作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2022, Vol. 48 ›› Issue (9): 45-54. doi: 10.19678/j.issn.1000-3428.0064071

• 热点与综述 • 上一篇    下一篇

面向单任务质量保障的移动群智感知任务分配

杨桂松1, 吴笑天1, 高丽2, 何杏宇1,3   

  1. 1. 上海理工大学 光电信息与计算机工程学院, 上海 200093;
    2. 上海理工大学 图书馆, 上海 200093;
    3. 上海理工大学 出版印刷与艺术设计学院, 上海 200093
  • 收稿日期:2022-03-02 修回日期:2022-04-27 发布日期:2022-09-08
  • 作者简介:杨桂松(1982—),男,副教授、博士,主研方向为物联网、普适计算;吴笑天,硕士研究生;高丽,副研究馆员、博士;何杏宇(通信作者),副教授、博士。
  • 基金资助:
    国家自然科学基金(61602305,61802257);上海市自然科学基金(18ZR1426000,19ZR1477600)。

Task Allocation Towards Individual Task Quality Assurance in Mobile Crowd Sensing

YANG Guisong1, WU Xiaotian1, GAO Li2, HE Xingyu1,3   

  1. 1. School of Optical-Electrical & Computer Engineering, University of Shanghai for Science & Technology, Shanghai 200093, China;
    2. Library Department, University of Shanghai for Science & Technology, Shanghai 200093, China;
    3. College of Communication & Art Design, University of Shanghai for Science & Technology, Shanghai 200093, China
  • Received:2022-03-02 Revised:2022-04-27 Published:2022-09-08

摘要: 在移动群智感知中,现有的任务分配方法大多关注平台的整体感知质量,未充分考虑任务对工人、预算等资源的竞争,无法有效保障大规模任务分配场景下每个任务的感知质量,从而导致平台资源利用率降低。针对该问题,提出一种面向单任务质量保障的任务分配方法。为高效利用平台预算,考虑任务的难度和位置以及工人的设备能耗和理性因素,设计平台的激励成本。为保障每个任务的感知质量,考虑任务间的资源竞争情况并设计2种衡量指标,分别是从任务的角度根据差异化感知质量需求设计任务覆盖效率,以及从工人的角度基于最大熵原理设计工人利用效率,将这2种衡量指标相结合作为平台的系统效用,在平台资源有限的情况下以平台系统效用最大化为优化目标,提出一种融合交叉和变异操作的天牛群(BSO)算法。实验结果表明,与PSO、GA等基线方法相比,BSO算法的系统效用最大值平均提升13.51%,寻优速度平均提高40.61%,利用该算法获取的具有最大系统效用的任务分配方案可以有效保障每个任务的感知质量。

关键词: 移动群智感知, 任务分配, 单任务质量保障, 系统效用, 天牛群算法

Abstract: In Mobile Crowd Sensing(MCS), most existing task allocation methods focus on the overall sensing quality of the platform and do not fully consider task competition for workers, budgets and other resources.This cannot effectively guarantee the sensing quality of each task in a large-scale task allocation scenario, resulting in a reduction of resource utilization of the platform.To solve this problem, a task allocation method for individual task quality assurance is proposed.To efficiently use the platform budget, the incentive cost of the platform is designed by considering the difficulty and location of tasks and the energy consumption and rationality of worker equipment.To ensure the sensing quality of each task, the resource competition between tasks is considered and two measurement indicators are implemented:the task coverage efficiency according to the differentiated perceived quality requirements from the perspective of tasks and the worker utilization efficiency based on the maximum entropy principle from the perspective of workers.These two measurement indicators are combined as the system utility of the platform.In the case of limited platform resources, a Beetle Swarm Optimization(BSO) algorithm that integrates crossover and mutation operations is proposed for maximizing the system utility of the platform.The experiments show that the maximum system utility and optimization speed of the BSO algorithm increases by 13.51% and 40.61% on average, respectively, compared with Particle Swarm Optimization(PSO), Genetic Algorithm(GA), and other baseline methods.The task allocation scheme with the maximum system utility obtained by this algorithm can effectively ensure the sensing quality of each task.

Key words: Mobile Crowd Sensing(MCS), task allocation, individual task quality assurance, system utility, Beetle Swarm Optimization(BSO) algorithm

中图分类号: