Year
2015
Abstract
The demands of radiation portal monitors attempting to detect vehicle born sources on system sensitivity are extreme. Given the small interaction time, from ~2 seconds for a source moving at 5 mph to less than .2 seconds for a source moving at highway speed, the demands on the system detection capability are large. To detect sources moving at highway speed a new approach has been developed. The Correlated Variation Method is a technique to improve detection sensitivity by examining the arriving count distribution. Counts coming from background are uncorrelated in time, and across detectors in multi-detector systems; counts arriving from signals correlate across detectors and during the time the source passes through the field of view. The algorithm breaks the total integration time into multiple smaller counting intervals. The probability that the counts came from background alone is calculated. These probabilities, over all detectors and over all sub-counting intervals are combined. The probability combination has the effect of multiplying all of the sub counting intervals rather than simply summing as is ordinarily done. The source count rate required for detection is greatly reduced with this method, while simultaneously exhibiting very low false alarm rate. This method has demonstrated sub-Currie limit detection with deployed detection systems. The method is predicted to be able to reliably detect gram levels of SNM in a vehicle moving at 5 mph with a four NaI log portal monitor, and also detect less than 20 gm levels of SNM in a vehicle moving at highway speed. The high sensitivity of the method allows systems to meet requirements with relatively few detectors, making systems employing the algorithm affordable