Elsevier

Pervasive and Mobile Computing

Volume 49, September 2018, Pages 111-125
Pervasive and Mobile Computing

The Curse of Sensing: Survey of techniques and challenges to cope with sparse and dense data in mobile crowd sensing for Internet of Things

https://doi.org/10.1016/j.pmcj.2018.06.009Get rights and content

Abstract

In this paper we present a survey on mobile crowdsensing (MCS) techniques that have been developed to address the Curse of Sensing problem i.e. propensity of MCS applications to generate sparse or dense data that can lead to significant gaps in the extracted knowledge. In order to do so, we identify features, based on the terminologies used in the literature, in order to develop a clear classifications among MCS and crowdsourcing applications and methods. Subsequently, we propose a taxonomy outlining both factors and objectives that need to be considered in designing MCS systems and have a direct impact on MCS applications’ tendency to fall into the Curse of Sensing. We then evaluate the majority of the research proposed in the field of MCS and addressing the Curse of Sensing problem with reference to the proposed taxonomy. Finally, we highlight the existing gaps in the literature and possible directions for future research.

Introduction

According to Ericsson’s Mobility Report [1], the total number of mobile phone subscriptions as of 2017 has equaled the world population (7.6 billion). It is predicted that by 2022, this will increase to 9 billion out of which 90% (over 8 billion) subscriptions will be 3G/LTE/5G based mobile smart devices. Another complimentary and converging paradigm that has fueled the development of many diverse applications in the recent years is the Internet of Things (IoT) [2]. It is fair to say that the increase in smart mobile phone subscriptions has further fostered and will continue to influence the growth of the IoT. According to Ericsson’s Internet of Things Forecast [1], in 2018, the mobile phone subscriptions is expected to surpass the number of IoT devices while by 2022, 70% of IoT devices will be powered by cellular technology providing them the fundamental access to Internet.

In the era of such phenomenal developments, Collective Awareness Paradigms (CAP) such as mobile crowdsensing (MCS) and crowdsourcing have regained significant attention from the research community. As mobile smart phones and the IoT become more ubiquitous, novel MCS applications (beyond current crowdsourced road navigation systems and Amazon Mechanical Turk-based applications) will start to emerge. We foresee that such future MCS applications for the IoT will have a significant influence in the data-driven decision making era. Nevertheless, many challenges lie ahead of these rapidly emerging technologies before mission-critical application (e.g. deciding to evacuate a building based on earthquake tremors reported by MCS and IoT sensor data) on top of these technologies can be developed and enter the plateau of productivity [3].

A fundamental challenge faced by MCS application for IoT is the Curse of Sensing. Inspired by Big Data’s Curse of Data, we define the Curse of Sensing as MCS applications’ inability to control sensing processes that may result in sparse or dense data. Sparse data may lead to insufficient information that will impact the ability to make data-driven decision and/or predictions; on the contrary, dense data could be influenced by Big Data’s Curse of Data problem, i.e. too much data leading to insights that may not reflect the real-world state and/or can lead to inaccurate predictions. Most of the recent surveys in the area of MCS focus on various aspects of mobile crowdsensing including defining the concept of crowdsensing [4], inspecting new applications [5], privacy [6], cost and QoS [7], data quality and credibility [8], incentive techniques [9] etc., with little or no attention to Curse of Sensing, that is impacted by issues such as participating rates and crowdsensing strategies [10]. However, there is a clear gap in the current literature in identifying techniques and challenges to address the Curse of Sensing problem. In summary, this survey paper will provide the following contributions:

  • Features to classify different CAPs: We identify features to classify crowdsourcing and MCS and, in particular, MCS into participatory and opportunistic (we provide definitions of these in Section 2). To the best of our knowledge, this is the first survey that distills the features that could be used to classify such applications, as, currently, works in the literature assume this to be known a priori or devote very little attention to the classification aspect, which is imperative in order to understand issues related with the Curse of Sensing.

  • Taxonomy of factors and objectives of MCS: We identify the factors and objectives that will have a direct impact on producing sparse or dense data in MCS applications.

  • Survey of techniques for coping with sparse and dense data in MCS: We provide a review of current approaches that have been carried out in recent years to address the Curse of Sensing problem. Here, we compare the techniques identified from the literature based on the previously established taxonomy.

  • Challenges in coping with sparse and dense data: We conclude the survey by identifying key challenges that are yet to be addressed in the literature that will aid and/or impact the Curse of Sensing problem in MCS applications.

The rest of the paper is organized as follows: in Section 2 we provide definitions from the literature and develop a taxonomy to classify crowdsourcing, participatory and opportunistic MCS; in Section 3 we identify the requirements and objectives of MCS; in Section 4 we provide a summary of existing approaches in the literature to cope with sparse and dense data; Section 5 provides future research challenges that will need to be addressed to solve the Curse of Sensing problem; finally, Section 6 concludes the survey.

Section snippets

Collective awareness paradigms: Background and motivation

We define Collective Awareness Paradigms (CAP) as paradigms that leverage the power of offloading tasks (as part of a campaign) to a crowd of individuals. The purpose is to collect data from crowds (large group of people), analyze and use such information for the benefit of the crowd itself [11]. CAPs were introduced in works like [12], where they are referred to as “collective intelligence”, based upon the fact that the aggregation of different points of view or observations leads to better

Taxonomy of mobile crowdsensing impacting sparse and dense data — The Curse of Sensing Problem

One of the major challenges in MCS that has not had much focus in the literature is dealing with the Curse of Sensing problem, i.e. dealing with deployments generating sparse and/or dense data. In this section we give a formal definition as well as a motivating scenario for the Curse of Sensing. In Section 2.3 we identified and briefly described the main feature to characterize and define MCS applications. In this section, we develop a taxonomy that expands on the definitions to further

Current state of the art in addressing the Curse of Sensing problem

In this section we focus on the diverse solutions and systems designed to cope with sparse and dense data in MCS. We use the term scenarios to refer to MCS applications similar to the ones described in Section 2.4. We describe each method in correlation with the factors and objectives as identified in the taxonomy (Section 3.2).

Future research landscape of MCS for Internet of Things

Designing a solution to develop MCS applications is a challenging task especially when MCS applications need to cope with dense and/or sparse data over which it has no direct control. In Section 4 we discussed the approaches that underpin the current research landscape and their relation with the factors and objectives of MCS described in Section 3. However, the approaches discussed in Section 4 partially consider the factors and objectives of MCS presented in Section 3. Hence, in this section

Conclusions

The advances in mobile devices technology coupled with the popularity of smart devices and the emerging wave of the Internet of Things paradigm has led to the development of large-scale application that can collect information about people and their environments in real-time from the physical world. Such applications are referred as Mobile Crowdsensing (MCS). Implementing a MCS system raises several challenges that need to be addressed for mainstream adoption. A fundamental challenge faced by

References (54)

  • ComptonM. et al.

    The ssn ontology of the w3c semantic sensor network incubator group

    Web Semant.: Sci. Servi. Agents World Wide Web

    (2012)
  • BoivinJ. et al.

    Are more data always better for factor analysis?

    J. Econometrics

    (2006)
  • N. Heuveldop, et al. Ericsson Mobility Report, Ericsson AB, Technol. Emerg. Business, Stockholm, Sweden, Tech. Rep....
  • GeorgakopoulosD. et al.

    Internet of things: from internet scale sensing to smart services

    Computing

    (2016)
  • K. Panetta, Top Trends in the Gartner Hype Cycle for Emerging Technologies. URL...
  • GantiR. et al.

    Mobile crowdsensing: current state and future challenges

    IEEE Commun. Mag.

    (2011)
  • MaH. et al.

    Opportunities in mobile crowd sensing

    Infocommun. J.

    (2015)
  • Vergara-LaurensI.J. et al.

    Privacy-preserving mechanisms for crowdsensing: Survey and research challenges

    IEEE Internet of Things J.

    (2016)
  • J. Liu, H. Shen, X. Zhang, A survey of mobile crowdsensing techniques: A critical component for the internet of things,...
  • F. Restuccia, N. Ghosh, S. Bhattacharjee, S. Das, T. Melodia, Quality of Information in Mobile Crowdsensing: Survey and...
  • ZhangX. et al.

    Incentives for mobile crowd sensing: a survey

    IEEE Commun. Surv. Tutor.

    (2016)
  • S.S. Kanhere, Participatory sensing: Crowdsourcing data from mobile smartphones in urban spaces, in: 2011 IEEE 12th...
  • W. Sherchan, P.P. Jayaraman, S. Krishnaswamy, A. Zaslavsky, S. Loke, A. Sinha, Using on-the-move mining for mobile...
  • MaloneT.W. et al.

    Harnessing crowds : Mapping the genome of collective intelligence

    MIT Sloan School Manag.

    (2009)
  • SurowieckiJ.

    The wisdom of crowds

    Amer. J. Phys.

    (2005)
  • HoweJ.

    The rise of crowdsourcing

    Wired Mag.

    (2006)
  • Estellés-ArolasE. et al.

    Towards an integrated crowdsourcing definition

    J. Inf. Sci.

    (2012)
  • GuoB. et al.

    Mobile crowd sensing and computing: The review of an emerging human-powered sensing paradigm

    ACM Comput. Surv.

    (2015)
  • N.D. Lane, S.B. Eisenman, M. Musolesi, E. Miluzzo, A.T. Campbell, Urban Sensing: opportunistic or participatory?, in:...
  • ChatzimilioudisG. et al.

    Crowdsourcing with smartphones

    IEEE Internet Comput.

    (2012)
  • P. Mohan, V. Padmanabhan, R. Ramjee, Nericell: Rich monitoring of road and traffic conditions using mobile smartphones,...
  • YavariA. et al.

    Contextualised service delivery in the internet of things: Parking recommender for smart cities

  • ChenX. et al.

    Crowdsourcing for on-street smart parking

  • MathurS. et al.

    Parknet: drive-by sensing of road-side parking statistics

  • SchurgersC. et al.

    Distributed on-demand address assignment in wireless sensor networks

    IEEE Trans. Parallel Distrib. Syst.

    (2002)
  • E.J. Candès, Compressive sampling, in: Proceedings of the International Congress of Mathematicians, 2006, pp....
  • LuoC. et al.

    Compressive data gathering for large-scale wireless sensor networks

    Mobicom

    (2009)
  • Cited by (23)

    • bioMCS 2.0: A distributed, energy-aware fog-based framework for data forwarding in mobile crowdsensing

      2021, Pervasive and Mobile Computing
      Citation Excerpt :

      Smart applications need to sense the physical environment in real-time and process the generated data to inform decision making. In recent times, mobile crowdsensing (MCS) [1] has emerged as an enabler of cost-effective sensing infrastructure. It allows citizens possessing smart devices like smartphones, tablets, wearables, etc. to collect rich sensory information about the surroundings.

    • Performance evaluation of hybrid crowdsensing systems with stateful CrowdSenSim 2.0 simulator

      2020, Computer Communications
      Citation Excerpt :

      Users might contribute freely to a sensing campaign or can be recruited through specific policies. The objectives of MCS research span over a number of areas, including quality and coverage of information over an area of interest, user recruitment, and incentive mechanisms [1,13]. Thus, such research areas typically propose optimization frameworks or algorithms that are evaluated standalone and often leave apart important components that impact on the correct modeling.

    • PS-Sim: A framework for scalable data simulation and incentivization in participatory sensing-based smart city applications

      2019, Pervasive and Mobile Computing
      Citation Excerpt :

      During recent times, overwhelming development in the smartphone technology and its ubiquitous nature have lead to rolling out of a plethora of mobile crowd-sensing (MCS) applications [1], which is considered to be one of the major driving forces for the realization of the smart city.

    View all citing articles on Scopus
    View full text