Siren: Contextaware Computing for Firefighting

1
Siren Context-aware Computing for Firefighting
Xiaodong Jiang, Nicholas Chen, Jason Hong, Kevin
Wang University of California, Berkeley Leila
Takayama James
Landay Stanford University
University of Washington
2
Motivation

• Fires cause great harm
• 1,755,000 fires in the United States in 1998
• 4000 deaths, 25000 injuries, 100 firefighter
  deaths / year
• Information intensive task
• Firefighting is making a lot of decisions on
  very little information
• Convergence of sensing and communication

3
Research Overview

• Goal
• Support interaction between firefighters working
  inside urban structures
• Approach User-centered Design

Design Principles 4 months of field study
Prototype Solution Off-the-shelf technology
Evaluation Quick feedback
4
Outline

• Field studies results
• From field to design
• Siren architecture
• Siren messaging application
• Evaluation
• Conclusion Future work

5
Field Studies

• Method
• 4 months, 3 depts
• Interviews inwork environment
• 1 field exercise
• 2 emergency calls
• Participants
• 1 assistant chief
• 4 battalion chiefs
• 2 captains, 2 engineers
• 5 firefighters

6
Fieldwork Results

• Firefighters have an incomplete picture
• Situational assessment is done by sending
  advanced team into fires
• Individual firefighter confined to a limited
  space
• Current communication systems are inadequate
• Require explicit control
• Broadcast and ephemeral
• Detached from information gathering
• Firefighters operate in harsh environments
• Inherently unreliable nature of any infrastructure

7
Communication
There is a lot of noise on the fire ground.
Youre inside the fire is burning it makes
noise theres breaking glass theres chain saws
above your head where theyre cutting a hole in
the roof theres other rigs coming in with
sirens blaring lots of radio traffic everybody
trying to radio at the same time.
8
Need for Tacit Communication

• Context-dependent sharing of mission-critical
  information
• See through the eyes of fellow firefighters
• Opportunistic, spontaneous, persistent
• When communication happens is context dependent
• Often triggered by external events, without human
  initiation
• Can be replayed and shared

9
Tacit Communication An Example
Abandon
10
From Field to Design

• Tacit communication
• Ad-hoc P2P
• Delayed multi-hop
• Context dependence
• Robustness multiple levels of redundancy
• Path redundancy
• Storage redundancy
• Version redundancy
• Feedback redundancy

11
Siren Architecture

• Adapted Context Fabric (Hong et. al.) for a P2P
  network of Wi-Fi-enabled Pocket PCs

12
Storage Manager

• An InfoSpace consists of tuples
• Each firefighter owns an InfoSpace containing
  data about himself, other peers and the
  environment
• Tags to specify expiration
• InfoSpace snapshot
• A view maintained by each device of entities on
  the Siren network
• Snapshot taken by aggregating sensor readings
  over a given time window
• InfoSpace operators (insert, retrieve) unify
  storage and communication

13
Multi-hop Communication Manager
Message id . Source . Time_Created Semantic
Slot 1 Semantic Slot 2 Semantic Slot 3
Type location Value 525 Soda Hall Confidence
70 Expiration

• Message format
• XML over HTTP
• Semantic slots
• Time-based message queuing routing
• One queue for each source
• Queue is sorted by time
• Keep forwarding to all neighbors until all
  semantic slots expire
• Version redundancy
• Not overwrite older messages as newer versions
  arrive why?

14
Context Rule Engine

• Take InfoSpace snapshots as input, generate user
  alerts
• Production system consisting of if-then rules
• Currently supports five types of alerts
• dangerous place
• danger to oneself
• next to dangerous place
• others in danger
• instructions

IF (firefighter F IN room A) AND (surrounding
temperature gt 1500F) THEN (generate_alert
(firefighter F1 in danger)) AND
(generate_alert (room A is a dangerous place))
15
Siren-based Messaging Application for Firefighters

• Running on Wi-Fi-enabled Pocket PCs
• A set of context rules
• A map-based user interface
• Use of Berkeley motes
• Location beacon
• Mobile sensing
• Environmental sensing

16
Video
17
Evaluation

• Used by a group of four firefighters in their
  fire house
• Positives
• Communication redundancy for large complex urban
  structures (hospitals, high rises, warehouses)
• Location tracking enhance firefighter safety
• Like UI with no manual input
• Concerns
• Ruggedness of technology
• Risks of new technology
• Deployment issues

18
Ruggedness of Technology
Its got to be tough youve got to be able to
drop it its got to be able to take extreme
heat its got to be able to get wet.
19
Risks of New Technologies

• Thermal Imaging Cameras
• PASS System

20
Deployment Issues

• Tighter integration with existing tools (use in
  conjunction with thermal imager to narrow search
  areas)
• Sensor deployment
• Most effective tools for us are always those we
  can directly access and maintain

21
Conclusion Future Work

• Conclusion
• User-centered design of Siren
• Support tacit communication needs
• Siren messaging application evaluation
• Future Work
• Dealing with uncertainty in sensor data
• Multimodal interaction techniques
• Deployment strategy

22
Thanks to Berkeley Fire Dept El Cerrito Fire
Dept Alameda Fire Dept NSF ITR,
CITRIS Lawrence Leung

• Xiaodong Jiang
• Nicholas Chen
• Jason I. Hong
• Kevin Wang
• Leila A. Takayama
• James A. Landay
• http//guir.berkeley.edu/siren