Engineering topics
Michalak, Sarah
Publications and source records attributed to Michalak, Sarah.
COMPARISON OF RADIO-FREQUENCY INTERFERENCE MITIGATION STRATEGIES FOR DISPERSED PULSE DETECTION
Impulsive radio-frequency signals from astronomical sources are dispersed by the frequency-dependent index of refraction of the interstellar media and so appear as chirped signals when they reach Earth. Searches for dispersed impulses have been limited by false detections due to radio-frequency interference (RFI) and, in some cases, artifacts of the instrumentation. Many authors have discussed techniques to excise or mitigate RFI in searches for fast transients, but comparisons between different approaches are lacking. This work develops RFI mitigation techniques for use in searches for dispersed pulses, employing data recorded in a 'Fly's Eye' mode of the Allen Telescope Array as a test case. We gauge the performance of several RFI mitigation techniques by adding dispersed signals to data containing RFI and comparing false alarm rates at the observed signal-to-noise ratios of the added signals. We find that Huber filtering is most effective at removing broadband interferers, while frequency centering is most effective at removing narrow frequency interferers. Neither of these methods is effective over a broad range of interferers. A method that combines Huber filtering and adaptive interference cancelation provides the lowest number of false positives over the interferers considered here. The methods developed here have application to other searches for dispersed pulses in incoherent spectra, especially those involving multiple beam systems.
Investigation and design of a network, I/O and storage SDC susceptibility analzer
This slide deck is intended for the Durmstrang Team Quarterly Review to be held January 25, 2011 in Baltimore, MD. The slides provide an overview of the Investigation and Design of a Network, I/O and Storage SDC Susceptibility Analyzer project. In addition, the slides provide the goals of the project, details of the work done to date and the project status and future work.
A fast and memory-sparing probabilistic selection algorithm for the GPU
A fast and memory-sparing probabilistic top-N selection algorithm is implemented on the GPU. This probabilistic algorithm gives a deterministic result and always terminates. The use of randomization reduces the amount of data that needs heavy processing, and so reduces both the memory requirements and the average time required for the algorithm. This algorithm is well-suited to more general parallel processors with multiple layers of memory hierarchy. Probabilistic Las Vegas algorithms of this kind are a form of stochastic optimization and can be especially useful for processors having a limited amount of fast memory available.