TaskMe: Toward a dynamic and quality-enhanced incentive mechanism for mobile crowd sensing

https://doi.org/10.1016/j.ijhcs.2016.09.002Get rights and content

Highlights

  • The dynamic budgeting approach based on spatiotemporal contexts increases the successful completion of tasks.

  • A multi-facet quality measurement method is proposed, where a combination of two factors—completion ratio and quality indicator—are used for quality measurement.

  • A novel reverse auction mechanism is proposed to enhance quality of sensing.

Abstract

Incentive is crucial to the success of mobile crowd sensing (MCS) systems. Over the different manners of incentives, providing monetary rewards has been proved quite useful. However, existing monetary-based incentive studies (e.g., the reverse auction based methods) mainly encourage user participation, whereas sensing quality is often neglected. First, the budget setting is static and may not meet the sensing contexts or user anticipation. Second, they do not measure the quality of data contributed. Third, the design of most incentive schemes is quantity- or cost-focused and not quality-oriented. To address these issues, we propose a novel MCS incentive mechanism called TaskMe. An LBSN (location-based social network)-powered model is leveraged for dynamic budgeting and proper worker selection, and a combination of multi-facet quality measurements and a multi-payment-enhanced reverse auction scheme are used to improve sensing quality. Experiments on several user studies and the crawled dataset validate TaskMe's effectiveness.

Introduction

Ubiquitous computing is moving from individual sensing to social and urban sensing. Mobile crowd sensing (MCS) has become a novel paradigm to achieve this (Ganti et al., 2011, Guo et al., 2015b). In MCS, an increasing number of average users are allowed to share local knowledge (e.g., noise level Stevens and D'Hondt, 2010, local utilities Nawaz et al., 2013) acquired using their smartphones. However, during the participatory sensing process, the employed smartphones will consume their own resources such as communication, computation, and energy. If they cannot gain sufficient rewards, users may refuse to participate in sensing tasks. Though there are several intrinsic or practical reasons that motivate people to contribute, such as self-efficacy, entertainment, and sensing service usage, in many cases the sensing tasks are time-consuming and thus external incentives are needed. To this end, incentives have become a significant MCS research area.

Generally, each MCS app has three entities, the platform, task requesters, and workers. Tasks are specified by task requesters and published at the platform, and workers are then recruited to perform the tasks. Each task consists of numerous elements, such as its target, spatio-temporal constraints, and the budget for recruiting workers. However, since MCS tasks have complex types and genres, it is difficult to have a unique incentive mechanism to support varied tasks. Numerous incentive mechanisms (Lee and Hoh, 2010, Reddy et al., 2010, Yang et al., 2012, Jaimes et al., 2012, Li and Cao, 2013, Feng et al., 2014, Kawajiri et al., 2014, Zhang et al., 2014, Zhao et al., 2014, Tuite et al., 2011, Rula and Bustamante, 2015) have been developed. They can be broadly divided into two major types: monetary and gamification. The former one offers workers monetary rewards (Lee and Hoh, 2010, Reddy et al., 2010, Feng et al., 2014) and the latter one attracts them with fun (Kawajiri et al., 2014, Tuite et al., 2011, Rula and Bustamante, 2015). As studied in Reddy et al. (2010), compared to gamification, monetary can better reinforce good data collection habits. The Reverse Auction (RA) method is a widely used method for monetary incentives in MCS. It allows people to sell their data based on their wills. Numerous incentive schemes are developed based on it (Lee and Hoh, 2010, Jaimes et al., 2012, Li and Cao, 2013, Feng et al., 2014, Zhao et al., 2014). However, quality of sensing is not considered in these studies. For example, the following aspects that can affect sensing quality are not considered.

(1) Budget setting and worker recruiting: The budget should be defined for monetary-based incentives. It is generally set statically in existing RA-based studies. However, the given budget may be too low/high to complete a task, especially when the task requester is unfamiliar with the target area. Without a reasonable budget setting, the quality of sensing of the task cannot be ensured, or over-payment may happen. It is thus crucial if the system can give a reference budget level, but this is difficult since many factors may affect the cost of a task, e.g., worker income status, workload, sensing environment, needed workers, etc. Given a reasonable budget, we should also select well-suited workers to perform it. Quality can be impacted when workers are not familiar with the target area or have to put in significant effort (e.g., she does not often visit that area) to perform the task.

(2) Data quality measurement: To enhance quality of sensing for the selected workers, the fundamental issue is finding an appropriate way to measure the quality of data contributed by them. Data quality is hard to measure as it is affected by many factors, such as the data type, accuracy, sensing context, etc. Data quality is also impacted by their distribution regarding task needs. Some tasks may have predefined points to cover. We can group the sensed data to the points to measure each worker's contribution. However, in many cases such as for crowd search tasks (Nawaz et al., 2013, Guo et al., 2015a, Ouyang et al., 2013, Mohan et al., 2008), the points cannot be pre-specified, and we should first learn the right points that should be covered. To improve quality of sensing, general metrics should be proposed to evaluate quality of contribution from workers (Goncalves et al., 2015a).

(3) Quality-enhanced incentives: Existing incentive schemes encourage active user participation and quantity of data, but the quality of sensing has attracted little attention. For instance, most of them execute bidding and winner selection before task performing, which cannot assure data quality because no competition is involved in the data collection process. Furthermore, quality is not considered in winner selection and payment. For instance, a well-performed worker who raises a relatively high bid regarding to her effort may lose in winner selection and gain no rewards. This will harm the enthusiasm of workers and impact quality of sensing for the task requests in the future.

To address the above issues, we have proposed TaskMe, a dynamic and quality-enhanced incentive mechanism for MCS tasks. It is founded on the reverse auction method and has the following contributions.

(1) An LBSN (location-based social network)-powered model for dynamic budgeting and worker selection: Based on the cross-community sensing paradigm (Guo et al., 2014), we leverage online crowdsourced data to steer offline crowd sensing. By extracting spatio-temporal contexts and human behavior patterns from online LBSN data (e.g., check-ins from FourSquare1 or JiePang2), the model can give a reference to define the budget of offline MCS tasks and select proper workers for them.

(2) Multi-facet quality measurement: Two common factors, completion ratio (CR) and quality indicator (QI), are identified for quality measurement. For CR, we consider situations with both predefined and non-defined points, and present the method for data grouping. For QI, we propose a crowd intelligence-based method for numeric-reading accuracy measurement and the context-based approach for picture quality estimation.

(3) Quality-enhanced incentives: A novel reverse auction mechanism is proposed to enhance quality of sensing, with strategies such as post bidding, quality-enhanced winner selection, and the multi-payment scheme.

Experiments have been conducted over several user studies and a crawled dataset from JiePang. The results indicate that the proposed mechanism not only supports dynamic budgeting and community-aware worker selection, but enhances quality of sensing by using the improved winner selection and multi-payment schemes.

Section snippets

Related work

The development of the TaskMe incentive mechanism has several closely related research areas, as discussed below.

An overview of TaskMe

TaskMe aims to develop a dynamic and quality-enhanced incentive mechanism for MCS tasks. This section presents an overview of our work.

The incentive mechanism design

Having given an overview of TaskMe, we present the detailed design of the incentive mechanism.

Evaluation and performance analysis

Having presented the methods proposed in TaskMe, in this section we present the results to validate our work.

Conclusion

We have presented TaskMe, a novel MCS incentive mechanism that supports quality-enhanced data collection. Different from traditional monetary-based incentives, to enhance sensing quality, we add more dynamics in our mechanism, such as dynamic budgeting, dynamic worker selection, and quality-based dynamic payment. Detailed design to the key components of TaskMe is presented, including the STP-based budget reference model, LBSN-based worker selection, multi-facet quality measurement,

Acknowledgments

This work was supported by the National Basic Research Program of China (No. 2015CB352400), and the National Natural Science Foundation of China (Nos. 61332005, 61373119, 61602230).

References (41)

  • R.K. Ganti et al.

    Mobile crowdsensingcurrent state and future challenges

    IEEE Commun. Mag.

    (2011)
  • J. Goldman et al.

    Participatory sensinga citizen-powered approach to illuminating the patterns that shape our world

    Foresight & Governance Project, White Paper

    (2009)
  • J. Goncalves et al.

    Worker performance in a situated crowdsourcing market

    Interact. Comput.

    (2015)
  • B. Guo et al.

    Fliermeeta mobile crowdsensing system for cross-space public information reposting, tagging, and sharing

    IEEE Trans. Mob. Comput.

    (2015)
  • B. Guo et al.

    Cross-community sensing and mining

    IEEE Commun. Mag.

    (2014)
  • Hasan, S., Zhan, X., Ukkusuri, S.V., 2013. Understanding urban human activity and mobility patterns using large-scale...
  • K. He et al.

    Single image haze removal using dark channel prior

    IEEE Trans. Pattern Anal. Mach. Intell.

    (2011)
  • Jaimes, L.G., Vergara-Laurens, I., Labrador, M.A., 2012. A location-based incentive mechanism for participatory sensing...
  • Kawajiri, R., Shimosaka, M., Kashima, H., 2014. Steered crowdsensing: incentive design towards quality-oriented...
  • D.S. Kerby

    The simple difference formulaan approach to teaching nonparametric correlation

    Innov. Teach.

    (2011)
  • Cited by (78)

    • A reliability and truth-aware based online digital data auction mechanism for cybersecurity in MCS

      2023, Future Generation Computer Systems
      Citation Excerpt :

      Sometimes, participants need to move to the specific place to sense data, which costs more [15,16]. Thus, many incentive mechanisms have been proposed to encourage participants to collect data [17–19], which can be classified into two categories. One is the offline incentive mechanism, assuming all participants are available for selection without considering the limitation of time and space [20,21].

    • Employing Social Participants for Timely Data Collection Using Pub/sub Solutions in Dynamic IoT Systems

      2023, Computer Networks
      Citation Excerpt :

      Service requesters publish sensing tasks and the budget to the platform [17–19]. Then the platform adopts publish/ subscribe model to collect data [20], which is to publish sensing task with the requirements of location, time, data attributes, payments and so on [21–23]. Once the participants are informed of the published tasks, they will subscribe sensing tasks which they can perform and bids to the platform [24–26].

    • Research on group awareness of networked collaboration within the design team and between teams

      2021, Advanced Engineering Informatics
      Citation Excerpt :

      Poppe et al. [42] proposed to use the virtual avatar to provide visual cues for team collaboration, making the communication process easier and more efficient. The essence of collaboration is that there are some connections between team members in completing collaborative tasks [43]. So these connections determine that members need to obtain the required information through communication during the collaboration process.

    View all citing articles on Scopus
    View full text