MetroSense Project: People-Centric Sensing at Scale Shane B. Eisenman

1
MetroSense Project People-Centric Sensing at
ScaleShane B. Eisenman, Nicholas D. Lane,
Emiliano Miluzzo, Ronald A. Peterson,
Gahng-Seop Ahn and Andrew T. CampbellElectrica
l Engineering, Columbia University Computer
Science, Dartmouth College
Mobile People-centric Applications
Campus Area Sensor Network
Motivation
Looking forward 10-20 years we envision Internet
scale sensing where the majority of the traffic
on the network is sensor data and the majority of
applications used every day by the general
populace integrates sensing and actuation in some
form. Sensing will be people-centric. We believe
the gap between the state of the art and our
future vision can be bridged through the
development of a new wireless sensor edge for the
Internet. To be truly impactful on the daily
lives of lay people, new applications must by
about people and the way they interact with their
surroundings. A focus on people-centric sensing
shifts design challenges no centralized control,
platform heterogeneity, batteries can be
recharged. Mobility presents both challenges and
unique opportunities.
BikeNet Sensing
Dual focus quantifying the cyclist experience
and facilitating in situ and post facto
information sharing between cyclists
SAP implementation Aruba AP-70 Moteiv Tmote
Invent
Network symbiosis we are deploying SAP on the
back of existing WiFi APs to reduce the time and
cost of system deployment. Each AP will continue
to fulfill its normal role while MetroSense taps
its spare resources.
SkiScape Sensing
Dual focus monitor ski resort trail conditions
and collect and store skier speed and location
statistics
MetroSense Architecture
Opportunistic Delegation Model
Opportunistic Operations Illustrated

• Goal is to leverage mobility when direct sensing
  from SAP is not possible. Scenario application
  requires sensed modality ß from space during
  t1, t2
• SAP delegates sensing responsibility ß for a
  given time to a mobile sensor (MS) in its radio
  range. The MS senses the target, if possible, and
  carries the result back to the SAP, if possible.

Server Tier
comms ground-truth sensing
opportunistic tasking
SAP Tier

• Direct vs. Indirect delegation
• If MSs available to a SAP are not fit to be
  direct delegates, a SAP may authorize a MS to
  delegate to another MS on the SAPs behalf
  (indirect delegation).
• Examples indirect tasking, virtual sensing range
  extension, indirect collection data muling
• Design challenges
• Limited rendezvous time
• Sensor selection challenging

opportunistic sensing
opportunistic collection
Sensor Tier
limited peering
MetroSense is a three-tiered physical
architecture Server Tier, Sensor Access Point
(SAP) Tier and Sensor Tier
Mobility gives rise to opportunistic sensor
networking opportunistic tasking, opportunistic
sensing, and opportunistic collection.
MetroSense comprises three main software
component sets Common Components, Core
Components and Resource Components
With support from ISTS, Intel and Nokia. Shane
Eisenman is supported by the Army Research Office
(ARO) under Award W911NF-04-1-0311 More
information on the MetroSense Project, including
publications, technical reports, and source code
from http//metrosense.cs.dartmouth.edu/.