群智感知需求不确定任务的资源分配方法  

A Resource Allocation Method for Task with Demand Uncertainty in Crowd Sensing

作　　者：姚秋言 赵丹 YAO Qiuyan;ZHAO Dan(School of Optical-Electrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093)

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093

出　　处：《计算机与数字工程》2024年第10期3019-3025,共7页Computer & Digital Engineering

摘　　要：对群智感知任务类型中的突发任务资源分配问题进行研究。首先分析突发任务的特点,建立突发任务需求不确定的多阶段随机规划模型,并使用三个指标:效率、效力和公平来衡量资源的分配,提出以最小化成本为目标的非线性优化问题。然后,针对优化问题,提出基于Q学习算法的资源分配方法,并与动态规划算法和启发式算法作对比。实验结果表明,Q学习算法在精度上优于启发式算法,在计算速度上优于动态规划算法。The problem of resource allocation for sudden tasks in crowd sensing has been studied.Firstly,the characteristics of sudden tasks are analyzed and a multi-period stochastic programming model with uncertain demand for sudden tasks is estab-lished.Three indicators are used to measure the allocation of resources,which are efficiency,effectiveness and fairness.A nonlin-ear optimization problem for minimizing cost is proposed.Then,aiming at the optimization problem,a resource allocation method based on Q learning algorithm has put forward and compare with dynamic programming algorithm and heuristic algorithm.Experi-mental results show that Q learning algorithm is better than heuristic algorithm in accuracy and dynamic programming algorithm in computing speed.

关 键 词：群智感知 需求不确定 多阶段随机规划 Q学习 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]