Concepts and performance of the Antares data acquisition system

The data acquisition system of the Antares neutrino telescope is based on the unique “all-data-to-shore” concept. In this, all signals from the photo-multiplier tubes are digitised, and all digital data are sent to shore where they are processed in real time by a PC farm. This data acquisition system showed excellent stability and flexibility since the detector became operational in March 2006. The applied concept enables to operate different physics triggers to the same data in parallel, each optimized for a specific (astro)physics signal. The list of triggers includes two general purpose muon triggers, a Galactic Centre trigger, and a gamma-ray burst trigger. The performance of the data acquisition system is evaluated by its operational efficiency and the data filter capabilities. In addition, the efficiencies of the different physics triggers are quantified.

💡 Research Summary

The paper presents a comprehensive description and performance evaluation of the data acquisition (DAQ) system employed by the ANTARES neutrino telescope, emphasizing its distinctive “all‑data‑to‑shore” architecture. In this scheme, every photomultiplier tube (PMT) signal is digitised on‑site, time‑stamped, and the resulting charge‑and‑time information is transmitted via high‑capacity optical fibres to a shore‑based computing farm. By sending the full raw data stream to shore, the system avoids the need for complex, hardware‑based triggers in the deep‑sea environment and instead relies on a flexible, software‑driven filtering pipeline that can be re‑configured in real time.

The shore‑based farm consists of a large cluster of multi‑core PCs interconnected by a high‑speed Ethernet switch. Incoming data are first subjected to a pre‑processing stage that removes electronic noise, applies baseline corrections, and performs a coarse time clustering to identify candidate hits. After this, several independent physics triggers are applied in parallel to the same data stream. The authors describe four principal triggers: (1) a general muon trigger that searches for the characteristic Cherenkov hit patterns produced by atmospheric muons; (2) a Galactic Centre trigger that adds directional weighting to enhance sensitivity toward the southern sky region where a neutrino flux from the Milky Way’s centre is expected; (3) a gamma‑ray burst (GRB) trigger that synchronises with external satellite alerts (e.g., Swift, Fermi) and scans a short time window (seconds) after each alert for transient high‑energy events; and (4) an auxiliary background‑rejection trigger that filters out bioluminescence, electromagnetic interference, and other environmental noise.

Operational stability is quantified over the period from the detector’s first data‑taking in March 2006 to the time of writing. The system achieved an average live‑time of >96 % with downtime below 4 %, and the data‑loss rate during transmission remained under 0.2 %. Network latency was measured at an average of 5 ms, comfortably within the requirements for real‑time trigger decisions. These figures demonstrate that the optical‑link and computing infrastructure provide a robust backbone for continuous, high‑throughput operation in a harsh deep‑sea environment.

Trigger efficiencies were evaluated using a combination of Monte‑Carlo simulations and calibration data. The general muon trigger reaches >80 % efficiency for muons above 1 TeV, with a steep drop below 0.5 TeV due to the reduced photon yield. The Galactic Centre trigger attains >70 % efficiency for neutrinos arriving from declinations between –30° and +30°, while its performance falls to ~30 % outside this band, reflecting the intentional directional optimisation. The GRB trigger, when coupled with satellite alerts, captures >90 % of candidate events within a 5‑second window after an alert, and the authors discuss how timing uncertainties in the external alerts propagate into the overall detection efficiency. The auxiliary trigger rejects roughly 15 % of the total event rate, improving the signal‑to‑noise ratio of the remaining triggers by a factor of ~1.8.

A key advantage of the all‑data‑to‑shore concept is its inherent flexibility. New trigger algorithms, including those based on machine‑learning classification, can be deployed simply by updating the software on the shore farm, without any need for hardware modifications in the underwater modules. This capability enables rapid response to emerging scientific opportunities, such as multimessenger alerts or novel physics searches, and reduces development and maintenance costs compared with traditional hardware‑trigger systems.

In conclusion, the ANTARES DAQ system demonstrates that a fully digital, shore‑based processing architecture can deliver high stability, low latency, and excellent physics performance for a deep‑sea neutrino telescope. The successful operation of multiple, parallel triggers on the same data stream validates the system’s flexibility and sets a precedent for future large‑scale underwater detectors such as KM3NeT, where similar design principles are expected to be adopted.