1. The Trigger and Data Acquisition System for the KM3NeT neutrino telescope Carmelo Pellegrino Tommaso Chiarusi INFN - Sezione di Bologna VLVnT 2015 | Rome, September 14-16 |

2. 2 KM3NeT: one experiment, two sites Astronomy Research with Cosmics in the Abyss (ARCA) Oscillation Research with Cosmics in the Abyss (ORCA) 115 strings in the French site vertical spacing: ~6 metres 2 x 115 strings + 8 towers*-deep core in the Italian site vertical spacing: 36 metres The common element is the Detection Unit (DU): vertical structure hosting 18 Digital Optical Modules (DOMs, talk of D. Real) each one equipped with 31 3” PMTs and 1 piezo acoustic sensor for the positioning system. *Talk by M. Favaro in this session

3. Control Unit (Slow Control) “Broadcasting” WR Switch Receiving Switch Infrastructure TriDAS Core (Fast Acquisition) TDC, Acoustic, Monitoring Slow Control and WR PTP optical splitting of downward SC stream (broadcast) Shore Station Transport Protocol: UDP for both SC and Fast Acquisition 3 Sea network Broadcast fibre Network infrastructure Asymmetric network layout DOM uplink fibre

4. 4 Data throughputs Optical data Acoustic data ARCA (230 strings) (Gbps)38.096.0420.0 (ORCA equivalent) ARCA (230 strings)58.01.980.80250.013.0

5. Time-slicing performed off-shore by DOMs Frame i Frame i+1 Frame n Frame... TS i TS i+1... TS n Time ΔTS: O(100 ms) A DOM frame contains all data from its 31 PMTs occurring in a TS Frame i Frame i+1 Frame n Frame... DOM k DOM N One full optical TS (data from all the detector) is sent to a specific filtering process (see ahead...) Time-slicing 5 Frame i Frame i+1 DOM j Frame i Frame i+1 Frame n Frame... DOM 1 Frame n Frame... ΔTµ: O(1 µs) ΔTS >> ΔTµ Probability of a muon split into 2 TS < 10 -5

6. Trigger and Data Acquisition System (TriDAS) 6 On-shore computing Off-shore elements

7. The DataQueue (DQ) is the software component of the TriDAS which provides the data aggregation and distribution layer for both acoustic and optical data. DataQueue Read-out of the data frames (UDP) from a predefined number of CLBs Reassembling of the Time Slice (TS, which is compiled by the off- shore electronics) Distribution of the TSs to the proper DataFilter 7

8. Route the data collected from all DQs referring to a precise Time Slice to the very same oDF. Acoustic data must be sent in a continuous stream, addressing all data from one DQ to a single aDF. Optical WorldAcoustic World DataQueue: Two flavours 8 If needed more Filtering Power, just add it!

9. The main task of the optical DataFilter (oDF) is to reduce the data-stream coming from the DOMs' PMTs by selecting events that are interesting for physics analyses. Apply trigger algorithms that find space-time correlations between hits. Keep a buffer of raw data for dumping in case of an external alert (follow-up trigger, e.g.: GRB...) Send data to the Data Writer Optical DataFilter 9

10. Detection a im Topologica lTrigger Simple Causality Trigger Sky Scan Trigger Tracking Stack- Analysis Vertex / Inertia muon ✔✔✔✔✔✔ showers ✔✔✔✔✔ slowly moving particles ✔✔✔✔✔ sources ✔✔✔✔ Other algorithms are under development... Filtering algorithms B. Bakker Thesis, Trigger studies for the Antares and KM3NeT neutrino telescopes, Nikhef 2011 10 Overall throughput reduction factor ~10 -4

11. The DataWriter collects the filtered data coming from the various oDFs and performs write operation on permanent storage media in a ROOT compatible data format. DataWriter 11

12. The acoustic DataFilter (aDF) performs online calculation of the Time-Of-Arrival (TOA) of the acoustic signals emitted by the ground-based acoustic beacon array. Calculate the TOA using the data from all the sensors (piezo + hydrophone) collected on-shore by a DQ Send data directly to the DB Writer Acoustic DataFilter 12 Talk by F. Simeone yesterday in the C session

13. Control Unit The CU controls the offshore detector and the onshore processes via a Finite State Machine Talk by C. Bozza in this session 13 The Control Unit (CU) is composed of 4 elements: Master Control Program (MCP): –Controls the Detector Manager, the Database Writer and the TriDAS Manager Detector Manager (DM): –Performs complete configuration of the CLBs (PMT HV, instruments, etc…) –Records all parameters and the occurring events –Provides online monitoring of sensible parameters/measurements –Controls the power system TriDAS Manager (TM): –Operates the TriDAS core computing farm DB Writer –Performs write operation to the remote Data Base

14. Monitoring Tools: ROyWeb, Rainbow Alga, ROyFit Python-based web server Python and C++ API classes Demo - http://royweb.km3net.dehttp://royweb.km3net.de Restless Oyster Web reached a usable status Rainbow Alga prototype ready to be used, but put on hold for now ROyFit is still in a redesign phase 14 Talk by T. Gal in this session

15. Conclusions KM3NeT will perform continuous data taking for at least 20 years in two sites, reaching a total throughput up to 500 Gbps per site; challenging techniques: Multi-PMT, Asymmetric network, Hybrid WR; the TriDAS has been designed to be modular and to scale with a growing up detector; it will be tested very soon, after the deployment of DU-1 in the Toulon site and of DU-2 in the Capo Passero site. 15

16. Thank you for your attention!