Richard Stottler, Stottler Henke Associates, Inc, Dr. Chris Bowman, Data Fusion & Neural Networks
Keywords: Electro Magnetic Interference (EMI), Satellite-Based EMI, EMI detection and Identification, Artificial Intelligence, Neural Networks, Mitigation, Rescheduling
Abstract:
Commanding, controlling, and maintaining the health of satellites requires a clear operating spectrum for communications. Electro Magnetic Interference (EMI) from other satellites can interfere with these communications. Determining which satellite is at fault improves space situational awareness and can be used to avoid the problem in the future. The Rfi detection And Prediction Tool, Optimizing Resources (RAPTOR) monitors the satellite communication antenna signals to detect EMI (also called RFI for Radio Frequency Interference) using a neural network trained on past cases of both normal communications and EMI events. RAPTOR maintains a database of satellites that have violated the reserved spectrum in the past. When satellite-based EMI is detected, RAPTOR first checks this list to determine if any are angularly close to the satellite being communicated with. Additionally, RAPTOR checks the Space Catalog to see if any of its active satellites are angularly close. RAPTOR also consults on-line databases to determine if the described operating frequencies of the satellites match the detected EMI and recommends candidates to be added to the known offenders database, accordingly. Based on detected EMI and predicted orbits and frequencies, RAPTOR automatically reschedules satellite communications to avoid current and future satellite-based EMI. It also includes an intuitive display for a global network of satellite communications antennas and their statuses including the status of their EM spectrum. RAPTOR has been prototyped and tested with real data (amplitudes versus frequency over time) for both satellite communication signals and is currently undergoing full-scale development. This paper describes the RAPTOR technologies and results of testing.
Date of Conference: September 20-23, 2016
Track: Poster