Workplace Applications of Sensor Networks

1
Workplace Applications of Sensor Networks

• Presentation by Nick Zuiderveld
• CS 510 Sensor Networks
• Professor Nirupama Bulusu

2
Introduction

• Buildings currently have monitoring equipment,
  but must be checked manually
• Sensor networks can make life easier
• Though power and networking is abundant, we still
  need low-power, self-configuring devices
• Additional wiring costs too expensive

3
Hardware for Workplace Nets

• Requirements
• Must be low-power devices, though the available
  power and network can be leveraged
• Workplace-related sensing equipment
• Human interaction with sensor network
• Interfacing sensor network with workplace networks

4
Sensor Nodes

• Mica-2
• The most common platform
• Optimizes power, cost, size
• Intel Mote (Imote)
• Increased processing capacity

Mica2, Mica2 dot Rene Motes
5
Comparison of Hardware Platforms

6
Display Nodes

• Simple interactions with network

Button Box
LCD Display Node
7
Handheld Nodes

• Laptops PDAs
• Can easily interact with sensor network

Canby Compact Flash Mote
8
Gateway Nodes

• Capable of bridging communication between sensor
  nodes and wired networks
• 400MHz proc
• gt10M RAM, nG Storage
• Interfaces with Mica2 802.11

Xscale Gateway Node
9
Conference Room Application

• Problem Overview
• In many modern offices, multiple cubicles make it
  difficult to hold impromptu meetings
• Conference rooms available for meetings, but
  often are reserved ahead of time
• Common for meetings to be shorter than
  reservation time, or meetings to be cancelled
  without cancelling the reservation
• Exhaustive search can locate empty room
• Application Idea
• Conference rooms already equipped with motion
  sensors
• Can use these sensors to determine room usage
  status

10
Architecture and Operation

• System consists of network of sensors deployed in
  and around conference rooms
• In-room sensors connected to motion sensors
• Gateway node receives sensor data
• Aggregates and stores data for access via web
• Users can simply check web via desktop or PDA to
  immediately find empty room

11
Conference Room Application

• Live Conference Room Occupancy
• (b) PDA Showing Room Status
• (c) Occupancy History Data

12
Architecture and Operation

• Motion detector data can also be compiled and
  used for future analysis
• Possible to analyze building usage patterns
• Can be used to design new buildings
• Same infrastructure can be reused to gather
  temperature and battery usage by each node
• Room reservation status available at each room
• Status nodes at room entrance can indicate
  current/future reservation status

13
Conference Room Application

(a) Motion Sensor Node (b) Reservation Status
Indicator
14
Application Challenges

• Power the most significant challenge
• Battery-powered devices used
• Analysis showed that eliminating power/network
  wiring at the sensors had significant cost
  savings
• Whenever possible, could leverage power outlets
• Synchronization allows nodes to sleep and still
  communicate at intervals
• Building maintenance already replaces such things
  as light bulbs periodically, so possible to
  provide node battery replacement every 6 months
  to a year

15
Communication Protocols

• Sensing
• Goal Deliver occupancy status of conference room
• Many-to-one delivery tree to sink node (web
  server)
• End-to-end reliability metric used for path
  selection
• Packet information
• Node number (1 byte), room number (2 bytes),
  occupancy (1 bit)
• Nodes along path append own node number
  occupancy status
• For each packet, web server obtains room number
  to node number mapping for original node, as well
  as occupancy status for several rooms

16
Communication Protocols

• Actuation
• Provides reservation info. to nodes outside each
  room
• Reservations change infrequently, so rarely push
  data
• Server tracks topology of data delivery tree
• Uses list of forwarding nodes in each occupancy
  status packet
• Reservation status packets can simply reverse
  path to transmit
• Packet Information
• Reservation status bitmap, timestamp for start
  of bitmap
• Every half hour, time synchronization performed
• Latency of a few seconds is sufficient

17
Communication Protocols

• Leverage Ethernet and 802.11 connectivity
• Sensor Network packets can be tunneled across
  IP-based infrastructure
• Stargate nodes deployed in several conference
  rooms on each floor and one at the sink node
• Each Stargate attached to mote to receive SN
  packets
• Stargates can be used for reliable path discovery
• Allows for more efficient communication,
  increased network reliability, and decreased
  energy expenditure for individual nodes

18
Follow-Me Application

• Problem
• Navigating an unknown place can be difficult
• Follow-Me an active visitor guidance system
• Sensor nodes deployed along walls at doors
• Nodes blink lights to indicate a path
• Breadcrumb Trail
• Key innovation
• Deployment-order approach to topology
  configuration

19
Hardware

• Button Boxes are the sensor nodes
• Deployment guidelines
• Should be one node at each office doorway
• Distance between two adjacent nodes should not be
  too large (sometimes placed in hallways)
• Touch-screen display at building entrance allows
  visitors to select destination

20
Follow-Me Application

Deployment Example
21
Protocols

• Traditional routing protocols specialize in path
  finding, problem is different for humans
• Needs to be the shortest path for a human to
  follow
• Logical topology the main technical challenge
• Deployment Order
• Algorithm to capture logical topology
• Present when network first configured
• Allows construction of complex topologies with
  minimal human interaction

22
Follow-Me Application

Radio Connectivity vs. Logical Topology
23
Deployment Order

• Methodology Concerns
• System must be easy to deploy and have low cost
• System must work well with building-like
  topologies
• Long, linear segments, parallel hallways,
  moderate density
• Typically, sensor nodes deployed sequentially
• If two nodes switched on one after the other in a
  short time, assume they are closest neighbors
• Linear path created, which can be used for
  visitor guidance
• Special problem of intersections needs to be
  addressed

24
Linear Paths

• Active State after node switched on, sends
  request for neighbor
• Receptive Will reply to connection request
  packets and establish links
• Passive Not involved in link operations. State
  for normal operation.

25
Intersection Handling

• Special case when nodes have more than two
  neighbors
• Button on each sensor node used to toggle node
  states

26
Intersection Handling

Intersection Example
27
Reconfiguration/Maintenance

• Fixing 1-hop node failures
• When node detects failed neighbor, it will try to
  skip neighbor and link to neighbors neighbors
  directly
• A link fix packet is broadcast containing
  failed ID
• Only failed nodes neighbors will respond with
  own IDs
• New links will be established, old links removed
• Accidentally changing network configuration
• Occurs when someone presses buttons on box
• Buttons need to be locked after configuration
  process
• Network managers may use a key to change config.

28
Routing

• Minimum-distance routing algorithm
• Uses logical topology to determine best path
• Flooding used to find forward paths,
  gradient-style routing for reverse paths
• Visitor arrives at lobby, selects destination
  from touch screen
• Destination node gathers routes / selects best
  path
• Metric Physical distance traveled
• (Nodes also support radio connectivity routing)

29
User Interaction

• Keep interface to system simple
• Synchronized blinking patterns across the network
  show path to visitor
• Moving light dots or lines communicate path /
  direction
• Interesting problem
• How to guide multiple visitors at the same time?
• Possible Solution Different colors / blinking
  patterns
• Also could limit indicators to those visible to
  visitor

30
Other Applications

• Voting App
• Provides feedback from an audience to a speaker
• Audience members given button box
• Votes sent to single destination, no vote
  aggregation
• ISI Security Application
• Balances privacy and security using different
  sensors
• Video sensor in lobby / motion detectors in
  hallways
• In case of theft, can map path of individual from
  crime scene back to lobby (using timestamps)
• Labscape
• Developed to improve work flow in cell biology
  lab
• Provides workflow automation in experiment
  preparation and execution

31
Reusable Tools / Techniques

• Routing
• All of the workplace applications explored
  require multi-hop wireless routing, despite
  available Ethernet
• Leveraging Existing Infrastructure
• e.g. Overlay routing to improve reliability
  (Conf.Room)
• Exploiting Simple External Interactions
• e.g. Sensing flashing LED on existing motion
  detector easier than detecting if person is in
  the room
• In Follow-Me, using deployment order to collect
  logical information easier than ground-up
  configuration

32
Conclusions

• Differences from outdoor applications
• Hardware/Software emphasizes user interaction
• Leverage existing infrastructure
• Techniques developed good for many applications
• Routing algorithms, overlay networks, easy
  configuration make sensor networks applicable to
  workplace applications
• Key Challenge
• Demonstrate that workplace applications can be
  beneficial through ease of use, ease of
  management, and productivity

33
Thanks

34
Sources

• Workplace Applications of Sensor Networks
• W. Steven Connor1, John Heidemann2, Lakshman
  Krishnamurthy1, Xi Wang2, Mark Yarvis1
• 1 Intel Research and Development
• 2 University of Southern California, Information
  Sciences Institute