面向移动群智感知的多任务分发算法

针对在移动群智感知中基于机会通信完成数据传输会消耗大量时间成本的问题,提出了一种基于中枢节点的多任务分发（HTA）算法。该算法利用节点在移动网络中社交关系属性不同的特点,通过中枢节点选择算法将部分节点作为中枢节点,并将其用于协助任务请求节点分发任务。在任务请求节点与中枢节点相遇时,同时给中枢节点本身和它的从属节点分配任务,并由中枢节点负责向从属节点分发任务与回收任务结果。基于The ONE模拟器进行实验,与在线任务分配（NTA）算法相比,HTA算法时间成本平均降低了24.9%,同时任务完成率平均提高150%。实验结果表明,HTA算法能够提高任务的完成速度,降低时间成本消耗。     

A QoS-sensitive task assignment algorithm for mobile crowdsensing

Recently, mobile crowdsensing has attracted many researchers’ attention due to the pervasiveness of smart phones. In this paper, we study the QoS-sensitive Task Assignment (QSTA) problem for mobile crowdsensing, which involves variable tasks and flexible rewards. Each user can conduct one or more tasks, and the number of tasks assigned to each user is flexible. The reward paid to the user is related to the assigned tasks. To ensure the sensing QoS of tasks, each task might be assigned to multiple users. We prove the QSTA problem to be NP-hard, and propose a greedy algorithm to solve this problem. Moreover, we also analyze the approximation ratio of the proposed algorithm. Additionally, we demonstrate the significant performance of the algorithm through simulations     

深度强化学习综述简

移动群智感知（mobile crowdsensing, MCS）是利用大规模移动智能设备进行数据收集、数据挖掘和智能决策的新范式,高效的任务分配方法是MCS获得高性能的关键. 传统的贪婪算法或蚂蚁算法假设工人和任务固定,不适用于工人和任务的位置、数量和时间动态变化的场景. 而且,现有任务分配方法通常由中央服务器收集工人和任务的信息进行决策,容易导致工人隐私泄露. 因此,提出具有隐私保护的深度强化学习（deep reinforcement learning, DRL）模型来获得优化的任务分配策略. 首先,将任务分配建模为多目标优化的动态规划问题,旨在最大化工人和平台的双向收益,实现纳什均衡. 其次,提出基于DRL的近端策略优化（proximal policy optimization, PPO）模型进行训练,学习模型参数. 最后,通过本地差分隐私方式,对工人位置等敏感信息加入随机噪声实现隐私保护,并由中央服务器训练整个模型,获得最优分配策略. 对收敛时间、最大收益和任务覆盖率等指标进行实验评估,在模拟数据集上的实验结果表明,与传统方法和其他基于DRL的方法对比,该方法在不同的评估指标上均有明显提升,并且能够保护工人的隐私.

     

基于区块链的边缘移动群智感知声誉更新方案

移动群智感知利用移动用户的智能终端设备以低成本获取大量感知数据,而恶意用户可能上传虚假数据以获取奖励。声誉管理是一种有效的解决办法,但是基于云服务器的移动群智感知系统存在高延迟、单点故障和隐私泄露问题。针对这些问题,结合区块链和边缘计算构建基于区块链的边缘移动群智感知系统,提出一种感知数据隐私保护的声誉更新方案,采用轻量级的隐私保护方法聚合感知数据,根据数据质量和历史任务表现更新声誉。该方案可有效抵抗恶意用户、降低时延,避免单点故障和保护数据隐私。仿真实验证明了所提方案的可行性和高效性,理论分析证明了系统的安全性。     

Analyzing crowdsourcing to teach mobile crowdsensing a few lessons

Cognition, Technology & Work - In recent years, mobile computing has shown so much potential that one can see a boundary-blurring expansion between the physical and the digital world. In this...     

群智感知中面向移动群体的参与者选择优化模型

随着短视频时代的来临,移动群智感知任务的视频化程度越来越高,在传统研究中常利用机会网络和移动网络激励任务的分发和数据的收集,但机会网络中节点移动的不可控性,以及视频任务内容传输的高代价性都使得这些方法的实用性大大降低.针对此问题,利用社会移动群体规律性的自主聚集、活动范围大等特点,提出一种面向社会移动群体的群智感知参与...     

移动群智感知中面向任务需求的用户选择激励机制

现有的移动群智感知激励机制大多以平台为中心或是以用户为中心进行设计，缺乏对感知任务需求的多维考虑，从而无法切实地以任务为中心进行用户选择，导致无法满足任务需求的最大化和多样化。针对上述问题，提出一种面向任务需求的用户选择激励机制TRIM，这是一种以任务为中心的设计方法。首先，感知平台根据任务需求发布感知任务，并从任务类型、时空特性以及感知报酬等多维度构建任务向量以最大化满足任务需求，而感知用户则基于意愿偏好、个人贡献值以及期望报酬等属性构建用户向量，实现个性化选择感知任务参与响应；然后，通过引入高效且隐私保护的余弦相似度计算协议（PCSC），计算任务和用户的相似度并根据相似度高低进行用户匹配筛选得到目标用户集，更好地满足感知任务需求的同时保护用户隐私不泄露；最后，通过仿真实验表明，在感知任务和感知用户的匹配过程中，与采用Paillier加密协议的激励机制相比，TRIM缩短了指数级增量的计算时间开销，提高了计算效率；与采用直接余弦相似度计算协议的激励机制相比，TRIM保证了感知用户的隐私安全，达到了98%的匹配精确度。     

基于压缩感知的移动群智感知任务分发机制

针对移动群智感知任务中区域全覆盖感知成本过高问题，提出基于压缩感知的移动群智感知任务分发（CS-TD）机制。首先提出了感知任务整体成本模型，该模型综合考虑了参与感知任务的节点个数、节点的感知次数与数据上传次数；然后基于成本模型，分析感知节点的日常移动轨迹，结合压缩感知数据采集技术，提出了一种基于感知节点轨迹的压缩感知采样方法；其次通过区域全覆盖最少节点（RCLN）算法，选出最佳节点集合，对节点进行任务分配，利用压缩感知技术恢复节点数据；最后在多次感知任务的迭代中对感知节点的可信程度进行评定，保证任务方案的最优性。对CS-TD分发模型进行多次实验验证，与已有的CrowdTasker算法相比，CS-TD算法平均成本降低了30%以上。CS-TD模型能有效降低感知节点的消耗，能在全覆盖感知任务中降低整体感知成本。     

移动感知环境下基于CSA-SSVR的交通状态预测方法

相较于传统感知网络,移动群智感知网络在部署和维护成本上有着较大优势,在智能交通系统中得到了越来越多的应用。交通状态的预测对交通管理系统具有重要意义,从移动群智感知环境下获取的车速数据出发,以支持向量回归算法(SVR)为基础,引入周期性算子,并采用布谷鸟算法(CSA)确定周期性SVR(SSVR)中的主要参数,提出了CSA-SSVR,对道路未来车速进行预测,据此判断道路的未来交通状态。实验表明,CSA-SSVR在移动群智感知环境下对于交通状态预测问题的准确性较高。     

Task reduction using regression-based missing data imputation in sparse mobile crowdsensing

The Journal of Supercomputing - Mobile Crowd Sensing (MCS) involves allocation of sensing tasks associated with an area of interest to a crowd of participants over time. Consequently, the...     

移动边缘群智感知动态隐私度量模型与评价机制

移动边缘群智感知中,用户执行感知任务采集数据所包含的隐私量是动态变化且不直观的,数据上传亦缺乏隐私风险预警值,提出一种动态隐私度量(DPM)模型.给出用户参与感知任务所获数据的结构化表示并转化成原始数值矩阵,引入隐私属性偏好与时效性因素实现对该矩阵的权重叠加,以度量数据所含隐私的动态变化,基于权重叠加后的矩阵合理计算用...     

Mobile agents for mobile tourists: a user evaluation of Gulliver's Genie

How mobile computing applications and services may be best designed, implemented and deployed remains the subject of much research. One alternative approach to developing software for mobile users that is receiving increasing attention from the research community is that of one based on intelligent agents. Recent advances in mobile computing technology have made such an approach feasible. We present an overview of the design and implementation of an archetypical mobile computing application, namely that of an electronic tourist guide. This guide is unique in that it comprises a suite of intelligent agents that conform to the strong intentional stance. However, the focus of this paper is primarily concerned with the results of detailed user evaluations conducted on this system. Within the literature, comprehensive evaluations of mobile context-sensitive systems are sparse and therefore, this paper seeks, in part, to address this deficiency.     

Secure mobile agents with controlled resources

Mobile agents often travel in a hostile environment where their security and privacy could be compromised by any party including remote hosts in which agents visit and get services. It is believed that the host visited by an agent should jointly sign a service agreement with the agent's owner; hence a proxy‐signing model was proposed in the literature, allowing every host in the agent system to sign a service agreement. We observe that this actually poses a serious problem whereby a host that should be excluded from an underlying agent network could also send a signed service agreement. In order to solve this problem, we propose two schemes achieving host authentication with controlled resources, where only selected hosts can be included in the agent network. We provide two schemes in this paper. The second scheme offers a smaller data size. We also define security models and provide rigorous security proofs to our schemes. Copyright © 2010 John Wiley & Sons, Ltd.     

Mobile agents

A lot of agents are executing on the Web, and some of them are starting to move around. While most agents are static (existing as a single process or thread on one host), others can pick up and move their code and data to a new host where they resume executing     

Flocking of mobile agents with fuzzy controller

Based on individual local observed information of the swarm, a fuzzy logical controller （FLC） of the attraction/repulsion function is built. Combining with the rate consensus algorithm, the swarming of the separation, cohesion and alignment is achieved. Based on the optimal controller theories, a capability target is used to measure the states of the system. In the computer simulation, it is shown that the aggregation of the multi-agent systems can get consensus following a mobile leader from the individual local perceived information.     

支持移动设备接入Gnutella网络的代理架构研究

针对当前移动设备无法加入Gnutella网络实现文件共享网络的现状,提出一种移动代理Gnutella网络架构的设计方案。利用移动代理作为移动设备的代理加入Gnutella网络,帮助移动设备实现文件的共享功能。架构能有效地减少移动设备的信息流量,并支持移动设备在Gnutella网络中实现可移动性。     

Integrating mobile agents into the mobile middleware

Mobile agents are a new paradigm for distributed computing that is especially well suited for mobile computing over global wireless networks. This paper describes the approach taken in the ACTS On TheMove project to integrate a mobile agent system into the Mobile Application Support Environment (MASE), a middleware for mobile computing. In this project, an existing mobile agent system was adapted for the requirements of mobile computing. We present the changes that had to be made to the agent system to adapt it to the wireless communication. We also present some of the application areas where a mobile agent system is suitable for mobile communication. We describe an agent based pre-fetcher application where an agent operaes disconnected from the user on the fixed network and prepares web pages for the anticipated next connection of the user using the Quality-of-Service trading functions available in MASE.