Fuzzy-Logic Radioisotope Identifier for Gamma Spectroscopy in Source Search

Abstract

Detection and identification of radioactive nuclear materials in urban searches can be fully performed with a portable gamma ray detector-spectrometer. Due to limited acquisition time and, as a consequence, low signal to noise ratio (SNR), development of fast and accurate real-time radioisotope identifier (RIID) algorithms is essential for automated source detection. In this paper, we evaluate the performance of fuzzy logic real-time radioisotope identification (FL-RIID) in several urban search scenarios. FL-RIID performance is tested on a database of searches consisting of injections of synthetic sources into experimental nuclear background spectra, acquired in one-second time intervals with a moving sodium iodide (NaI) gamma radiation detector-spectrometer. Performance of FL-RIID is benchmarked against that of maximum-likelihood (ML) fitting method. Demonstrated advantages of FL-RIID over ML in search applications include lower false alarm rate and faster execution time.