In this issue, we describe a growing technology area at PSI in radiation detection systems that is addressing important national concerns.
Innovation can occur both in original discovery and also by finding new solutions to existing problems via technologies developed originally for other applications. Finding good solutions to problems where others have not is particularly rewarding.
PSI’s broad program base in radiological security is derived from our strong background in spectrally-based automatic target recognition. Our Poisson Clutter Split algorithm has been used to provide significant enhancements in detection sensitivity and false alarm rejection across detection platforms ranging from standoff to portals to hand-held.
Our technology development efforts seek to increase the capability of detection systems in low signal-to-noise regimes encountered in both portal and urban detection applications. Through funding from the Domestic Nuclear Detection Office (DNDO), PSI has recently demonstrated a significant performance enhancement in Advanced Spectroscopic Portals (ASP), Standoff Radiation Detection Systems (SORDS) and handheld isotope identifiers (RIIDs) through the use of new advanced detection and identification algorithms.
PSI’s Poisson Clutter Split (PCS) algorithm is a novel approach for radiological back-ground estimation that improves the detection and discrimination capability of medium resolution detectors. The algorithm processes energy spectra and performs clutter suppression, reducing noise levels in gamma-ray spectra enabling significant enhancements in detection and identification of low activity threats with spectral target recognition algorithms. The performance is achieved at the short integration times (0.5 – 1 second) necessary for operation in a high throughput and dynamic environment.
The key innovation of PCS lies in: 1) the use of a novel, nonlinear probabilistic representation of radiological backgrounds; 2) accurate modeling of gamma counts based on Poisson statistics and; 3) the use of the Generalized Likelihood Ratio Test to simultaneously perform isotope detection and identification. The variability among radiological background spectra collected at different locations over time can be attributed to two mechanisms:
Background clutter, i.e. changes in the energy-dependent count rate due to variations in isotopic composition at different locales, weather conditions, etc.
The random process of radioactive decay, described by Poisson statistics
The PCS algorithm was developed to accurately mitigate both sources of randomness in the radiological spectra by combining an accurate model of the Poisson processes and an efficient representation of background clutter within a single probabilistic framework. The resulting background estimation (model) is then used to assess the presence of a threat signature in any acquired medium-resolution spectrum, while at the same time performing isotope identification via correlation against a known set of library spectra.
The PCS algorithm is currently implemented in C ++ as an executable code for real-time processing. It has been compiled for Windows OS, Linux and Peta Linux. The code makes use of multicore architectures and parallel processing. PCS has also been integrated with a dual core ARM architecture in a 2" x 3.5" Sensor Interface Module. This capability allows PCS-based isotope detection and identification with small hand-held devices, while affording long battery life.
PCS implemented for real-time processing on an ARM architecture in a 2” x 3.5” Sensor Interface Module (SIM) communicating to both a hand-held scintillation detector and an Android smartphone for alarm display with isotope identification provided at 1 Hz update rate.
Android-based user interface showing PCS isotope identification results
The PCS algorithm has been integrated with ASP, SORDS and RIID sensor units and evaluated in field trials. It has successfully demonstrated the following:
For the SORDS surge and surveillance mission, PCS demonstrated the ability to achieve a factor of 3 increase in detection sensitivity over standard methods currently employed on such systems, while operating in cluttered environments at mission-relevant 1-in-8 hour false alarm rates. Ongoing DNDO sponsored efforts are focused on using environmental information provided by orthogonal sensing technologies to dynamically adjust the background models utilized by the PCS framework. Such an approach has already demonstrated a factor of ~ 2 improvement in sensitivity over the baseline PCS algorithm.
For portal applications, PCS demonstrated the ability to achieve high sensitivity with a reduced number of detectors and at relevant constant false alarm operating conditions, while allowing for vehicle traffic moving through the portal at speeds as high as 35 mph. This performance represents a significant improvement in ASP throughput relative to the current ~5 mph limited operation.
PCS demonstrated real-time (1Hz) iso-tope identification capability for handheld devices, which is an important factor to end users who require a fast assessment of alarms due to real threats or nuisance isotopes.
The benefit provided by this development and application of the PCS algorithm to these detection systems was recognized in an award for outstanding service in support of the mission of the DNDO. The award was presented during a ceremony at the 2014 Symposium on Radiation Measurements and Applications (SORMA XV). PSI has also been awarded a prime contract in support of DNDO’s Radiation Awareness Interdiction Network (RAIN) program. Under this effort, the PCS capability is being integrated with commercial off-the-shelf (COTS) gamma detectors to develop, demonstrate, and characterize an early warning system to monitor vehicular traffic and provide actionable information to aid in the interdiction of radiological and nuclear threats.
For more information on PSI’s Radiation Detection System Development, please contact Dr. Bogdan Cosofret firstname.lastname@example.org
PSI has been selected by the Advanced Research Projects Agency-Energy (ARPA-E) to develop and demonstrate the RMLD™ Sentry for continuously monitoring, locating, and quantifying volumetric leak rates of methane, a potent greenhouse gas, at natural gas production sites. PSI will work in collaboration with Heath Consultants Inc., Thorlabs, Princeton University, and The University of Houston on this development and demonstration of the RMLD™ Sentry.
PSI has been awarded a research contract from the Department of Energy to develop a High Speed VNIR/SWIR Hyperspectral Imager for Quantifying Terrestrial Ecosystems. PSI is developing a compact, airworthy hyperspectral imager to enable dynamic vegetation trait mapping in support of more accurate Earth system models. The vegetation trait mapping solution is based on PSI’s novel, compact, rugged, hyperspectral imager.
Editor Donna Lamb
Contributors: B. Cosofret and BD Green
Copy Right 2015 a publication of :
Physical Sciences Inc.
20 New England Business Center
Andover, MA 01810-1077
t 978.689.0003 f 978.689.3232