PaperlayerView exact revision on arXiv

Source-linked AI summary

The CHIME Fast Radio Burst Project: System Overview

The CHIME/FRB Collaboration, :, M. Amiri, K. Bandura, P. Berger, M. Bhardwaj, M. M. Boyce, P. J. Boyle, C. Brar, M. Burhanpurkar, P. Chawla, J. Chowdhury, J. F. Cliche, M. D. Cranmer, D. Cubranic, M. Deng, N. Denman, M. Dobbs, M. Fandino, E. Fonseca, B. M. Gaensler, U. Giri, A. J. Gilbert, D. C. Good, S. Guliani, M. Halpern, G. Hinshaw, C. Hofer, A. Josephy, V. M. Kaspi, T. L. Landecker, D. Lang, H. Liao, K. W. Masui, J. Mena-Parra, A. Naidu, L. B. Newburgh, C. Ng, C. Patel, U. -L Pen, T. Pinsonneault-Marotte, Z. Pleunis, M. Rafiei Ravandi, S. M. Ransom, A. Renard, P. Scholz, K Sigurdson, S. R. Siegel, K. M. Smith, I. H. Stairs, S. P. Tendulkar, K. Vanderlinde, D. V. Wiebe

arXiv:1803.11235v1astro-ph.IMastro-ph.HE

TL;DR

FRB surveys face sparse detections, uncertain sources, and poor localization despite a high inferred event rate. This paper describes CHIME/FRB, which upgrades CHIME with wide-field beamforming and high-resolution real-time searches; predicted rates range from 0.6–11 bursts per day under revised assumptions, while the project is positioned to advance FRB studies.

  • Problem

    FRBs have uncertain origins, limited localization, and far fewer detections than their inferred sky rate, motivating a wide-field, sensitive survey.

  • Method

    CHIME/FRB forms 1024 beams, searches 16k-channel data in real time with dedispersion and machine-learning RFI classification, and uses optimized processing for low latency.

  • Results

    0.6–11 bursts per day are predicted under the revised rate analysis, with 0.3–2 bursts per day if FRBs are detectable only at 700–800 MHz.

  • Takeaways & Limitations

    CHIME/FRB should enable broad FRB surveys, repeat-burst studies, and potential progress toward localization, host identification, and redshift determination.

  • Takeaways & Limitations

    CHIME/FRB’s angular resolution alone is insufficient for host-galaxy identification, and H I absorption redshifts may be undetectable for FRBs in dwarf galaxies.

Abstract

from arXiv · show

The Canadian Hydrogen Intensity Mapping Experiment (CHIME) is a novel transit radio telescope operating across the 400-800-MHz band. CHIME is comprised of four 20-m x 100-m semi-cylindrical paraboloid reflectors, each of which has 256 dual-polarization feeds suspended along its axis, giving it a >200 square degree field-of-view. This, combined with wide bandwidth, high sensitivity, and a powerful correlator makes CHIME an excellent instrument for the detection of Fast Radio Bursts (FRBs). The CHIME Fast Radio Burst Project (CHIME/FRB) will search beam-formed, high time-and frequency-resolution data in real time for FRBs in the CHIME field-of-view. Here we describe the CHIME/FRB backend, including the real-time FRB search and detection software pipeline as well as the planned offline analyses. We estimate a CHIME/FRB detection rate of 2-42 FRBs/sky/day normalizing to the rate estimated at 1.4-GHz by Vander Wiel et al. (2016). Likely science outcomes of CHIME/FRB are also discussed. CHIME/FRB is currently operational in a commissioning phase, with science operations expected to commence in the latter half of 2018.

1. FAST RADIO BURSTS

FRBs are brief, dispersed radio bursts whose origins remain unknown, while existing surveys detect few events despite a much higher inferred sky rate. CHIME/FRB is designed to address these limitations using CHIME’s large field of view and multiple formed beams.

  • FRB phenomenon: FRBs are millisecond-duration radio bursts with dispersion measures that can greatly exceed Galactic expectations, suggesting extragalactic origins.The first recognized burst had a 30 Jy brightness, duration below 5 ms, and DM 375 pc cm−3 versus a Galactic maximum of 25 pc cm−3 along that sightline.
  • Detection challenge: Fewer than three dozen FRBs had been detected despite an inferred 1.4-GHz sky rate of several hundred to a few thousand bursts per day.The discrepancy is attributed largely to the small fields of view of existing telescopes.
  • Open questions: FRB sources remain unknown, and typical positional uncertainties of many arcminutes hinder identification of multi-wavelength counterparts and host galaxies.Models commonly involve compact objects, with some proposing cataclysmic events.
  • Repeating FRBs: FRB 121102 is the only source described here as repeating, with many bursts at a consistent DM and interferometric localization to a dwarf galaxy.Its repetition demonstrates that at least this source did not originate in a cataclysmic event.
  • CHIME/FRB motivation: CHIME/FRB uses CHIME’s >200 sq. deg. field of view, collecting area, bandwidth, and 1024 independent beams to detect multiple FRBs in parallel.The project repurposes a telescope originally designed for hydrogen-intensity mapping while operating an FRB detector alongside its other programs.

2. THE CHIME TELESCOPE STRUCTURE, FEEDS, AND ANALOG SIGNAL PATH

CHIME is a stationary, cylindrical transit telescope optimized for wide sky coverage and rapid mapping, with four reflectors and densely distributed dual-polarization feeds. Its analog signal path carries 2048 feed signals to receiver-hut F-Engines for digitization and channelization.

  • Design motivation: CHIME’s design was driven by all-Northern-hemisphere hydrogen-intensity mapping, requiring many feeds and a stationary telescope for fast, low-cost sky coverage.The original mapping target spans redshift z = 0.8–2.5.
  • Key properties: The telescope operates across 400–800 MHz with a 2.5°–1.3° field of view and 1024 beams.The listed beam width is 40′–20′, while the FRB search uses 0.983 ms time resolution and 24.4 kHz frequency resolution.
  • Telescope structure: CHIME comprises four stationary 20-m-wide, 100-m-long cylindrical paraboloidal reflectors aligned North–South.The 5-m focal length gives f/D = 0.25 and places the feeds in the aperture plane to reduce cross-talk and ground pick-up.
  • Feeds: 256 dual-polarization feeds on each cylinder produce 2048 signal paths for digital beam generation.Feeds are spaced by 30 cm along 80 m of each focal line.
  • Analog signal path: Equal-length 50 m coaxial cables carry feed signals to shielded, liquid-cooled receiver huts, where they are amplified, filtered, digitized, and split into 1024 frequency channels.The focal-line components themselves are not temperature controlled.

3. UPGRADED CHIME CORRELATOR

The upgraded correlator adds the high time and frequency resolution and spatially discrete beams required for FRB detection. Its L0 pipeline forms 1024 beams and produces finely channelized data, while the broader system routes these streams to FRB-search processing.

  • FRB requirements: FRB detection requires millisecond-scale time resolution, high frequency resolution for dedispersion, and spatially localized beams beyond the baseline correlator’s BAO-oriented output.Without dedispersion, dispersion by free electrons can render FRB signals undetectable at CHIME frequencies.
  • Frequency resolution: CHIME/FRB uses 1-ms cadence and 16k frequency channels to minimize intra-channel dispersion smearing.The native 1024-channel resolution would produce greater smearing, motivating upchannelization.
  • Correlator upgrade: The correlator required upgrades because its 6.5 Tb/s F-Engine output could not readily be duplicated or distributed beyond the X-Engine.Additional FRB processing therefore occurs inside the correlator system.
  • F-Engine: The F-Engine digitizes 2048 timestreams and channelizes 400 MHz into 1024 bins of 390 kHz each before data are reorganized for X-Engine processing.The F-Engine’s total digitized data rate is 13.1 Tb/s.
  • L0 beamforming: The L0 pipeline forms 256 North–South FFT beams and four East–West phased beams, yielding 1024 static beams that tile the primary beam.Voltage samples are transformed, frequency-downsampled, averaged, and polarization-summed into Stokes-I data.

4. CHIME/FRB INSTRUMENT AND SOFTWARE PIPELINE

CHIME/FRB is a real-time, four-level FRB search system processing high-resolution data from 1024 formed beams. Its bonsai-based dedispersion and candidate pipeline address extreme computational demands while targeting low latency and sensitivity.

  • Instrument and pipeline: CHIME/FRB receives 16k frequency channels at 1-ms cadence for 1024 beams and processes candidates through L1–L4 stages.L1 performs per-beam RFI rejection and dedispersion; later stages consolidate, classify, and act on detections.
  • Candidate identification: The dedispersion transform searches DM and arrival time alongside spectral index, scattering time, and intrinsic width, producing a 5D SNR array.Candidate cells are thresholded at 10σ, and local maxima are isolated before L1 headers are generated.
  • Computational challenge: 1.5 PB/day and 10^11 SNR values per second make the CHIME/FRB search substantially larger than previous FRB searches.The input and computation rates are also comparable to the planned Square Kilometer Array search.
  • Dedispersion: Tree dedispersion reduces computational cost to O(TF log F), compared with O(TF^2) for direct dedispersion.Tree methods are parametrically faster but can suffer from approximation-related SNR loss.
  • Dedispersion: bonsai uses blocked, cache-tuned tree dedispersion to achieve a factor ∼30 speed-up on multi-core machines.Its incremental operation can trigger within seconds even when dispersion delays are much longer, supporting a 20-s baseband buffer.
  • Validation: Initial injection and pulsar analyses find the L1 pipeline produces fewer RFI false positives and better detects faint signals than RRATTrap/PRESTO comparisons.These findings are described as initial analyses rather than a final evaluation.

4.4. L2: Grouping of Multi-Beam Events

L2 consolidates per-beam candidate reports into multi-beam events, refines their positions, estimates fluxes, and separates RFI from astrophysical events.

  • Grouping of Multi-Beam Events: L2 buffers L1 headers, groups candidates by time, DM, and sky position, and uses simplified DBSCAN clustering with O(n log n) complexity.Beam-adjacency thresholds enforce spatial connectivity while DM and time thresholds reflect coarse-grained uncertainties.
  • Grouping of Multi-Beam Events: After clustering, events are classified as RFI or astrophysical, and each L2 header contains one or more grouped L1 headers.RFI events are sent directly to the L4 database.
  • Position refinement: For multi-beam events, relative beam SNRs are matched to a sky-position and spectral-index lookup table, then χ2 minimization refines the estimate and uncertainty region.Single-beam events use a separate precomputed mapping.
  • Flux estimation: Real-time flux estimates use the single-pulse radiometer equation with bonsai SNR and pulse width, corrected for beam sensitivity and RFI-masked bandwidth.The sky temperature is estimated from a reprocessed 408-MHz map scaled to 600 MHz.

4.5. L3: Identification of Extragalactic Events and Action Determination

L3 identifies astrophysical events by comparing their measured properties with known sources and Galactic DM models, then applies predefined response rules.

  • Identification of Extragalactic Events and Action Determination: L3 uses determined DM and localized position to distinguish Galactic from extragalactic events and known from unknown sources.Automatic recognition allows alerts to target FRBs rather than Galactic objects.
  • Source association: The known-source database includes pulsars, RRATs, and FRBs, and is updated as CHIME/FRB and other surveys discover new sources.Astrophysical events are compared with catalogued source positions and DMs while accounting for measurement error.
  • Extragalactic classification: For unassociated events, Galactic DM models provide a maximum line-of-sight DM, with model disagreement combined with L1 uncertainty as a systematic error.The resulting full uncertainty is used in the extragalactic classification condition.
  • Action determination: Predefined rules can ignore events, archive headers, request intensity or baseband data, initiate CHIME/Pulsar monitoring, or send alerts.Rules may be configured for individual sources or source groups.
  • Action determination: L4 executes L3-selected actions and archives headers and analysis products for every event passing L1, including both astrophysical and RFI events.The archive is paired with a web interface for event search, visualization, downloads, and user classification.

4.7. Intensity-Data Call-back

Because CHIME/FRB cannot retain all intensity data, L1 maintains a multiresolution ring buffer so L3 can retrieve recent data for selected events.

  • Intensity-Data Call-back: L1 stores recent beam intensity data in buffers and retrieves it when L3 identifies an interesting event.The detected and immediately adjacent beams are saved independently.
  • Intensity-Data Call-back: At DM 13,000 pc cm−3, a pulse can sweep through the CHIME band in ∼250 s, exceeding available full-resolution memory.This long delay motivates progressive downsampling in the buffer.
  • Intensity-Data Call-back: The telescoping ring buffer retains 60 s at full resolution, 120 s downsampled by two, and 240 s downsampled by four.Older data are progressively stored at coarser time resolution to optimize memory use.
  • Intensity-Data Call-back: Called-back intensity data are written to a network-shared archiver for offline analysis and visualization, including refined event parameters and localization.Dynamic spectra are generated for single-pulse visualization.

4.8. RFI Excision

CHIME/FRB uses staged RFI excision, combining intensity preprocessing, candidate-level classification, and beam-group classification to distinguish terrestrial signals from astrophysical transients.

  • RFI mitigation uses different criteria at multiple pipeline stages to minimize terrestrial-signal misclassification.
  • L1: L1 removes RFI by zero-DM detrending and subtraction, while clipping amplitude and standard-deviation outliers.
  • L1: Nine clipping transforms—five at 3σ and four at 5σ—are iterated six times, with the sequence repeated twice and followed by polynomial and spline detrending.
  • Candidate-level RFI excision: A support vector machine classifies candidate events using local SNR fall-off at nearby sub-optimal DMs and global SNR behavior.
  • L2: L2 groups detections across neighboring beams and DM-time thresholds, then classifies them from SNR distributions and grouped-beam shapes.
  • L2: All L1 candidates reaching L2 are stored regardless of classification, enabling later classifier reclassification and retraining.

5.1. Real-time Alert System

CHIME/FRB plans rapid, interoperable FRB alerts to coordinate observations across radio bands and searches for repeating bursts or counterparts.

  • Rapid FRB reporting supports coordination with other radio-band searches, repeat-burst investigations, and searches for electromagnetic or gravitational-wave counterparts.
  • The alert system will use the VOEvent implementation recommended by Petroff et al. to remain compatible with FRB searches worldwide.
  • After initial human confirmation, CHIME/FRB plans to announce every detected FRB with its significance and other relevant characteristics in real time.The planned latency from arrival of the lowest-frequency signal is approximately 2–3 s.

5.2. Calibration, Sensitivity, & Completeness Measurement

CHIME/FRB plans calibration, monitoring, and injection-based testing to quantify survey exposure, sensitivity, and completeness over time, sky location, and burst properties.

  • The calibration and monitoring program has three stages: CHIME phase calibration, pipeline-performance monitoring, and sensitivity and completeness measurement.
  • Calibration: Complex gains will be determined near real time from transits of bright effective point sources such as Cassiopeia A and Cygnus A.
  • Sensitivity and completeness: FRB-rate and property-distribution measurements require sensitivity and exposure tracked by time and sky location, plus completeness characterized by fluence, width, and DM.
  • Exposure monitoring: Exposure accounting tracks usable feeds, active L0 nodes, UDP packet loss, masked data, active searching nodes, reporting nodes, and TCP delays.
  • Exposure monitoring: Metrics will be reported every few minutes, stored in a time-tagged database, and scraped to calculate CHIME/FRB sky sensitivity and exposure.
  • Injection testing: Known-property FRB injections will test RFI excision and filtering while accounting for coherent-beam spatial and spectral responses.
  • Injection testing: A randomly selected 4×4 grid among 1024 coherent beams will sample different North-South locations in the primary beam.

5.3. Planned Triggered Baseband Recording System

The planned triggered baseband system will preserve high-resolution FRB data for improved localization and detailed studies of burst structure, propagation, polarization, and magnetic environments.

  • Triggered baseband recordings will provide polarization sensitivity, improved source localization, finer spectral and temporal resolution, and opportunities for FRB VLBI.
  • System design: Baseband data from all 2048 correlator inputs will be buffered before beamforming and written to disk after an FRB-search trigger.
  • Localization: Interferometric baseband localization will improve on formed-beam width, yielding arcminute uncertainties for most bursts.
  • Time resolution: Coherent dedispersion will expose FRB structure well below the search engine’s 1-ms resolution, constraining emission physics.
  • Propagation effects: Baseband data will enable detailed measurements of scattering and scintillation timescales that constrain FRB astrophysical environments.
  • Polarization: Full-Stokes data will measure polarization and, for linearly polarized bursts, permit Faraday rotation-measure calculations probing the source environment, host galaxy, and intervening medium.

7. PREDICTED CHIME FRB EVENT RATE AND ENVISIONED FRB SCIENCE

CHIME/FRB targets the poorly understood low-frequency FRB population with broad sky coverage and revised rate estimates. Its expected detections could support population studies, localization follow-up, and redshift measurements, while angular resolution limits host identification alone.

  • None of the observed FRBs had been detected below 400 MHz, leaving the causes of the apparent low-frequency dearth unresolved.
  • 0.6–11 bursts per day are predicted for CHIME/FRB using the revised all-sky rate, Euclidean flux distribution, 3-ms intrinsic widths, and Galactic-like scattering timescales.The estimate could be 0.3–2 bursts per day if FRBs are detectable only in the 700–800 MHz portion of the band.
  • Thousands of FRBs could be detected over an envisioned ∼3 yr project lifetime, enabling uniform-sensitivity studies of sky, DM, and logN/logS distributions.The dataset could also support correlations among DM, flux, scattering measure, and width, as well as cosmological and intergalactic-medium studies.
  • CHIME/FRB’s angular resolution alone is insufficient for host-galaxy identification, but repeat bursts can be followed up interferometrically for localization and host identification.VLBI using CHIME/FRB baseband data and outtrigger stations are additional localization avenues.
  • H I absorption could independently provide host-galaxy redshifts for bright FRBs, with SNR > ∼15 expected approximately 10% of the time under specified assumptions.The method likely fails for FRBs in dwarf galaxies such as the host of FRB 121102, and spectral stacking could help for repeaters.

8. CONCLUSIONS

The CHIME/FRB project extends a telescope designed for independent science into a real-time FRB detector. Its broad survey capabilities and follow-up strategy could substantially advance understanding of FRBs, although host-galaxy and redshift identification require additional observations.

  • CHIME/FRB has the potential to yield major progress in understanding the FRB mystery.
  • CHIME’s software-based architecture is presented as scalable with computing and storage growth, potentially enabling deeper future FRB surveys.
  • CHIME/FRB cannot identify host galaxies and redshifts alone because of its relatively poor localization capabilities.Interferometric follow-up, especially for repeating FRBs, is being pursued as a localization path.
  • The telescope’s utility also includes simultaneous pulsar timing through CHIME/Pulsar.
Loading 1803.11235v1…