Opportunistic crowd-sensing online task allocation method based on user pre-grouping

Abstract

The invention discloses an opportunistic crowd-sensing online task allocation method based on user pre-grouping. The opportunistic crowd-sensing system has multiple task requesters and multiple users who can execute perception tasks. Each task requester has one or more tasks, each task is indivisible, and each task needs to be completed by one user. The task requester and the user move randomly in the target environment. The allocation of perception tasks and the return of perception results can only be performed when the task requester and the user directly meet in space. The interaction of perception tasks and results is performed through short-range wireless communication. The invention mainly includes: a user pre-grouping method combining task load and encounter rules, an intra-group online task allocation method based on the first idle user priority allocation and the maximum load task priority allocation, and the goal is to minimize the maximum expected completion time of all tasks of all task requesters.

Description

Technical Field

[0001] The present invention relates to the field of data perception technology, and in particular to an opportunistic crowd-sensing online task allocation method based on user pre-grouping. Background Art

[0002] In recent years, with the rapid development of information technology, mobile smart devices have become more and more popular. Mobile devices such as smart phones and watches have various types of sensors such as global positioning systems, thermometers, distance sensors, light sensors, cameras, etc., and have powerful computing capabilities. Users carrying these devices move around in urban areas and make full use of the idle resources of these devices, which has led to the birth of mobile crowd intelligence perception. Since there is no need to deploy a large number of fixed sensor nodes and it has a high spatiotemporal coverage, mobile crowd intelligence perception has a wide range of applications in environmental monitoring, traffic management, and smart cities. However, in the mobile crowd intelligence perception system, the total task load of different task requesters often has a large difference, and there is competition for user use between each other. The goal of the present invention is to minimize the maximum expected completion time of all tasks of all task requesters. Summary of the invention

[0003] The present invention relates to an opportunistic crowd-sensing online task allocation method based on user pre-grouping, which is characterized by: an opportunistic crowd-sensing system has multiple task requesters and multiple users who can execute perception tasks, each task requester has one or more tasks, each task is indivisible and needs to be completed by one user, the task requester and the user move randomly in a target environment, the allocation of perception tasks and the return of perception results can only be performed when the task requester and the user directly meet in space, and the interaction of perception tasks and results is performed through short-range wireless communication methods such as DSRC / 802.11p / Bluetooth / WiFi, etc. The method mainly includes: a user pre-grouping method combining task load and encounter rules, and an intra-group online task allocation method based on first idle user priority allocation and maximum load task priority allocation.

[0004] The application environment of the present invention is as follows:

[0005] Each task requester i stores locally the expected meeting time interval EMT between it and each user j ij ;

[0006] The expected completion time of task k assigned to user j by task requester i, denoted as T ik ,have:

[0007]

[0008] Where S ij is the set of tasks assigned to user j by task requester i, sp is the set of tasks S ij The pth task in τ p It is a task p The processing time required; before user j completes task k, he must first complete the tasks that have been assigned to him by task requester i and are ranked before task k, s p ≤s k It means that when user j completes task p, task k is completed before task k; ij is the expected encounter time required for task requester i to assign the task to user j and for user j to return the perception result to task requester i. If task requester i assigns the task to user j and happens to directly meet him in space, then δ ij =EMT ij , otherwise δ ij =2EMT ij ;

[0009] The expected processing time required for user j to complete the task assigned by task requester i, recorded as EPT ij :

[0010]

[0011] where τ ik It is a task ik The required processing time is proportional to the load of the task;

[0012] The first idle user of task requester i is the EPT among all users assigned to the task requester iq The user with the smallest value. Here, user q is a user in the task requester user group.

[0013] The expected task completion time M of task requester i i is the maximum expected processing time of all users in the task requester user group, that is:

[0014]

[0015] in is the set of users assigned to task requester i after user pre-grouping, and Represents task requester i and user u ix Expected meeting time.

[0016] User pre-grouping method based on task load and encounter pattern:

[0017] • Phase 1: In this phase, a user is assigned to each task requester. The specific process is as follows: Each time, the task requester with the largest total task load is assigned a user with the shortest expected meeting interval with it, and then the user is deleted from the set of optional users. This process is continued until each task requester's user group has one and only one user; Assuming that each task requester assigns all its tasks to the only user in its user group, on this basis, the expected task completion time of each task requester is calculated;

[0018] • Phase 2:

[0019] Based on Phase 1, continue with the following steps:

[0020] 1) When there are users who have not been grouped, perform the following steps;

[0021] 2) Assign a user who has not been grouped yet, that is, the user with the shortest expected meeting interval, to the task requester with the largest expected completion time; then virtually allocate all tasks of the task requester within the group, with the largest task being allocated first and then being allocated to the first idle user. Continue this virtual allocation process until all tasks are virtually allocated. On this basis, update the expected completion time of the task requester's task;

[0022] 3) If there are no ungrouped users, then end; otherwise, go to step 2).

[0023] The online task allocation method within the group is based on the priority allocation of the first idle user and the priority allocation of the largest load task:

[0024] a) Whenever a task requester encounters a user belonging to its own user group and has not yet assigned a task to the user, it starts to assign a virtual task;

[0025] b) The task requester virtually allocates the remaining tasks to all users in his group who have not yet been assigned any tasks. The principle is that the task with the largest load is allocated first, and it is allocated to the first idle user. This process continues until all tasks are virtually allocated;

[0026] c) For the current user, the task requester actually assigns all the tasks virtually assigned to the user in the previous step to the user, and still holds the remaining tasks, waiting to be assigned when encountering other users in the group who have not yet been assigned tasks in the future.

Claims

1. An opportunistic crowd-sensing online task allocation method based on user pre-grouping, characterized by: The opportunistic crowd-sensing system has multiple task requesters and multiple users who can perform perception tasks. Each task requester has one or more tasks. Each task is indivisible and needs to be completed by one user. The task requester and the user move randomly in the target environment. The allocation of perception tasks and the return of perception results can only be carried out when the task requester and the user meet directly in space. The interaction of perception tasks and results is carried out through short-range wireless communication. It mainly includes: a user pre-grouping method combining task load and encounter rules, an in-group online task allocation method based on the first idle user priority allocation and the maximum load task priority allocation; in Each task requester i stores locally the expected meeting time interval EMT between it and each user j ij ; ●The expected completion time of task k assigned to user j by task requester i, denoted as T ik ,have: Where S ij is the set of tasks assigned to user j by task requester i, s p is the set of tasks S ij The pth task in τ p It is a task p The required processing time is proportional to the load of the task. Before completing task k, user j must first complete the tasks that are assigned to the user by task requester i and are ranked before task k. p ≤s k It means that when user j completes task p, task k is completed before task k; ij is the expected encounter time required for task requester i to assign the task to user j and for user j to return the perception result to task requester i. If task requester i assigns the task to user j and happens to directly encounter him in space, then d ij =EMT ij , otherwise ij =2EMT ij ; ●The expected processing time required for user j to complete the task assigned by task requester i, denoted as EPT ij : where τ ik It is a task ik The required processing time is proportional to the load of the task; ●The first idle user of task requester i is the EPT among all users assigned to the task requester iq The user with the smallest value. Here, user q is a user in the task requester user group. ●The expected task completion time M of task requester i i is the maximum expected processing time of all users in the task requester user group, that is: in is the set of users assigned to task requester i after user pre-grouping, and Represents task requester i and user u ix Expected meeting time; User pre-grouping method based on task load and encounter pattern: ● Phase 1: In this phase, a user is assigned to each task requester. The specific process is as follows: Each time, the task requester with the largest total task load is assigned a user with the shortest expected meeting interval with it, and then the user is removed from the set of optional users. This process is continued until each task requester's user group has one and only one user; assuming that each task requester assigns all its tasks to the only user in its user group, on this basis, the expected task completion time of each task requester is calculated; Phase 2: Based on Phase 1, continue with the following steps: 1) When there are users who have not been grouped, perform the following steps; 2) Assign a user who has not been grouped yet, that is, the user with the shortest expected meeting interval, to the task requester with the largest expected completion time; then virtually allocate all tasks of the task requester within the group, with the largest task being allocated first and then being allocated to the first idle user. Continue this virtual allocation process until all tasks are virtually allocated. On this basis, update the expected completion time of the task requester's task; 3) If there are no ungrouped users, then end; otherwise, go to step 1.

2. The opportunistic crowd-sensing online task allocation method based on user pre-grouping according to claim 1 is characterized in that: The online task allocation method within the group is based on the priority allocation of the first idle user and the priority allocation of the largest load task: a) Whenever a task requester encounters a user belonging to its own user group and has not yet assigned a task to the user, it starts to assign a virtual task. The specific steps are as follows; b) The task requester virtually allocates the remaining tasks to all users in his group who have not yet been assigned any tasks. The principle is to prioritize the tasks with the largest load and allocate them to the first idle user. This process continues until all tasks have been virtually allocated. c) For the currently encountered user, the task requester actually assigns all the tasks virtually assigned to the user in the previous step to the user, and still holds the remaining tasks, waiting to be assigned when encountering other users in the group who have not yet been assigned tasks in the future.

Images