RFI Mitigation and Burst Detection with a Digital Receiver RFI2004

Transcription

1 RFI Mitigation and Burst Detection with a Digital Receiver RFI2004 July 16-18, 2004 Cedric Dumez-Viou1,4, Andrée Coffre4,Pierre Colom2, Laurent Denis4, Alain Lecacheux2, Jean-Michel Martin3, Philippe Ravier1, Rodolphe Weber1, Philippe Zarka2 1 LESI, Polytech'Orleans - Université d'orléans 2 LESIA, Observatoire de Paris/CNRS 3 GEPI, Observatoire de Paris/CNRS 4 Station de Radioastronomie de Nançay 1

2 RSR: Radio Spectrum Reconquest 2 High dynamic receiver ( > 70dB ) Highly reconfigurable for multipurpose use

3 RSR: Radio Spectrum Reconquest 3 High dynamic receiver ( > 70dB ) Highly reconfigurable for multipurpose use

4 Current implementations Sigma detectors for Time-Frequency Blanking T-FDMA RFI mitigation Wideband RFI mitigation 4 Fine Structures Future works...

5 Given constraints 5 Implementation of basic methods to be used today. RFIs detection based on power estimation. Real-time algorithms. - Only a small part of the signal is available. - Limited data-flow.

6 Blanking of T-FDMA modulation: principle 6

7 Blanking of T-FDMA modulation: principle 7

8 Blanking of T-FDMA modulation: principle 8

9 Blanking of T-FDMA modulation: principle 9

10 Blanking of T-FDMA modulation: principle Time Original Data 10 Frequency

11 Blanking of T-FDMA modulation: principle Time Original Data Full Spectrum Blanking 11 Frequency

12 Blanking of T-FDMA modulation: principle Time Original Data Full Spectrum Blanking Timefrequency Block blanking 12 Frequency

13 Blanking of T-FDMA modulation: method T-FDMA are narrow band signals --> efficient mean and variance estimate using median filtering or robust mean. 13 Full spectrum blanking: - 20% loss of data. - Excellent rejection of RFI. Time-frequency Block blanking: - 2.5% loss of data. - Unequal equivalent time of integration.

14 Blanking of T-FDMA modulation: IIIZw35 OH megamazer results 14

15 Blanking of Wideband RFIs: principle 15

16 Blanking of Wideband RFIs: principle 16

17 Blanking of Wideband RFIs: principle 17

18 Blanking of Wideband RFIs method 18 Lightnings, Radars,... Occupy the full band or most of it. Median filtering not applicable. Estimation of mean and variance compromised.

19 Blanking of Wideband RFIs method 19 Lightnings, Radars,... Occupy the full band or most of it. Median filtering not applicable. Estimation of mean and variance compromised.

20 Blanking of Wideband RFIs method 20 Lightnings, Radars,... Occupy the full band or most of it. Median filtering not applicable. Estimation of mean and variance compromised.

21 Blanking of Wideband RFIs method Lightnings, Radars,... Occupy the full band or most of it. 21 Median filtering not applicable. Estimation of mean and variance compromised Wideband BUT not continuous in time. Some spectrum are free of RFIs Time averaging to find clean spectrum

22 Blanking of Wideband RFIs results 22

23 Blanking of Wideband RFIs results 23 Blanking simulation on real data. Current results allow 10 minutes of integration without signs of RFIs. Further testes needed to evaluate RFI rejection.

24 Fines Structures: principle Solar and Jovian radio-emissions Very brief «Chirps» (~100ms). High frequency and time resolution required to understand the underlying physics High volume of data (50Go/h) --> Backup of interesting events only. Automatic recorder 24 Detection of RFI-like SOIs, Corrupted by RFIs (frequency lines and wide-band signals), With real-time constraints.

25 Fines Structures: method 25

26 Fines Structures: results 26

27 More advanced methods for the future... Encouraging results for real-time RFI mitigation. Should work shortly for daily-use observation. Classifier (SVM, neural networks, expert systems,...). 27 Automatic determination of RFIs characteristics. Selection of the appropriate parameters for blanking. Implementation of cyclostationary detector. Highlight hidden periodicities of signals. Present in telecommunications. Absent in most natural signals.

28 More advanced methods for the future... 28