Backpressure meets taxes: Faithful data collection in stochastic mobile phone sensing systems

S. (Shusen) Yang, U. (Usman) Adeel, J. (Julie) McCann

    Research output: Contribution to journalConference article

    122 Downloads (Pure)

    Abstract

    The use of sensor-enabled smart phones is considered to be a promising solution to large-scale urban data collection. In current approaches to mobile phone sensing systems (MPSS), phones directly transmit their sensor readings through cellular radios to the server. However, this simple solution suffers from not only significant costs in terms of energy and mobile data usage, but also produces heavy traffic loads on bandwidth-limited cellular networks. To address this issue, this paper investigates cost-effective data collection solutions for MPSS using hybrid cellular and opportunistic short-range communications. We first develop an adaptive and distribute algorithm OptMPSS to maximize phone user financial rewards accounting for their costs across the MPSS. To incentivize phone users to participate, while not subverting the behavior of OptMPSS, we then propose BMT, the first algorithm that merges stochastic Lyapunov optimization with mechanism design theory. We show that our proven incentive compatible approaches achieve an asymptotically optimal gross profit for all phone users. Experiments with Android phones and trace-driven simulations verify our theoretical analysis and demonstrate that our approach manages to improve the system performance significantly (around 100%) while confirming that our system achieves incentive compatibility, individual rationality, and server profitability.
    Original languageEnglish
    JournalProceedings - IEEE INFOCOM
    DOIs
    Publication statusPublished - 24 Aug 2015
    EventIEEE 2015 Conference on Computer Communications (INFOCOM) - Kowloon, Hong Kong, China
    Duration: 26 Apr 20151 May 2015

    Bibliographical note

    Author can archive post-print. For full details see http://www.sherpa.ac.uk/romeo/issn/2327-4662/ [Accessed 05/04/2018]

    Fingerprint Dive into the research topics of 'Backpressure meets taxes: Faithful data collection in stochastic mobile phone sensing systems'. Together they form a unique fingerprint.

  • Profiles

    Cite this