A Distributed eResearch Tool for Evaluating Source Backtracking Algorithms Shanika Karunasekera, John Beaton and Adrian Dimech Department of Computer Science and Software Engineering The University of Melbourne VIC 3010, Australia Email: {shanika, john, agdimech}@csse.unimelb.edu.au Alex Skvortsov and Ajith Gunatilaka Human Protection and Performance Division Defence Science and Technology Organisation 506 Lorimer Street, Fishermans Bend VIC 3207, Australia Email: {alex.skvortsov, ajith.gunatilaka}@dsto.defence.gov.au Abstract—This paper presents an eResearch tool for evaluating and comparing Chemical, Biological, Radiological and Nuclear (CBRN) source-backtracking algorithms. The tool enables users to define a CBRN threat scenario to be simulated, which includes the geographical area of concern specified through a map, the simulated threat sources and their placement, and the configuration of sensors. Users can define multiple searchers who act as clients that interact with a central server and search for the simulated radiological source or sources. Based on the simulated measurements received from the searchers, and a chosen source backtracking algorithm, the central server updates the source estimates and issues control vectors to guide each searcher. I. I NTRODUCTION Localization of hazardous sources is important for eco- logical monitoring (pollution control), defence applications (chemical, biological, radiological, and nuclear (CBRN) de- fence) and security systems (protection lines). Prompt (close to real-time) determination of the characteristics of a source based on point measurements or data from a sensor network, and prediction of its long-term impact on the environment (situation awareness) are crucial to mitigate the harm from these hazardous sources. For instance, in the case of a so called “dirty bomb” or a radiological dispersal device, locating the radiological material before it is dispersed can prevent the emission of radioactive particles into the atmosphere; this can avoid harmful radiation exposure to people. It also averts the need for expensive clean up. Therefore, CBRN data fusion algorithms for source detection, localization, and tracking have recently become a topic of intensive research [1], [2], [3], [4], [5], [6], [7], [8], [9]. Source backtracking algorithms that are capable of local- izing hazardous sources rapidly and accurately may benefit first responders (for example, the Incident Response Regiment of the Australian Defence Forces (ADF) [10]) by improv- ing search efficiency and reducing their time of exposure to harmful substances. Indeed, in order to convert this re- search into an operational capability, extensive testing and evaluation are required so that performance criteria such as localization accuracy, search time and the resulting exposure of the searchers can be assessed. Unfortunately, due to the hazardous nature of the sources, lab and field validation of these algorithms is costly and, often, even unfeasible because of risks of environmental contamination or human exposure. Even when feasible, these experiments are often of limited scope [11], [12]. These considerations motivated the development of a special software tool that would allow researchers and practitioners to simulate source backtracking in realistic environments (e.g. complex urban settings) and undertake research on algorithm efficiency, scalability, inter-comparison and validation. The proposed tool should allow the seamless integration of dif- ferent algorithms implemented on a variety of software plat- forms without adapting the original source code to a common programming language. This will not only improve the source code reuse and may significantly reduce development efforts, but also enhance research collaboration. Another purpose of the eResearch tool is to facilitate a dialog with a commander (military or the first responder) about future requirements and performance benchmarks for the source backtracking system through some practical experience with these systems in the “virtual reality”. We expect that this experience will draw the commander’s attention to such areas as risk mitigation, value of accurate information, and optimum resource allocation. In a military context, this is called seminar wargaming and has become the de-facto standard to define the future capabilities of Australian Defence Forces [13]. This paper describes a distributed eResearch tool we devel- oped to enable objective evaluation and comparison of source detection and backtracking algorithms, and enhance research collaboration. The rest of the paper is organized as follows. Section II provides an overview of the system, including the functional and non-functional requirements. Section III describes the sys- tem architecture. Section IV describes some implementation aspects. Section V discusses the system behavior, and provides an evaluation of the system. II. SYSTEM OVERVIEW The system was designed and developed jointly by the Human Protection and Performance Division (HPPD) of the Defence Science and Technology Organisation (DSTO), Aus- tralia and the Department of Computer Science and Software Engineering of The University of Melbourne. This was an extension of DSTO’s previous work in CBRN data fusion and Operations Research [8], [7], [3]. DSTO researchers 2010 Sixth IEEE International Conference on e–Science 978-0-7695-4290-4 2010 U.S. Government Work Not Protected by U.S. Copyright DOI 10.1109/eScience.2010.16 17