88 Chapter 5   Stationary Sensors

Figure 5.1  3-plane chart for Smart Network Infrastructure for Fire Resilience (SNIFFR).

to data processing computers; and (4) optimal architectural layout of processing computers given constraints imposed by fire
size and location. In order to decipher the information acquired from the sensor networks, the Model and Decision Theory
thrusts and Super-Real-Time test bed will enable development of (5) computer models that interpret the sensor signals and
convert them for use in predictive tools; (6) computer inverse models identifying the real-time fire state and super real-time
forecasts; and (7) framework for uncertainty estimates inherent in data and predictions in sensor and computer models. Ulti-
mately, deciphered information must be delivered to people or response subsystems. To this end, the Communications and
Decision thrusts and Human Behavior test bed will enable development of (8) protocols/software to extract only necessary
information for guidance; and (9) communication protocols/software to transmit and display critical information.
The resulting SNIFFR system, outlined in Figure 5.2, could prevent an incipient fire, and in the case of a developing
fire, assist in fire fighting efforts by forecasting and providing safe egress routes and available time to occupants. The
fire-shielded sensors will wirelessly communicate the measurements to a central processing unit, where they will drive
the real-time analysis to forecast the future condition of the fire, structure, and tenability. This forecast, along with critical
­decision-making information, will be communicated back to the occupants and fire fighters in “super-real-time.” This
forecast will be dynamically updated as new sensor data become available.

5.6.2.4  Fundamental Research Plan

Sensors  Fire sensors and repurposed fire sensors can be classified as thermal (temperature and heat flux), chemical (CO,
hydrocarbons, gases, O2%, photoelectric or ionization smoke detectors), optical (visible or infrared mounted in buildings
or unmanned aerial systems), and structural (stain gauges, deflection cameras, and plumb bobs) [1–9]. All of these sensors
provide useful but different information about the fire environment and have different cost, power, and data processing
requirements. Optical techniques are limited to the field of view and thermal and chemical sensors depend on the gas flow.

SFF15_CH05_077_096.indd 88 4/23/15 11:43 AM

 5.6  Perceived Priorities for Research 89

Figure 5.2  SNIFFR system framework applied to WUI. Items 1–9 highlight enabling insights described in 5.6.2.3.

Thus different types of sensors are required to “sniff” a fire. Further, the fire sensors must often be wireless, networked [10,
11], and fire-hardened [12, 13]. If a fire starts, an effective sensor network must continuously (1) monitor the environmental
conditions, (2) identify and track occupants and fire fighters, and (3) determine the state of the infrastructure under fire
insult (e.g., building, transportation), and (4) monitor the interdependent health of the environment, people, and structures.
The raw data from sensors also need to give a spatial and temporal context before forwarding for analysis. Data-driven
predictive fire sub-models provide an example of this approach.
Based on SNIFFR system requirements, it is apparent that a large numbers of sensors would positively impact the
uncertainty propagation barrier for accurate information; yet vast quantities of data are inherent to this approach, creating
the expansive data barrier for timely information. The cost of continuously monitoring all the sensors all the time will be
addressed by developing smart sensing methodologies, which could involve some of the low or zero power sensors along
with the environmental sensors like temperature and humidity in the safe-state, and by activating the rest of the sensors
if a fire-state develops. Decision theory will be applied using value of information (VOI) models to determine where and
which sensor inputs have the most impact.
Networks  Sensors, network management centers, and base stations that are located both on site and in hardware and
software of the underlying Wireless Sensor Network (WSN) will be jointly designed to guarantee power-optimal performance
[14,15] of the mobile nodes and resiliency of the overall network [16]. Distributed detection and estimation systems will
address the fragmented network and uncertainty propagation barriers, enabling accurate and swift fire detection, while also
resolving information for safe evacuation and fire fighting. The fusion of measurements collected by sensors that are of
different types and placed at distinct locations will be accomplished using a model-based approach [17] that encapsulates

SFF15_CH05_077_096.indd 89 4/23/15 11:43 AM

 90 Chapter 5   Stationary Sensors

the underlying physics of fire growth and propagation via distributed filtering algorithms that are scalable and dependable in
the presence of intermittent [18] and fragmented networks [19, 20]. The implementation of these functionalities will follow
the Keppler workflow system [21] of real-time data processing. Decision theory activities applied to the networks include
characterizing the statistical properties of the WSN for incorporation into the workflow system.

Models  Computer-based models have been applied to fire protection engineering problems and have emerged over the
past three decades as powerful tools for basic understanding, system design, and forensic analysis. Models are routinely
applied to simulate fire spread and smoke transport; the evacuation of building occupants in response to a fire; and the
structural resilience of buildings to fire loads. Current fire models are limited in scope because of the large uncertainties
associated with both the accuracy of physical models and uncertainties in many of the input parameters to the fire problem.
A promising approach to overcome the limitations found in numerical simulations of fires is data assimilation (DA).
DA consists of integrating computer-based fire modeling with sensor technology [22–32]. DA provides an optimized
approach to overcome the barriers associated with fire complexity and human behavior barriers (i.e., complex multi-
physics systems influenced by a large number of unknown or partially-known input parameters). While still original in
the field of fire, DA is an established approach in several scientific areas such as numerical weather predictions. The idea
of data assimilation has been explored recently by several fire research groups, both for wildland and for building fire
applications. Data-driven fast physics fire models will be developed and evaluated to (1) forecast fire spread and fire plume
dynamics (Forest test bed); (2) detect incipient fires, optimize the response to a growing fire in terms of providing safe
egress, assisting first responders, activating building systems, for example, fire suppression systems and HVAC systems
(Building and Transportation test beds); and (3) monitor the health of a structure and/or an engineering system (Building
and Transportation test beds) supported by fire fundamental experiments. To support the use of these models for decision
making, parameter inference tools will be used to produce stochastic models to assimilate the sensor data. Compact
mathematical representations to model under-sampled probability distributions and computationally efficient models for
uncertainty propagation will be developed.

Communications  The effectiveness of fire warnings transmitted to occupants is an important area of research, as people
do not respond well to non-voice signals (e.g., alarms, sirens, and bells). Further, decisions about evacuating from fire-
focused warnings may involve delays due to searches for additional information [32]. Brennan [33] underscored that
fire fatalities tend to include a disproportionate number of elderly and disabled, that is, high-risk demographic groups
in society. To address the human behavior barrier, the proposed research program will use experimental settings along
with field tests of our sensor communications technologies at the Maryland Fire and Rescue Institute (MFRI) or Purdue’s
Homeland Security Institute (HIS) to understand the best way to communicate information, not just alarm, to residents
across a spectrum of ages and abilities. Collaboration will add value to the occupant focused communications and incident
commander focused decision theory thrusts.

Decision Theory  Decision theory will be applied to develop a framework that (1) identifies the information used by
decision makers (DMs) in the system test bed environments, (2) characterizes the sources and properties of uncertainty
in sensor-data, models, and networks, and (3) communicates information relevant to the alternative strategies that a DM
is considering [34]. The decision theory approach directly addresses the human behavior barrier while guiding overall
SNIFFR development. The focus will be on DM incident commanders (ICs) engaged in fighting forest, building, or
transportation fires. These ICs are responsible for deploying resources to mitigate the fire and evacuating civilians from
the fire threat. Research on naturalistic decision making (NDM) and decision analysis will be used, considering decision
processes under stress and use of sensor data and other information to increase situation awareness and inform options
analysis [35–36]. Expert incident commanders from the three test bed environments will be surveyed to encapsulate their
information gathering, assessment, and decision-making strategies into mathematical models, through which the type of
information that most affects their decisions can be identified along with the sensitivity of decisions to this information.
In the SNIFFR system, sensors, sensor networks, models, and communication networks provide information to increase
situation awareness. Since sensor signals do not perfectly map to the state of the environment, there is noise in networked
signals, and models contain bias and model parameter uncertainty, one can consider that the nominal outputs are enveloped
in uncertainty. To address this, the overarching objective of the decision theory thrust is to best synthesize the data, network
characteristics, and models to reliably communicate the system state given the state of uncertainty across all planes. The
decision theory has been integrated throughout the thrust area and test bed discussions.

SFF15_CH05_077_096.indd 90 4/23/15 11:43 AM

 5.6  Perceived Priorities for Research 91

5.6.2.5  Technology Development

Hot Sensor  The SNIFFR sensors will be exposed to laboratory controlled conditions simulating fire sources (e.g.,
combustion gases, soot particles, flame emission) and nuisance sources (e.g., steam, ambient light) to determine sensor
response (i.e., what the sensor “sees”) with a focus on discrimination between true fire and nuisance signatures. This test
bed will explore detector innovations, imaging innovations, and Unmanned Aerial Systems innovations. Critical items such
as network continuity, degradation of sensor performance, and ultimately sensor failure will be monitored and can serve
as additional key data points for the SNIFFR. The observed fire conditions will be simulated and sensor responses will be
correlated with those of a sensor model for calibration and improvement of the latter.
Rugged Network  Accurate and reliable processing of sensor data must be provided (sometimes under extremely
tight energy consumption constraints, depending on the application) for the rugged sensor networking required for the
SNIFFR. Signal processing design methodologies and digital subsystem architectures for system-on-chip integration will
be explored using the variety of sensors employed in the rugged network test beds. Critical challenges in jointly providing
ultra-low power operation, secure wireless communication, accurate sensor signal processing, and efficient integration
with heterogeneous sensing devices will be addressed. Extensive previous work by the investigators in developing energy
efficient application-specific integrated circuits and signal processing techniques for surveillance-oriented sensor networks
[14, 38] provide a solid foundation for these efforts. Decision theory efforts will focus on developing VOI models to
determine which networks are most important and require the highest reliability and redundancy.
Super-Real-Time  Prototype software cyber-infrastructure will be developed for this test bed that integrates data-driven
estimation and prediction models with a sensing system that will “sniff” environmental conditions and selectively disseminate
fire predictions to incident commanders and fire fighters in the field. Data assimilation and parameter estimation techniques
will be used to address the need to reduce large dimensional sensor data to a parametric lower dimension suitable for rapid
fire spread analysis, interpretation, and alert purposes, allowing for a more efficient (mobile) communication of events for
decision-making and crisis management. Large-scale simulations will be devised, refined, and evaluated (on the massively
parallel platforms available to the proposers when needed). The simulations include consideration of the physical geometry
(e.g., in wildfires terrain, weather, vegetation) as well as parameters articulated from historical, observational, and sensory
data. The Kepler scientific workflow system will be used to coordinate the execution of real-time data processing and fire
modeling tools on distributed computing environments. Decision theory will be applied to identify time scales required for
meaningful data assimilation (sensor, model, networks) so as to ultimately affect decision makers. Extensive testing of the
concept of the super-real-time sensing and data-driven predictive fire models will be conducted in a virtual test bed [10,
39]. A data communication layer with links to archives, experimental data, and heterogeneous sensor data networks will
be developed. The test bed would integrate portals for dissemination of data to different end users that include scientists,
first responders, and public notification of user-defined real-time alerts via various receivers and Web 2.0–based public
systems. This test bed will address the need for expansive data and sensor driven models, leading to predictive modeling
assistance for decision makers.
Human Behavior  Social science has shown that the response to sensor information and fire warnings varies dramatically
with levels of social capital, demographic conditions, and historical experiences with similar warning systems [40–45]. The
investigators will test the responses of both first responders and residents to our SNIFFR system through multiple approaches,
including qualitative process tracing of past disasters (to better understand what factors are relevant), agent based modeling
(to predict crowd and group behaviors in safe virtual environments), and field experiments of the technology. Evacuation
rates and speed from residents along with response time and appropriate tactics from first responders will be used as core
performance metrics. Once these metrics surpass those of current technologies, the SNIFFR system will be ready.

5.6.2.6  SNIFFR System Development

The SNIFFR addresses important fire scenarios as shown in Figure 5.1. In forest fires (time scale of days), the SNIFFR
provides information to adapt for optimal resource allocation as conditions change. In complex buildings (time scale of tens
of minutes), the SNIFFR locates hazardous conditions and directs occupants along safe evacuation routes. In transportation
fires (time scale of minutes), the SNIFFR diagnoses the fire source and provides information to response systems for restoring
the system to a safe state. Decision theory will be employed using IC expertise to construct decision-tree/influence diagram
models for these scenarios. Further, sensitivity analysis on decision models will be performed to identify the most critical

SFF15_CH05_077_096.indd 91 4/23/15 11:43 AM

 92 Chapter 5   Stationary Sensors

information flows. Finally, forecasting post-processing and visualization tools to best communicate ensemble predictions
will be developed.
Forests  The San Diego backcountry instrumented with the densest county-wide sensor array available in the world will
serve as a test bed to analyze the SNIFFR effectiveness in protecting WUI. Archived weather data from this network will
be compared with corresponding WUI fire simulation using the Kepler system (available at the collaborating San Diego
Supercomputer Center and described in the Super-Real-Time test bed). Further, home ignition data from IBHS and NIST
will be utilized to assess the realism and performance of the ignition behavior of the WUI fire simulations. The SNIFFR
system must also address how the information will be interpreted and passed to incident commanders and individual
emergency responders. Guidance for emergency responders will be addressed based upon the decision theory activities
mentioned previously in this section.
Buildings  The efficacy of the SNIFFR system to locate people, provide people with meaningful information, and propose
evacuation strategies in complex buildings will be performed through a combined physical and virtual system test bed.
Large-scale experiments would be performed at available facilities (e.g., at university or industrial partners) having mock-
ups of multi-room spaces. Tests would be conducted monitoring fire conditions to evaluate the SNIFFR system detection
algorithms in the incipient fire predictive performance in the developing fire. These fire tests would be extended to large-
scale experiments in buildings with comparable footprints as high-rise buildings (historically available when scheduled
for significant renovation or demolition). With regard to people, it will be important to understand human response to fire
in complex buildings so that people can be provided with meaningful information and guided in ways that are readily
accepted. SNIFFR communications will be evaluated in an available small-scale virtual reality (VR) “cave” where a
subject (occupant or emergency responder), with VR goggles and perhaps other interactive feedback devices, navigates
simulated game-like environments (of the full-scale building tests) with and without guidance from corresponding SNIFFR
communications. Determining effective guidance for emergency responders will be addressed based upon the decision
theory activities mentioned previously.
Transportation  The transportation test bed will consist of three components: a ground vehicle, an aircraft, and a coal mine.
The order of components corresponds to progressively larger and more complicated environments that will be used to test
progressively broader ranges of the SNIFFR technologies. To implement the ground vehicle component, a decommissioned
bus will be acquired and placed at the Maryland Fire and Rescue Institute training grounds. The SNIFFR sensing system
will be installed in the vehicle, and its ability to diagnose incipient fires associated with simulated malfunction of electrical
and mechanical components will be examined. The aircraft component will be based on a widebody aircraft mockup
located at the fire testing facility of the FAA Technical Center. Experiments will be conducted in this mockup to establish
the SNIFFR’s capability to detect in-flight fires, forecast fire growth (based on sensor input), and communicate this
information to the flight crew. The last component of the transportation test bed will be realized using the NIOSH coal
mine fire testing facility. This component is selected to mimic fire scenarios in underground transportation infrastructure
(i.e., subways and tunnels). Experiments designed to simulate accidental equipment fire, such as conveyor belt or electrical
generator fires, will be conducted. The ability of the SNIFFR to detect these fires at early stages, correctly forecast fire
development, communicate essential hazard information and tunnel environment to optimize egress will be tested and
validated. It should be noted that the last two components utilize testing facilities of the organizations, whose goals are
consistent with that of the SNIFFR. The proposed experiments will be incorporated into these organizations’ routine test
schedules to leverage their resources.

5.6.2.7  SNIFFR Impact

The SNIFFR workforce development efforts seek to build a stronger pipeline in engineering and science through inno-
vative research training, university education, and precollege activities. This pipeline is essential for our stakeholders
in general and advances in fire resilience technology in particular. The strong appeal of our exciting multidisciplinary
test beds (e.g., SRT visualization or airplane test bed) will be used to attract undergraduate and PhD research assistants.
Fire Protection Engineering undergraduate students will continue to be attracted to these research opportunities, and
returning members of the military and existing members of the fire service will also be targeted for our doctoral pro-
grams. Upon graduation, these individuals will serve as the agents of change to revolutionize the traditional fire fighting

SFF15_CH05_077_096.indd 92 4/23/15 11:43 AM

 5.7  References 93

industry. These students will intern at ERC member institutions and enrich cross-disciplinary projects. The nanoHUB
framework will be utilized to create an “international learning network” FireHUB. The rich web-based features will be
used to provide simulation tools for researchers, curricula, and activities for outreach to K-12 and community colleges,
and NSF Research Experience for Teachers (RET) resources. A Young Scholars Program would leverage FIRST Robot-
ics Competition for SNIFFR-enabled fire challenges building on the strong robotics and unmanned vehicle interests at
partner and collaborating institutions.
The Innovation Ecosystem is the structure that enables technology and information to flow between individuals, com-
panies, and institutions partnered within the ERC to foster the transformation of discovery into viable new offerings (i.e.,
innovation) for commercial economic growth [46,47]. The ERC will institute Industry and Practitioner (I&P) membership
through financial commitments (and possibly other resources) with an Industrial Advisory Board. Semi-annual workshops
with I&P members will continuously “pull” the ERC research agenda into focus while strengthening existing strong and
functioning relationships (e.g., insurance, sensor and control companies) for champion-based recruiting. The networked
sensors, insurance, and EFR sectors will be the first value chain target for ERC membership as these groups already have a
high level of research engagement. Strong university-based innovation and entrepreneurial resources are available to drive
the commercialization process at university and ERC partner institutions (e.g., MTech). Commercialization is critical to
achieve a balance between the knowledge and commercial economies required for a sustainable ERC.

5.7 References
1. Hamins A., Averill J., Bryner, N., Gann, R., Butry, healthcare applications. Journal of Medical Systems.
D., Davis, R., Amon, F., Gilman, J., Maranghides, A., 36: 93–101.
Madrzykowski, D., Manzello, S., and W. Mell. 2012. 10. Kamel Boulos, M. N., Resch, B., Crowley, D. N., Bres-
Reducing the Risk of Fire in Buildings and Commu- lin, J. G., Sohn, G., Burtner, R., Pike, W.A., Jezierski,
nities: A Strategic Roadmap to Guide and Prioritize E., and K. Y. S. Chuang. 2011. Crowdsourcing, citizen
Research. Gaithersburg, MD: National Institute of Stan- sensing and sensor web technologies for public and
dards and Technology. environmental health surveillance and crisis manage-
2. The business value of BIM for infrastructure. Smart ment: trends, OGC standards and application examples.
Market Report. http://download.autodesk.com/us/bim_ International Journal of Health Geographics. 10.
infra/Business_Value_of_BIM_for_Infrastructure_ 11. Sendra, S., Granell, E., Lloret, J., and J. J. P. C.
SMR_2012.pdf. Rodrigues. 2014. Smart collaborative mobile system
3. Anonymous. 2009. Business Value of BIM, McGraw for taking care of disabled and elderly people. Mobile
Hill Construction Smart Market Report. New York: Networks and Applications. 19: 287–302.
McGraw Hill. 12. Dekdouk, A. 2013. A mobile-based automation system
4. Roman, J. 2014. Data driven. NFPA Journal. Nov/Dec. for maintenance inspection and lifesaving support in a
5. Karter, M. J. 2010. Fire Loss in the United States smart ICT building. Communications in Computer and
During 2009. Quincy, MA: National Fire Protection Information Science. 413.
Association. 13. Nakamura, T., Kogo, K., Fujimura, J., Tsudaka, K.,
6. Ahrens, M. 2010. Home Structure Fires. Quincy, MA: Wada, T., Ohtsuki, K., and H. Okada. 2013. Develop-
National Fire Protection Association. ment of emergency rescue evacuation support system
(ERESS) in panic-type disasters: disaster detection
7. Macias, E., Suarez, A., and J. Lloret. 2013. Mobile sens-
by positioning area of terminals. Proceedings of the
ing systems. Sensors (Switzerland), 13: 17292–17321.
International Conference on Parallel Processing.
8. Christin, D., Reinhardt, A., Kanhere, S. S., and M. Hol- 931–936.
lick. 2011. A survey on privacy in mobile participatory
14. Roggen, D., Wirz, M., Tröster, G., and D. Helbing.
sensing applications. Journal of Systems and Software.
2011. Recognition of crowd behavior from mobile sen-
84: 1928–1946.
sors with pattern analysis and graph clustering methods.
9. Al Ameen, M., J. Liu, and K. Kwak. 2012. Security Networks and Heterogeneous Media. 6: 521–544.
and privacy issues in wireless sensor networks for

SFF15_CH05_077_096.indd 93 4/23/15 11:43 AM

 94 Chapter 5   Stationary Sensors

5.8  Case Study References for 5.6.4

1. Gottuk, Daniel T., et al. “Advanced Fire Detection Using 14. Shen, C., W.L. Plishker, D. Ko, S.S. Bhattacharyya, and
Multi-Signature Alarm Algorithms.” Fire Safety Journal N. Goldsman. “Energy-driven distribution of signal pro-
37.4 (2002): 381–394. cessing applications across wireless sensor networks.”
2. Zhigang, Liu, and Andrew K. Kim. “Review of recent ACM Transactions on Sensor Networks 6, no. 3 (June
developments in fire detection technologies.” Journal 2010): 32. DOI:10.1145/1754414.1754420
of Fire Protection Engineering 13.2 (2003): 129–151. 15. Ok, C., S. Lee, P. Mitra, and S. Kumara. “Distributed
3. Pohle, R., et al. “Fire detection with low power FET routing in wireless sensor networks using energy welfare
gas sensors.” Sensors and Actuators B (Chemical) 120, ­metric.” Information Sciences 180, no. 9 (2010): 1656–1670.
no. 2 (2007): 669–672. 16. Bhattacharyya, S.S., E. Deprettere, R. Leupers, and
4. Hagen, B. C., and J.A. Milke. “Use of gaseous fire sig- J. Takala. Handbook of Signal Processing Systems
natures as a mean to detect fires.” Fire Safety Journal (Springer) 2nd Edition (2013).
34, no. 1 (2000): 55–67. 17. Lee, S., S. Kumara, and N. Gautam. “Market-based
5. Yuan, Feiniu. “A fast accumulative motion orientation model predictive control for large-scale information
model based on integral image for video smoke detec- networks: completion time and value of solution.” IEEE
tion.” Patter Recognition Letters 29.7 (2008): 925–932. Transactions on Automation Science and Engineering
5, no. 4 (2008): 630–640.
6. Merino, L, F. Cabaleero, and A. ollero. “Applications of
Networked Unmanned Aerial Vehicles to Cooperative 18. Gupta, V., and N.C. Martins. “Optimal tracking control
Fire Detection using Grid-based Data Fusion Tech- across erasure communication links in the presence of
niques.” SYROCO’06, 8th International IFAC Sympo- preview.” International Journal of Robust and Nonlin-
sium on Robot Control. 2006. ear Control 19, no. 16 (November 2009): 1837–1850.
7. Blunck, D.L., and J.P. Gore. “A Study of Narrowband 19. Park, S., and N.C. Martins. “An Augmented Observer
Radiation Intensity Measurements from Subsonic for the Distributed Estimation Problems for LTI Sys-
Exhaust Plumes.” AIAA Journal, 2010. tems.” Proceedings of the American Control Confer-
ence. 2012.
8. Trouvé, Arnaud — organizer and point of contact.
Cyber-Based Combustion Science. 20. Lipsa, G., and N.C. Martins. “Remote State Estimation
With Communication Costs for First-Order LTI Sys-
9. Sivathanu, Y.R., R. Joseph, J. Lim, Y. Zheng, and J.P.
tems.” IEEE Transactions on Automatic Control 56,
Gore. “Rapid Scanning Infrared / Near Infrared Spec-
no. 9 (2011).
trometer: Phase 2 SBIR Final Report.” GCR, NIST,
99–777. 21. Ludäscher, B., et al. “Scientific workflow management
and the Kepler system.” Concurrency Computation
10. Loh, K. J., J. Kim, J.P. Lymnch, N.W.S. Kam, and
Practice and Experience 18, no. 10 (2006): 1038–1065.
N.A. Kotov. “Multifunctional layer-by-layer carbon
­nanotube-polyelectrolyte thin films for strain and cor- 22. Mandel, J., Beezely, and J.D. “Morphing Ensemble
rosion sensing.” Smart Materials and Structures 16, Kalman filter.” Tellus 60A (2007): 131–140.
no. 2 (2007): 429–438. 23. Mandel, J., et al. “A wildland fire model with data
11. Trouvé, A., D. C. Haworth, J. H. Miller, L. K. Su, and assimilation.” Mathematics and Computers in Simula-
A. Violi, Cyber-Based Combustion Science Report on tion 79 (2008): 584–606.
the NSF Workshop on Cyber-Based Combustion Sci- 24. Leblanc, M., and A. Trouvé. “Inverse zone modeling of
ence, April 19–20, 2006, National Science Foundation enclosure fire dynamics.” Proc. 6th U.S. Combustion
Headquarters, Arlington VA. Sponsored by the National Meeting. Ann Arbor, MI.
Science Foundation (NSF). 25. Cowlard, A., W. Jahn, C. Abecassis-Empis, G. Rein,
12. Zaruba, G.V., M. Huber, F.A. Kamangar, and I. Chlam- and J.L. Torero. “Sensor assisted fire fighting.” Fire
tac. “Indoor location tracking using RSSI readings from Technology 46 (2010): 719–741.
a single Wi-Fi access point.” Wireless Networks 13, 26. Jahn, W. “Inverse Modeling to Forecast Enclosure Fire
no. 2 (2007): 221–235. Dynamics.” PhD thesis (University of Edinburg, UK),
13. Fielder, R.S., and K.L. Stinson-Bagby. “High-temper- 2010.
ature fiber optic sensors for harsh environment appli- 27. Mandel, J., J.D. Beezley, and A.K. Kochansky. “Cou-
cations.” Proceeding of the SPIE — The International pled atmosphere-wildland fire modeling with WRF 3.3
Society for Optical Engineering 5272 (2003): 190–196.

SFF15_CH05_077_096.indd 94 4/23/15 11:43 AM

 5.9  Additional Reading 95

and SFIRE 2011.” Geoscientific Model Development 4 http://download.clib.psu.ac.th/datawebclib/e_resource/

(2011): 591–610. trial_database/WileyInterScienceCD/pdf/BDM/
28. Jahn, W., and G., Torero, J.L. Rein. “Forecasting fire BDM_4.pdf, accessed July 27, 2013.
dynamics using inverse computational fluid dynamics 38. Shen, C, R. Kupershtok, S. Adl, S.S. Bhattacharyya, N.
and tangent linearisation.” Advances in Engineering Goldsman, and M. Peckerar. “Sensor support systems
Software 47 (2012): 114–126. for asymmetric threat countermeasures.” IEEE Sensors
29. Rochoux, M.C., B. Delmotte, B. Cuenot, S. Ricci, and Journal 8, no. 6 (2008): 682–692.
A. Trouvé. “Regional-scale simulations of wildland fire 39. Gregory, O.J., and Y. Tao. “Ceramic temperature sensors
spread informed by real-time flame front observations.” for harsh environments.” IEEE Sensors Journal 5, no. 5
Proc. Combust. Inst. 34 (2013): 2641–2647. (2005): 833–838.
30. Rochoux, M.C., et al. “Data assimilation applied to 40. Martin, Ingrid M., Holly Bender, and Carol Raish.
combustion.” C.R. Méchanique 341 (2013): 266–276. “What Motivates Individuals to Protect Themselves
31. Rochoux, M.C., C. Emery, S. Ricci, B. Cuenot, and from Risks: The Case of Wildland Fires.” Risk Analy-
A. Trouvé. “Towards predictive simulation of wildfire sis 27, no. 4 (2007).
spread at regional scale using ensemble-based data 41. Jakes, Pamela. “Homeowners, Communities, and
assimilation to correct the fire front position.” Fire Wildfire: Science Findings from the National Fire Plan
Safety Science — Proc. Eleventh International Sympo- Proceedings of the Ninth International Symposium on
sium. International Association for Fire Safety Science. Society and Resource Management.” United States
32. Meacham, Brian J. “Integrating Human Factors Issues Department of Agriculture, 2002. http://www.na.fs
into Engineered Fire Safety Design.” Fire and Materials .fed.us/fire/cwpp/research_reports/gtr_nc_231.pdf
23 (1999): 273–279 42. Kobes, Magrethe, Ira Hesloot, Bauke de Vries, and Jos.
33. Brennan, P. “Victims and Survivors in Fatal Residen- Post. “Building safety and human behavior in fire: A
tial Building Fires.” Fire and Materials 23 (1999): literature review.” Fire Safety Journal 45, no. 1 (2011):
304–310. 1–11.
34. Cowell, R.G., Dawid, P., Lauritzen, S.L., and Spiegel- 43. Bateman, J.M., and B. Edwards. “Gender and evacu-
halter, D.J.; Probabilistic Networks and Expert Systems: ation: A closer look at why women are more likely to
Exact Computational Methods for Bayesian Networks; evacuate for hurricanes.” Natural Hazards Review 3,
Springer; 1999 no. 3 (2002): 107–117.
35. Bratvold, R., Bickel, J.E., and Lohne, H.P. “ Value of 44. Dow, Kirstin, and Susan L. Cutter. “Crying Wolf:
information in the oil and gas industry: past, present, Repeat Responses to Hurricane Evacuation Orders.”
and future” SPE Reservoir Evaluation & Engineering, Coast Management 26, no. 4 (1998): 237–52.
2009 45. Hasan, Samiul, Satish Ukkusuri, Hugh Gladwin, and
36. Lipshitz, R., and O. Strauss. “Coping with Uncertainty: Pamela Murray-Tuite. “Behavioral Model to Under-
A Naturalistic Decision-Making Analysis.” Organiza- stand Household-Lvel Hurricane Evacuation Decision
tional Behavior and Human Decision Processes 69, Making.” Journal of Transportation Engineering, May
no. 2 (February 1997): 149–163. http://www.oat.ethz 2011: 341–348.
.ch/education/Autumn_term_07/Material_on_Empirical_ 46. Jackson, Deborah. “What is an innovation ecosystem?”
Methods/Lipshitzuncertainty.pdf, accessed July 27, 2013. National Science Foundation.
37. Lipshitz, R., G. Kelin, and J. and Salas, E. Orasanu. 47. Keely, Larry, Helen Walters, Ryan Pikkel, and Brian
“Taking Stock of Naturalistic Decision Making.” Jour- Quinn. “Ten Types of Innovation: The Discipline of
nal of Behavioral Decision Making 14 (2001): 331–352. Building Breakthroughs.”

5.9  Additional Reading

Aralt, T. T. 2008. Cellular phones as indicators of the number Cisco visual networking index: global mobile data traffic
of people inside a tunnel during a fire. 15th World Con- forecast update, 2013–2018. Cisco. http://www.cisco
gress on Intelligent Transport Systems and ITS America .com/c/en/us/solutions/collateral/service-provider/
Annual Meeting 2008. 1454–1461. visual-networking-index-vni/white_paper_c11-520862.
html

SFF15_CH05_077_096.indd 95 4/23/15 11:43 AM

 96 Chapter 5   Stationary Sensors

Kanhere, S. S. 2013. Participatory sensing: crowdsourcing Liu, M. 2013. A study of mobile sensing using smartphones.
data from mobile smartphones in urban spaces. Lecture International Journal of Distributed Sensor Networks.
Notes in Computer Science. 7753. Nasar, J., Hecht, P., and R. Wener, R. 2007. Call if you
Koukoumidis, E., M. Martonosi, and L. –S. Peh. 2012. have trouble: mobile phones and safety among college
Leveraging smartphone cameras for collaborative road students. International Journal of Urban and Regional
advisories. IEEE Transactions on Mobile Computing. Research. 31: 863–873.
11: 707–723. Wu, O., Briggs, A., Kemp, T., Gray, A., MacIntyre, K., Rowley,
Lane, N. D., Miluzzo, E., Lu, H., Peebles, D., Choudhury, J., and K. Willett. 2012. Mobile phone use for contacting
T., and A. T. Campbell. 2010. A survey of mobile phone emergency services in life-threatening circumstances.
sensing. IEEE Communications Magazine 48: 140–150. Journal of Emergency Medicine. 42: 291–298.

SFF15_CH05_077_096.indd 96 4/23/15 11:43 AM

 97Head 97
6.2  Planning and AnalysisRunning

ChapteR
Data Collections
1
6
Ed Plaugher Abstract
Chair, NFPA Technical
Existing database collections already provide important information used by today’s
Committee on Data Exchange
fire service, not only on the fireground but in all aspects of their duties, including pre-
for the Fire Service
Conway, South Carolina fire planning, code enforcement, fire inspection, education, and investigations. Today,
the evolution of cyber-physical systems is providing access to massive amounts of
Leo Neumeyer new available data. These databases address key operational parameters such as fire
loss records, fire fighting resources, building information modeling, building sup-
Akualab, Inc.
Palo Alto, California porting infrastructure, and so forth. Current and evolving data collections along with
trends for future databases are an important consideration for Smart Fire Fighting.

Keywords 6.1 Overview
database Responders today are facing increasingly complex and difficult incidents while serving
data collection their dynamic and multi-faceted communities. Since the advent of computer-aided dis-
data patch, efforts to streamline and enhance incident response with data systems are being
undertaken. Footprinted in these efforts are the challenges of an ever increasing myriad
of responses. Designing updates and maintaining the complete data-supported framework
of duties, as well as interrelationships with the response environment, is key to the data
systems planned and utilized throughout the entire response community.
Phases of an incident (from pre-incident response, through the life of an incident,
through recovery from the incident) are key in a discussion of the modern data collection
and analysis system. Since the events of September 11, 2001, a national framework for
response and recovery has been established, which gives us the framework for the discus-
sions in this chapter.
Utilizing the guiding principles of response in this chapter — planning, prepared-
ness, response, and recovery — our discussions will shadow the efforts set out in the
proposed NFPA guide NFPA 951, Guide to Building and Utilizing Digital Information
(see Figure 6.1).

6.2  Planning and Analysis

6.2.1  Capability Assessment

Capability assessment data programs match real-world hazards and response assessments
to the data packages within a community. Items such as analysis of travel times and
station locations have been developed by utilizing well-established models of response
such as NFPA 1710, Standard for the Organization and Deployment of Fire Suppression

97

SFF15_CH06_097_108.indd 97 4/23/15 11:43 AM