MCEER logo

MCEER Bulletin, Volume 21, Number 3, Fall 2007

skip navigation

Volume 21, Number 3, Fall 2007

mceer logo
ub 2020 banner with images of bridge collapse, wtc, water after hurricane katrina and fire after northridge earthquake

As MCEER graduates from the National Science Foundation’s Engineering Research Centers Program, it will continue its mission to create disaster resilient communities. As part of these future efforts, MCEER will help lead the University at Buffalo’s Strategic Strength in Extreme Events: Mitigation and Response. The University has identified eight different Strategic Strengths as part of its planning process for growth through the year 2020 (see article). As a facilitator of this Strategic Strength, MCEER will tap additional expertise in complex multidisciplinary projects related to extreme events at UB. The article below describes one such relevant project carried out by faculty in Industrial & Systems Engineering through the Center for Multisource Information Fusion (CMIF). We plan to regularly feature other projects in future issues of the MCEER Bulletin.

Dispatching and Routing of Emergency Vehicles in Disaster Mitigation Using Data Fusion

flow chart showing interaction

Interaction between different federates of the simulation model

The use of data fusion in disaster mitigation was the topic of a multi-year multidisciplinary project carried out through the Center for Multisource Information Fusion at UB and sponsored by the U.S. Air Force Office of Scientific Research. A simulation engine was created to answer this basic question through the development and testing of various levels and types of data fusion techniques. The project was conducted by a large team of scientists led by Dr. Peter Scott of the Department of Computer Science and Engineering.

A key part of this project was research on investigating the problem of optimizing the available resources to minimize the response time for a casualty pickup or a delivery and to maximize the life expectancy of casualties. This part of the effort was led by Dr. Rajan Batta and conducted by a team consisting of two former doctoral students, Arun Jotshi and Qiang Gong. Results are summarized in their doctoral dissertations and in four research articles:

The steps involved in this part of the research effort are described below.

Information about different entities—casualties, roads, emergency vehicles (EVs), clusters (casualties located in the same geographic area), hospitals—is assumed to flow from multiple sources. Data fusion was used to develop appropriate estimates, which formed the basis of an efficient dispatching and routing methodology. A Best Exit-Entry approach was used for efficient shortest path calculations, and travel delays due to congestion and road damage were incorporated. Three different dispatch decisions were modeled—dispatch to a casualty (in absence of clusters), dispatch to a cluster (and subsequently to one of its cells), and dispatch to a hospital.

Experimentation on parameter settings was illustrated with a case study of a post-earthquake disaster relief scenario in Northridge, California. Seven sets of parameters were investigated to illustrate the developed methodologies. Changing the parameters had a direct effect on the corresponding dispatching decisions. The results demonstrate that even though the average distance to a casualty pick up location and a cluster is comparable across the three scenarios, the average response time to a casualty is almost double the average response time to a cluster.

Sensors report information on casualties and the status of roads. These consist of satellites, sensor systems embedded in the infrastructure (e.g., surveillance cameras), police, civilians and reports (phone calls) received at the Emergency Response Centers. Information begins to flow from multiple sources to the fusion center. Variability of the sources generally leads to imprecision of the raw data, which then makes it difficult to determine the precise situation. After all original data have been collected, the Data Fusion Center fuses it and reports resulting estimates to the Dispatcher-Router center.

Information on each disaster-generated patient is composed of his/her location and injury class. For such purposes, four casualty classes are defined: Type 1 (mildly injured); type 2 (moderately injured); type 3 (severely injured); and type 4 (mortally injured). The casualties are distributed randomly across the affected area.

Information on each link (road segment) constitutes the level of damage and congestion on the link during a particular interval of time. Each level associated with a link is assigned a probability that measures the likelihood that the information is accurate.

In the case of a major disaster, the discovery rate of casualties is very high. A queue is thus created in which the casualties wait for service by an Emergency Vehicle (EV); hence, the need for an efficient dispatching and routing strategy.

The figure summarizes the interactions between the different components of the simulation software that was developed in this research.