Review on Various Architectural Models in Mobile Crowdsensing
Mobile Crowdsensing : Current State and Future Challenges
description
Transcript of Mobile Crowdsensing : Current State and Future Challenges
Mobile Crowdsensing: Current State andFuture Challenges
Raghu K. Ganti, Fan Ye, and Hui Lei
IBM T. J. Watson Research Center, Hawthorne, NY
Presented by:Sheekha Khetan
Introduction
• Mobile Crowdsensing - individuals with sensing and computing devices collectively share information to measure and map phenomena of common interest.
• Devices - smartphones, music players, and in-vehicle sensing devices
Introduction
Phen
omen
a
Individualmovement patterns, modes of
transportation ,and activities .
Communitypollution (air/noise) levels in a neighborhood, real-time traffic
patterns, pot holes on roads, road closures and transit timings.
Introduction
• Community sensing is popularly called participatory sensing or opportunistic sensing
, Participatory sensing - individuals are actively involved in contributing sensor data
, Opportunistic sensing - autonomous and user involvement is minimal
• Research Challenges, Localized analytics, Resource limitations, Privacy, Aggregate analytics, Architecture
MOBILE CROWDSENSING APPLICATIONS
MCS Applications
Categories
Environmental
Natural Environment
Infrastructure Public Infrastructure
Social Personal information
MCS: UNIQUE CHARACTERISTICS
• Multi-modality sensing capabilities• Deployed in the field• The dynamic conditions in the
collection of mobile devices• Privacy • Energy• Cost• Efforts
LOCALIZED ANALYTICS
LOCALIZED ANALYTICS
• Less amount and appropriately summarized data
• Reduces the amount of processing that the backend has to perform
LOCALIZED ANALYTICS – Challenges
• Some applications may be delay sensitive
• Data mediation, such as filtering of outliers, elimination of noise, or makeup for data gaps.
• Context inference– Transportation mode– Kinetic modes of humans– Social settings
LOCALIZED ANALYTICS – Challenges
• Due to highly dynamic nature, modeling and predicting the energy, bandwidth requirements to accomplish a particular task becomes much more difficult
• Identifying and scheduling sensing and communication jobs among them under resource constraints becomes more difficult as well
• Interdependencies between various types of sensory data due to multi-modality sensing capability
• The existence of multiple concurrent applications that require data of different types also complicates resource allocation
Questions that need to be answered
• How do multiple applications on the same device utilize energy, bandwidth, and computation resources without significantly affecting the data quality of each other?
• How does scheduling of sensing tasks occur across multiple devices with diverse sensing capabilities and availabilities (which can change dynamically)?
Privacy
• Potentially collect sensitive sensor data pertaining to individuals– the routes they take in daily commutes– their home and work locations
Approaches to privacy
• Anonymization – One of the approaches – Still not reliable
• Cryptographic techniquesRequire the generation and maintenance of multiple keys
• Perturbation based approach– adds noise to sensor data before sharing
AGGREGATE ANALYTICS
• These analytics detect patterns in the sensor data from large number of mobile devices– coordinate the traffic lights– public works maintenance
ARCHITECTURE
• Existing MCS applications take an ”application silo” approach where each application is built from scratch without any common component even though they face many common challenges. Such an architecture hinders the development of new MCS applications