed that a detailed description of well-known related structures or functions would cause ambiguous interpretation of the present disclosure, such detailed description will be omitted.

[0057] Figure 2 3 is a core method diagram illustrating a location scheduling method for mobile computing resources in an edge environment according to an embodiment of the present disclosure.

[0058] like Figure 1 As shown in the figure, most fixed-location base stations are distributed around the theme park. In mobile edge computing, deploying edge servers at every base station would lead to an imbalanced workload between base stations and reduce the number of user devices served. When user devices offload tasks to edge servers, some base stations near the entrance can be fully utilized, but other base stations along the roads outside the park are underloaded, meaning their computing resources are wasted. Even with a better server deployment solution, fixed deployment solutions still face the problem of inefficient computing resource utilization in the long run. In contrast, mobile edge servers can be dispatched to hotspots to handle sudden requests. During periods of low request volume, some robots can continue to work while the remaining robots return to the warehouse for maintenance or other work.

[0059] This solution aims to effectively and efficiently dispatch mobile edge servers. Mobile edge servers can be deployed closer to people, providing services to as many user devices as possible while also improving server utilization through high flexibility and adaptability to complex terrain.

[0060] Table 1 Problem scenario modeling variables

[0061]

[0062]

[0063] The communication model is as follows: Let (u m ,v m ) and (u n ,v n ) represent the coordinates of task m and mobile edge server n respectively, then the distance between the server and task m is as follows:

[0064]

[0065] The path loss of the channel can be modeled as:

[0066]

[0067] in, It is the path loss index obtained by comprehensively considering the line-of-sight (LOS) and non-line-of-sight (NLOS) channels, and PL(d0) is the path loss at the reference distance d0.

[0068] The data transfer rate can be derived from the following formula:

[0069]

[0070] Where B0 is the channel bandwidth, P TX is the transmission power of the user equipment, N0 is the noise power, h mn is the total channel gain, which includes the effects of path loss, shadowing, and multipath fast fading.

[0071] The calculation model is as follows: we can set x mn Indicates whether task m is offloaded to edge server n (x mn =1 means yes, x mn =No). Since the computing power of each edge server is limited, if task m is offloaded to edge server n, the task delay By communication time (i.e. the time between sending a task and receiving the result) and computational latency Composition can be expressed as:

[0072]

[0073] in is the communication delay, B(d mn ) is the distance d mn Assuming that each task has the same task size A, this model can be extended to the case where the task size is variable and these tasks can be evenly divided into small tasks. is the queuing delay of task offloading, where ω = P + P' is the workload P (in CPU cycles) of task m plus the current workload P' of the assigned edge server n, Q E The CPU processing power of the edge server (in CPU cycles / second). Each edge server has an upper limit on the number of pods it can hold. If the number of offloaded tasks exceeds the number of pods, the tasks will be queued and executed on a first-come, first-served (FCFS) basis.

[0074] Due to the uneven geographical distribution of the crowd and task requests, this solution considers the flexible deployment of mobile edge servers to provide task offloading services in time and space. The location of the mobile edge servers needs to be coordinated with the ever-changing task requests. Specifically, the edge servers equipped with robots should be scheduled to provide services to dense crowds with more requests. The goal of this solution is to serve as many tasks as possible within a given task deadline. If all tasks are assigned to a single server, although the server facility cost is the lowest, due to the limited computing power of the server, the server cannot complete all tasks within the deadline. If this solution inefficiently assigns tasks to multiple servers, the deployed computing resources (number of servers) will be wasted. This characteristic (tasks and robot-equipped servers) makes it possible to formulate it as a bin packing problem.

[0075] The bin packing problem, a classic optimization problem, has been widely studied. Its core objective is to rationally distribute a set of items into a set of bins. In traditional bin packing problems, the goal is typically to minimize the number of bins required given a fixed number of items, or to maximize a predetermined profit given a fixed number of bins. In server-based scheduling, mobile servers are scheduled based on a distribution of tasks, with profit defined as the number of tasks successfully served within their deadlines. The goal of this solution is to determine the location of mobile edge servers and their corresponding communication ranges to maximize the number of tasks that can be served.

[0076] In order to maximize the number of tasks served in each time slot, the variable-sized bin packing problem with geographical constraints (Geo-VBP) can be formulated. The specific constraints are as follows:

[0077] Objective function:

[0078]

[0079] This function represents the solution's goal to maximize the value of the mobile edge server's coordinates and communication range. in is the collection of all tasks, is the set of all mobile edge servers, x mn Indicates whether task m is assigned to server n.

[0080] The constraints are as follows:

[0081]

[0082] Formula (3.6) indicates that for any m in set M and any n in set N, x mn It can only take 0 or 1, which clarifies the duality of the distribution relationship between tasks and servers, that is, a task is either assigned to a server or not assigned. Formula (3.7) means that for the set Each server n in the communication range r n Must be within the minimum communication range r min and the maximum communication range r max This is a reasonable limitation on the communication range of the server based on the hardware capabilities of the server and the actual application scenario. Formula (3.8) shows that for each server n in the set N, S n is defined as the set of all tasks assigned to server n, that is, when x mn = 1, task m belongs to the set S n This formula is used to clarify the tasks that each server is responsible for processing. Formula (3.9) is an indicator function. When the number of tasks assigned to server n |S n When | is greater than 0, the function value is 1, otherwise it is 0. The entire formula indicates that the number of servers assigned tasks cannot exceed φ, which represents the maximum number of mobile edge servers that can be scheduled, and limits the number of servers used from the perspective of resource limitations. Formula (3.10) is used to constrain the delay of task processing. Where A / B(d mn ),Right now represents the communication delay (A is the amount of task data, B(d mn ) is the distance d between server n and task m mn associated data transfer rates), Indicates the calculation delay (|S n | is the number of tasks assigned to server n, P is the computational workload of a single task, Q E is the task processing capability of the server), Is the predefined task deadline to ensure that the task is completed within the specified time. Formula (3.11) represents that for each server n in the set N and its assigned task set S n For each task m in the server, the distance d between task m and server n is mn Must be less than or equal to the communication range r of server n n, ensuring that the server can effectively communicate with the assigned tasks. Formula (3.12) indicates that for any task m in the set, the sum of the number of servers assigned to it is at most 1. In other words, a task can only be processed by one server at a time, avoiding repeated or chaotic task allocation. To simplify modeling, this solution stipulates that only one task can be processed simultaneously in a pod. When the number of pods in a server reaches the upper limit, it means that the server has reached the upper limit of its task parallel processing capacity, and new tasks need to be queued for processing.

[0083] The bin packing problem has been proven to be NP-hard. Several well-known one-dimensional bin packing algorithms exist, but these algorithms do not consider geographic constraints or similar factors. For example, in the Next-Fit Decreasing algorithm, items are rearranged in non-increasing order based on their size. If the total size of the assigned items does not exceed the bin capacity, the items are packed into the current bin; otherwise, a new bin is opened. However, in the mobile edge server scheduling problem, geographic constraints must be considered for tasks assigned to the same edge server. Tasks covered by the same edge server must be located within a circular area; that is, the distance between any two tasks must not exceed the diameter of the communication coverage area. On the other hand, mobile edge servers (carried by quadruped robots) can adjust their coverage area to meet task requirements in hotspots or sparsely populated areas. The greater the distance between tasks and servers, the greater the communication delay due to lower data transmission rates. Existing location-based clustering algorithms are not applicable because both server capacity and communication range need to be considered. To solve the geographic variable-size bin packing (Geo-VBP) problem, this paper proposes an online scheduling scheme to determine the station locations of deployed quadruped robots.

[0084]

[0085] As shown in Algorithm 1, the mobile edge server (on the quadruped robot) proactively synchronizes user equipment (UE) information with other mobile servers and determines their potential resting locations. Considering the geographic constraints in Geo-VBP, Algorithm 2, the quadruped robot resting location decision algorithm, is designed as a greedy algorithm to consolidate hotspots and assign tasks to edge servers. Afterward, the mobile edge server (on the quadruped robot) arrives at its resting location and begins processing offloaded tasks.

[0086] To identify and track dynamic hotspots, researchers have explored various approaches, such as leveraging machine learning techniques and hotspot monitoring through wireless sensors. However, these approaches often face two major challenges: they either require significant computational resources or incur high implementation costs. Furthermore, existing algorithms do not fully consider practical constraints in terms of communication and computation. In generating a heat map, tasks are merged or clustered under different influence radii, and the task intensities are graphically displayed. As the influence radius of a task increases, the task intensities in the overlapping area accumulate. This means that if a server is deployed in the overlapping area and the corresponding influence radius is used as its communication range, all tasks that affect the location of that server can be covered by this server. This situation is similar to the geographic constraints in the geographic variable-size bin packing (Geo-VBP) problem.

[0087]

[0088]

[0089] As shown in Algorithm 2, the proposed quadruped robot stay location decision (HOLD) scheme is based on merging hotspots, sequentially integrating and aggregating user devices, and assigning them to mobile edge servers (carried by the quadruped robot). The service area is discretized into J grids with a discretization radius of j.

[0090] In each time slot, the task intensity of grid j∈J can be expressed as follows according to the Euclidean distance:

[0091]

[0092] where ε mj ∈{0,1} indicates whether the distance between task m and the center of grid j is within radius r. (1 means within the radius, 0 means not within the radius). Once the capacity of a grid (defined as the maximum number of tasks that can be served) reaches the upper limit, a mobile edge server (equipped with a quadruped robot) can be deployed at the center of the grid, and the tasks that have been served will be removed. If the upper limit has not been reached, r will continue to increase, and the specific process is as follows Figure 3 As shown, the above verification process is repeated. Each time a server is deployed or r changes, the heat map will be updated. Since the last task processed by the server determines whether all assigned tasks can be completed on time, the estimated value of C(r) is:

[0093]

[0094] If r reaches the maximum communication range, servers will be deployed to these grids in the order of decreasing task intensity, regardless of whether the task intensity of the current grid exceeds the server capacity. Repeat the above steps until the number of servers reaches the initial server budget or all tasks are processed, such as Figure 4 To ensure server utilization, the algorithm will terminate when the number of remaining tasks is lower than the threshold θ·C(r).

[0095] This solution uses Δr to represent the increment of radius r.

[0096]

[0097] After iterations, the parking locations of the mobile servers will be determined one by one.

[0098] The above algorithm is a process of determining a set of mobile server placement locations within each determined time slot. This solution discretizes the process of server location movement, using a determined time step (e.g., five minutes) as the time slot interval.

[0099] Technical Effects

[0100] This solution verifies the effectiveness of the above-mentioned method through simulation testing.

[0101] Table 2 Hardware Configuration

[0102]

[0103]

[0104] Table 3 Software Configuration Table

[0105]

[0106] Figure 4The figure shows the number of service tasks under different deployment strategies at different time periods. The black dashed line in the figure represents the total number of tasks. It can be seen that the service volume is low from midnight to 8:00 AM, with approximately 100 task requests per period. The number of service requests gradually increases from 8:00 AM to 12:00 PM. The peak period for the entire day is from 12:00 PM to 6:00 PM, with the number of tasks remaining high. The number of tasks then gradually decreases, returning to a lower level at 10:00 PM. The blue curve represents the number of tasks that can be completed by the on-demand dynamic deployment strategy studied in this chapter. This solution uses uniform fixed deployment, on-demand fixed deployment, spiral deployment, and trajectory planning deployment as control groups, represented by the green solid line, red solid line, cyan dashed line, and purple dashed line, respectively. The figure shows that on-demand dynamic deployment completes more tasks in most time periods. The task completion trends of the other solutions are generally consistent with the total number of tasks. On-demand fixed deployment performs best, while the naive fixed uniform deployment performs worst. Spiral deployment and trajectory planning deployment have similar performance.

[0107] The peak period of tasks is from 12:00 to 18:00 in a day. Figure 5 As shown, the service task percentages of various deployment schemes during peak hours are relatively stable compared to other periods. On-demand dynamic deployment has an average service task percentage of 91.3%, while on-demand fixed deployment, which performs best in terms of service task quantity, only achieves an average service task percentage of 83.1%. On-demand dynamic deployment outperforms it by approximately 8.2%. When the number of servers is small, spiral deployment and trajectory planning deployment have similar performance and both perform poorly, with average service task percentages ranging from 77% to 79%. Uniform fixed deployment performs the worst. The service task percentages of various schemes during other periods vary significantly, as they are more susceptible to randomness when the number of tasks is small.

[0108] In summary, the present invention provides a mobile computing resource scheduling method based on greedy thinking, which aims to achieve higher resource utilization of the server by defining the task and resource model in the edge environment and proposing a series of edge server location scheduling algorithms, so as to maximize the number of service tasks under the condition of limited server resources.

[0109] While some embodiments of the present disclosure have been shown and described, it will be appreciated by those skilled in the art that changes may be made to these embodiments without departing from the principles and spirit of the disclosure, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for location scheduling of mobile computing resources in an edge environment, characterized in that: First, we build a communication system architecture consisting of user terminals and mobile edge servers. The set of mobile edge servers is represented as N, the set of user tasks is represented as M, the mobile edge server is represented as n, and the user task is represented as m. The communication system architecture is then modeled as a communication model, and a variable-size bin packing problem with geographical constraints is formulated for the location scheduling of mobile computing resources in edge environments, with constraints and objective functions established; An online scheduling scheme is designed for the variable-size packing problem with geographical constraints. The specific process is as follows: the mobile edge server actively synchronizes user device information with other mobile servers and determines its potential resting location; considering the geographical constraints, a greedy algorithm is designed for the quadruped robot's resting location decision, which merges hot spots and assigns tasks to edge servers; then, the mobile edge server arrives at its resting location and starts processing the offloaded tasks, and finally determines the deployed residence location.

2. The method for location scheduling of mobile computing resources in an edge environment according to claim 1, wherein: The mobile edge server is mounted on a quadruped robot.

3. The method for location scheduling of mobile computing resources in an edge environment according to claim 1, wherein: The specific construction method of the communication model is: using (u m ,v m ) and (u n ,v n ) represent the coordinates of task m and mobile edge server n, respectively. The distance between the server and task m is as follows: The path loss of the channel is modeled as: in, It is the path loss index obtained by comprehensively considering the line-of-sight and non-line-of-sight channels, and PL(d0) is the path loss at the reference distance d0.

4. The method for location scheduling of mobile computing resources in an edge environment according to claim 3, wherein: The data transfer rate can be derived from the following formula: Where B0 is the channel bandwidth, P TX is the transmission power of the user equipment, N0 is the noise power, h mn is the total channel gain including the effects of path loss, shadowing, and multipath fast fading; Let x mn Indicates whether task m is offloaded to edge server n, x mn =1 means yes, x mn =No; due to the limited computing power of each edge server, if task m is offloaded to edge server n, the task delay By communication time and calculation delay composition: in is the communication delay, B(d mn ) is the distance d mn Communication efficiency when Further extended to the case where the task size is variable, the variable-size task can be evenly divided into small tasks; is the queuing delay of task offloading, where ω = P + P' is the workload P of task m plus the current workload P' of the assigned edge server n, Q E It is the CPU processing capacity of the edge server. Each edge server sets an upper limit on the number of pods. If the number of offloaded tasks exceeds the number of pods, the tasks will be queued and executed on a first-come, first-served basis.

5. The method for location scheduling of mobile computing resources in an edge environment according to claim 4, wherein: The objective function is: The goal is to maximize in is the collection of all tasks, is the set of all mobile edge servers, x mn Indicates whether task m is assigned to server n.

6. The method for location scheduling of mobile computing resources in an edge environment according to claim 4, wherein: The constraints are specifically: That is, for any m in set M, any n in set N, x mn It can only take the value 0 or 1, which clarifies the binary nature of the assignment relationship between tasks and servers. That is, a task is either assigned to a server or not assigned. That is, for the set Each server n in the communication range r n Must be within the minimum communication range r min and the maximum communication range r max between; j is for the set Each server n, S n Defined as the set of all tasks assigned to server n, when x mn = 1, task m belongs to the set S n ; That is, when the number of tasks assigned to server n |S n When | is greater than 0, the function value is 1, otherwise it is 0. The entire formula means that the number of servers assigned tasks cannot exceed φ; Used to constrain the delay of task processing, where represents the communication delay, A is the task data volume, B(d mn ) is the distance d between server n and task m mn The relevant data transmission speed, Indicates computation delay, |S n | is the number of tasks assigned to server n, P is the computational workload of a single task, Q E is the task processing capability of the server, is a predefined task deadline; That is, for the set Each server n and its assigned task set S n For each task m in the server, the distance d between task m and server n is mn Must be less than or equal to the communication range r of server n n ; That is, for any task m in the set, the sum of the number of servers it is assigned to is at most 1. A task can only be processed by one server at a time, avoiding repeated or chaotic assignment of tasks.

7. The method for location scheduling of mobile computing resources in an edge environment according to claim 1, wherein: The specific implementation of the quadruped robot's stop position decision algorithm is as follows: first, hotspot areas are merged, user devices are aggregated in sequence, and mobile edge servers are allocated. The service area is discretized into J grids with a discretization radius of j; In each time slot, the task intensity of grid j∈J is expressed as follows according to the Euclidean distance: where ε mj ∈{0,1} indicates whether the distance between task m and the center of grid j is within radius r. When the capacity of a grid reaches its upper limit, a mobile edge server is deployed at the center of the grid, and the tasks that have been served are removed. If the upper limit has not been reached, the radius r is increased and the above verification process is repeated. The heat is updated every time a server is deployed or r changes. The estimated value of C(r) is: Where C(r) represents the number of tasks completed by a single mobile edge server within the communication range r. The maximum number of tasks that can be processed under r, if r reaches the maximum communication range, no matter whether the task intensity of the current grid exceeds the server capacity, servers will be deployed to these grids in the order of decreasing task intensity; Repeat the above steps until the number of servers reaches the initial server budget or all tasks are processed. The y operation is terminated when the number of remaining tasks is lower than the threshold θ·C(r).

Images