参与式众包的QoI与节能策略探索

177 浏览量更新于2024-07-14 收藏 561KB PDF 举报

"参与式众包的QoI和节能" 这篇研究论文探讨了参与式众包中的质量-of-Information（QoI）和节能问题，它已被接受发表在未来的《IEEE Transactions on Vehicular Technology》期刊上。文章的作者包括Chi Harold Liu、Jun Fan、Pan Hui、Jie Wu和Kin K. Leung，他们都是IEEE的成员或院士。随着智能手机的普及，传统的众包概念正在向一种新的、任务导向型的参与式应用转变。这种新型的众包模式，被称为“公民科学”努力，旨在利用大众的力量进行知识探索，理解人类行为，并对人们的观点进行测量和评估。为了支持这些场景，论文提出了一种创新的网络管理框架，以提高参与式众包的效率和能源效率。在抽象部分，作者指出，他们设计的框架着重于优化用户参与的体验（QoI）同时减少能耗，这对于依赖移动设备如智能手机的众包平台至关重要。由于这些设备通常电池寿命有限且运行时会消耗大量能源，因此在不牺牲任务完成质量的前提下实现节能成为了一个挑战。论文可能详细讨论了以下几个方面： 1. **QoI模型**：定义并量化了参与式众包中的QoI指标，可能包括任务完成的准确性、响应时间、用户满意度等，这些都是衡量众包服务质量的关键因素。 2. **节能策略**：提出了各种方法来降低众包过程中的能量消耗，比如智能的任务调度、功耗优化的通信协议、以及利用设备的休眠模式等。 3. **网络管理框架**：描述了一个整体架构，可能包括任务分配机制、资源管理模块、以及适应性算法，这些都能确保高效且节能的众包操作。 4. **实验与评估**：通过模拟或真实世界的数据集进行实验，验证了所提出的框架在提高QoI和节能方面的有效性，可能比较了不同策略下的性能。 5. **未来研究方向**：可能还指出了当前研究的局限性和未来可能的研究方向，如更复杂的环境下的QoI优化，或者考虑更多用户行为因素的影响。这篇论文对于理解如何在保持高质量众包服务的同时，最大限度地减少能源消耗，对移动计算和物联网领域的研究者和开发者具有重要的参考价值。其结果可能有助于推动更加绿色和可持续的众包实践。

http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI

10.1109/TVT.2014.2367029, IEEE Transactions on Vehicular Technology

Particularly in (4), parameter p

, ∀i ∈ N exactly describes the

property of each user, connecting the contributed information

with its recognized value. At the same time, it also indicates

the QoI difference among users. On the basis of (4), if we

consider the decreasing marginal returns in the amount of

information, another example can be:

(l) = ϑ

i=1

(l)/

i=1

(l), (6)

where ϑ ∈ (0, 1] is the decay coefﬁcient, and  is the scaling

factor.

As our proposal is based on the mathematical paradigm

of Gur Game, our solution is transparent to any speciﬁc

form of function used in the QoI model, whether or not it

is discontinuous, multimodal, or concave, etc. In this paper,

we adopt the fusion functions in (4) and (6), and verify the

adaptability of our solution in the Section VII.

B. System Flow

Gur Game [20], [21] was proposed to use in distribute

systems who wish a collection of agents to cooperate on a

task. Each agent is associated with a ﬁnite state automaton

that independently guides the agent’s action, while taking

into account the collective feedback that eventually captures

the composite effect of all agents’ actions. Compare to our

considered participatory crowdsourcing scenarios, the partic-

ipant’s smart device in this case acts as the “agent”, where

the associated automaton can be easily deployed through a

piece of software in the mobile OS. The “task” translates

exactly to our focused social studies crowdsourced from a

co-located group of participants; and the “composite effect

of all agents’ actions” is then the result of the participants’

action upon returning answers to the querier. Therefore, we

believe that the fundamentals of Gur Game serve as the ideal

engine algorithmically due to its robustness, simplicity and

decentralized features. To make it particularly suitable for

crowdsourcing, in this paper we largely extend the existing

Gur Game and use it as part of the overall system ﬂow shown

in Fig. 1.

The system ﬂow consists of two stages. The ﬁrst stage

relates to the interaction between the information center of

the network platform and the Gur Game engine programmed

in user’s smart device. The inputs are the user’s smart device

residual energy levels and multiple QoI requirements of the

request. At each iteration step of the Gur Game, the smart

device sends its preliminary action (as a result of our proposed

Gur Game algorithm) back to the platform. The latter then

calculates the pay-off value based on the received collective

pieces of information from all participants, and propagates it

back to Gur Game engine of each user. Based on this feedback,

the automaton changes its current state and generates the new

action, i.e., the level of information contribution. Therefore,

the Gur Game engine uses the trial-and-error method to

produce the best result at each step and iteratively achieves

the overall optimum to fulﬁll the QoI requirements in aware

of energy efﬁciency.

Sept. 5, 2008

June 29, 2009

a 3-day time period

phone usages

questionnaire request

91 time periods

time

ĂĂĂĂ

lunch dinner lunch lunch dinnerdinner

ĂĂ

Fig. 2. The illustration of quantizing the entire duration into 91 time periods

and when considered participatory crowdsourcing request arrives.

The second stage relates to the interaction between the credit

center of the network platform and the bidding module of

user’s smart device. After the ﬁrst stage, the platform obtains

each user’s preliminary action. Then, the users send their bids

to the platform, which represent their expected paid credits for

unit amount of information contribution. The network platform

selects the users who can meet the QoI requirements of the

request and help reduce total paid credits as the active users

for questionnaire answering, receives their ﬁnal information

contribution determined by the Gur Game engine in the ﬁrst

stage, and pay their credits according to the previous bids. We

next describe a detailed implementation and solution of this

framework.

IV. A CASE STUDY

This section ﬁrst provides an overview of the used Social

Evolution data set gathered by MIT Media Lab [29], and

then illustrates how we use it to motivate our participatory

crowdsoucing application. The data set is generated by an

application on 80 undergraduates’ smart devices, who move

around the campus. It collects the phone usages and student

locations from October 2008 to June 2009. The phone usage

data consist of 3.15 million records of Bluetooth scans, 3.63

million scans of WLAN access-points, 61,100 call records, and

47,700 logged SMS events. Also, students provide ofﬂine, self-

report answers related to their health habits, diet and exercise,

weight changes, and political opinions during the presidential

election campaign. In our simulation, we use the phone usage

data, and the self-report answers on the health condition to

motivate and form the participatory crowdsourcing process as

described below.

A. Phone Usage Data

We extract the phone usage records from September 5, 2008

to June 29, 2009, in a total of 273 consecutive days. The

data include 49,906 voice call records with the calling time,

duration, caller and callee information, 33,148 SMS events

including the sending time, sender and receiver information.

Then, to facilitate our simulation, we quantize the entire 273

days into 91 independent time periods each of which involves

a 3-day usages, as shown in Fig. 2. We assume that phone

usages between two consecutive periods do not interrelate with

each other, and since the phone battery status is not provided

in the data set we simply assume that at the beginning of any

time period the device is fully charged. As a result, we have

91 time periods, each of which is associated a set of phone

usages, and this is used to evaluate how different phone data

impact on the proposed algorithm. Besides, we also make the

following two assumptions:

剩余16页未读，继续阅读

weixin_38692666

粉丝: 6
资源: 914

参与式众包的QoI与节能策略探索

知识图谱的众包构建

众包测试的威力

众包下的电子商务的发展与研究

车辆众包概述 纤细介绍

面向众包任务的高质量参与者推荐研究——以猪八戒平台为例

kppw众包建站系统

什么是JAD吗？什么是参与式设计？

冲压车间节能金点子有那些

节能减排大赛相关项目

最新资源

车辆众包概述纤细介绍