diff --git a/applications/lofar2/doc/prestudy/station2_sdp_transient_detection.txt b/applications/lofar2/doc/prestudy/station2_sdp_transient_detection.txt index 30f46842d564676cacad3c309be623dded55680b..3595c039b001fb8ee6c359d10bba9a19b42d3100 100644 --- a/applications/lofar2/doc/prestudy/station2_sdp_transient_detection.txt +++ b/applications/lofar2/doc/prestudy/station2_sdp_transient_detection.txt @@ -1,3 +1,14 @@ + +Cosmic rays at LOFAR, proefschrift A. Corstanje, sep 2019: +Cosmic rays arrive on Earth from all directions. These high-energy particles (10**9 to 10**20 eV) +are atomic nuclei.This is the field of astroparticle physics. At > 10**14 eV the particles is +too rare to measure directly. Instead they can be mesured indirectly, because they cause a secondary +air shower of particles when they interact with nuclei in molecules in the atmosphere. The +air showers produce radio signals, that arrive on the ground as short pulses in the range of 30 -80 +MHz. LOFAR detected first cosmic ray in June 2011 and has detected 2717 pulses up to end 2018. The +radio pulse is short (10 - 100 ns). The timing of the pulse can be better than the sample period +and depends on the SNR. + 1) Transient detection (TDet) Design - no self triggering yet for MVP @@ -71,7 +82,7 @@ CR, LI pulse duration in comlex envelope is ~50 - 100ns . CR ~ 1 pulse per min LI binning in time, to limit maximum message rate -. bin time fixed in time of 10 - 100 us +. bin time fixed in time of 10 - 100 us --> 10000 - 1000 bins/s . only report strongest event during bin . typically each bin will show an event during a flash @@ -85,3 +96,75 @@ Message . timestamp of crossing threshold (or of maximum after crossing threshold until uncrossing threshold?) . max level of pulse is not needed + +5) Planning topics Transient detection + +- Comparison with LOFAR1 [1, 5] + . IIR filter with programmable coefficients to implement: + - HPF (> 30MHz), BPF + . An IIR ellip BPF needs about order 10 to achieve-40 dB at 30 and 80 MHz, so 5 biquads + - or notches (one at 15 MHz short wave band, one at 88 MHz FM band) + . filtering of FM band increases SNR, short wave band filtering increases stability of SNR + . in LOFAR1 we implemented a three stage filter in which the parameters b0, b1, b2, a1, a2 can + be set. By choosing different values for the parameters one can implement a high-pass, + low-pass or notch filter. + . With one biquad per notch the attenuation at f0 is strong, but the notch bandwidth is small + (~3 MHz at -20dB, ~10 MHz at -10dB) + . peak detect if abs > mu + k * sigma --> simplifies to --> abs > k2 mu for Gaussian signal + . binning for LI, because 10000 triggers / flash, so only trigger strongest per bin (= time slot) +- Comparison cosmic ray and lightning [1, 2] + . Check that LI binning is transparant to CR + . Format of detection message +- Requirements, MVP + . number of signal inputs per FPGA (max 12 single pol) + . one polarization per antenne ? + . CR signals are small and often vary significantly over the distance of a single station. + Number antennas of the order of 20 are required distributed across the station [1] +- Detection algorithm analysis in Python: + . Operate on raw data or on subbands ? --> ik denk raw want CR pulse duurt < 20 ns + . BPF for RFI suppression: 30 - 80 MHz bandpass, possibly with notch filters + . complex envelope, abs amplitude calculation, running average, threshold detection + . Verification (parameter tuning) with existing TBB data from LOFAR1 +- L4 SDPFW Decision: TDet +- L5 SDPFW Design: TDet + . Architecture + - Block diagram + - Signal levels and quantization + . Monitor and control + . Implementation + - Efficient mapping on DSP multipliers + - New components + . Verification in simulation + . Verification on hardware +- L2 ICD SC-SDP OPC-UA interface + . CP and MP + . Dynamic behavior +- L1 ICD STAT-CEP interface + . Detection message ICD: + - format (destination, application info) and control (enable per signal input, rate) + - forced message every 30 s to show alive + + +6) TBuf planning for comparison [7]: +- R_booked = 1.7: Booked time in days = R_booked * N_story_points +- R_lead = 2.1: Lead time in days = R_lead * N_story_points +- SDP Design: N_story_points = 29 story points --> N_booked_days = 49 +- SDP Realisation: N_story_points = 93 story points --> N_booked_days = 158 +- SDP hours: (49 + 158) * 8 = 1656 hours + +References: + [1] https://support.astron.nl/confluence/display/L2M/2023-04-18+Meeting+notes+Preparation+CR+group+meeting?preview=/117413179/117413191/Cosmic_Rays_Firmware_Implementation_Document%20comments%20BH.pdf + [2] https://support.astron.nl/confluence/display/L2M/2023-04-13+Meeting+notes+update+session+with+Cosmic+Ray+group?preview=/117411807/117411963/20230413%20Comparison%20LIFT%20vs%20CR%20use%20cases.pdf + [3] https://www.sgo.fi/KAIRA/LOFAR-ASTRON-MAN-064_v12.pdf --> in Documents/LOFAR2/References/LOFAR1_station/ + + [4] Doorduin. TBB Transient Detector User Manual. Technical Report LOFAR-ASTRON-DDD-056, + ASTRON, Dwingeloo (The Netherlands), November 2009. + [5] A. Horneffer and S. Laf`ebre. The LOFAR VHECR Trigger. Technical Report LOFAR-ASTRON- + SDD-055, ASTRON, Dwingeloo (The Netherlands), March 2007 --> in https://support.astron.nl/confluence/display/L2M/Temporary+storage+of+documents+and+papers + + [6] Air_shower_measurements_with_LOFAR_2009.pdf + + [7] https://support.astron.nl/confluence/display/SBe/SDP+Transient+buffer+planning+experience+and+estimate + +VHECR = Very High Energy Cosmic Rays +