Parallel optimization of signal detection in active magnetospheric signal injection experiments

Michael Gowanlock, Justin D. Li, Cody M. Rude, Victor Pankratius

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

Signal detection and extraction requires substantial manual parameter tuning at different stages in the processing pipeline. Time-series data depends on domain-specific signal properties, necessitating unique parameter selection for a given problem. The large potential search space makes this parameter selection process time-consuming and subject to variability. We introduce a technique to search and prune such parameter search spaces in parallel and select parameters for time series filters using breadth- and depth-first search strategies to increase the likelihood of detecting signals of interest in the field of magnetospheric physics. We focus on studying geomagnetic activity in the extremely and very low frequency ranges (ELF/VLF) using ELF/VLF transmissions from Siple Station, Antarctica, received at Québec, Canada. Our technique successfully detects amplified transmissions and achieves substantial speedup performance gains as compared to an exhaustive parameter search. We present examples where our algorithmic approach reduces the search from hundreds of seconds down to less than 1 s, with a ranked signal detection in the top 99th percentile, thus making it valuable for real-time monitoring. We also present empirical performance models quantifying the trade-off between the quality of signal recovered and the algorithm response time required for signal extraction. In the future, improved signal extraction in scenarios like the Siple experiment will enable better real-time diagnostics of conditions of the Earth's magnetosphere for monitoring space weather activity.

Original languageEnglish (US)
Pages (from-to)107-116
Number of pages10
JournalComputers and Geosciences
Volume114
DOIs
StatePublished - May 2018

Keywords

  • Parallel parameter search
  • Signal processing
  • Siple station experiment
  • Time series analysis

ASJC Scopus subject areas

  • Information Systems
  • Computers in Earth Sciences

Fingerprint

Dive into the research topics of 'Parallel optimization of signal detection in active magnetospheric signal injection experiments'. Together they form a unique fingerprint.

Cite this