Uploaded image for project: 'SAFe Program'
  1. SAFe Program
  2. SP-1673

Initial prototype of RT RFI flagging pipeline functionality (considering interfaces that can be tested with the AA0.5 system)

Details

    • Enabler
    • Should have
    • PI11
    • COM SDP SW
    • None
    • Data Processing
    • Hide

      If we develop a prototype real-time RFI pipeline, we can understand and develop the required interfaces for the early AA releases, which will provide high-throughput and effective removal of RFI to optimise calibration.

      Show
      If we develop a prototype real-time RFI pipeline, we can understand and develop the required interfaces for the early AA releases, which will provide high-throughput and effective removal of RFI to optimise calibration.
    • Hide
      • The prototype code should be provided within a suitable repository with required documentation for usage with information of how to access the simulation test data
      • Documentation in confluence reporting on the investigation of interfaces
      Show
      The prototype code should be provided within a suitable repository with required documentation for usage with information of how to access the simulation test data Documentation in confluence reporting on the investigation of interfaces
    • 3
    • 3
    • 8.333
    • Team_HIPPO
    • Sprint 5
    • Hide

      The prototype of the pseudo-real-time pipeline was implemented using C++ and stored in the SKA repository. Input data format was chosen to be the measurement set from CASA as it is a standard, and simulated data from the SIM team was provided in Measurement set format. The RFI algorithm itself is an initial implementation of the sum threshold algorithm. The implementation can read input data in a measurement set format, perform RFI mitigation and update the input measurement set data with RFI flags.

      The data used for testing is a simulation of a Perth DTV transmitter. A basic form of tests was performed where each flagged visibility is checked with the RFI location, which the SIM team also provided. The current test shows that 30% of flags correctly flags RFI present in the input data, while 70% of flags were false positive. However, the RFI pattern is switched on and off randomly, and the sum threshold algorithm flags larger chunks of data which may cause this discrepancy. Since the RFI mitigation implemented contains only a sum threshold algorithm, it is expected that RFI mitigation will improve with more advanced methods.

      Repository: https://gitlab.com/ska-telescope/ska-post-correlation-rfi-flagger/-/tree/master
      RFI mitigation description: https://confluence.skatelescope.org/display/SE/Pseudo+Real-time+RFI+Flagging+Prototype+Design
      Summary of implemented RFI flagger: https://confluence.skatelescope.org/display/SE/Initial+Implementation+of+the+RFI+Flagger
      Data used: https://confluence.skatelescope.org/display/SE/RFI+simulations+for+the+Real-Time+pipeline

      Future work:
      Investigate resulting flags to identify the source of a large number of false-positive flags.
      Investigate simulated data and correct RFI flags provided by SIM for discrepancies
      Implement more advanced RFI mitigation algorithms

      Show
      The prototype of the pseudo-real-time pipeline was implemented using C++ and stored in the SKA repository. Input data format was chosen to be the measurement set from CASA as it is a standard, and simulated data from the SIM team was provided in Measurement set format. The RFI algorithm itself is an initial implementation of the sum threshold algorithm. The implementation can read input data in a measurement set format, perform RFI mitigation and update the input measurement set data with RFI flags. The data used for testing is a simulation of a Perth DTV transmitter. A basic form of tests was performed where each flagged visibility is checked with the RFI location, which the SIM team also provided. The current test shows that 30% of flags correctly flags RFI present in the input data, while 70% of flags were false positive. However, the RFI pattern is switched on and off randomly, and the sum threshold algorithm flags larger chunks of data which may cause this discrepancy. Since the RFI mitigation implemented contains only a sum threshold algorithm, it is expected that RFI mitigation will improve with more advanced methods. Repository: https://gitlab.com/ska-telescope/ska-post-correlation-rfi-flagger/-/tree/master RFI mitigation description: https://confluence.skatelescope.org/display/SE/Pseudo+Real-time+RFI+Flagging+Prototype+Design Summary of implemented RFI flagger: https://confluence.skatelescope.org/display/SE/Initial+Implementation+of+the+RFI+Flagger Data used: https://confluence.skatelescope.org/display/SE/RFI+simulations+for+the+Real-Time+pipeline Future work: Investigate resulting flags to identify the source of a large number of false-positive flags. Investigate simulated data and correct RFI flags provided by SIM for discrepancies Implement more advanced RFI mitigation algorithms
    • 14.2
    • Stories Completed, Integrated, Outcomes Reviewed, Demonstrated, Accepted by FO
    • PI22 - UNCOVERED

    Description

      A minimal prototype making use of the simulations developed under SP-1670 to provide real-time flagging functionality. This will likely focus on the interfaces that will be required for use with the AA0.5 system. Given the real-time needs, this will necessitate use of relatively simple (in comparison to the Batch RFI processing) algorithms to be performant. This could consider current implementations of similar systems at e.g. LOFAR and MeerKAT. This work can potentially make use of the output from SP-1554 which includes initial assessment of the MeerKAT RFI processing pipelines https://confluence.skatelescope.org/display/SE/RFI+Management+System+in+MeerKAT.

      This is not expected to be integrated into the SDP system in PI11, but must be working towards achieving this by PI13-14 (or before if possible). 

      Attachments

        Issue Links

          Structure

            Activity

              People

                D.Fenech Fenech, Danielle
                f.graser Graser, Ferdl
                Votes:
                0 Vote for this issue
                Watchers:
                0 Start watching this issue

                Feature Progress

                  Story Point Burn-up: (100.00%)

                  Feature Estimate: 3.0

                  IssuesStory Points
                  To Do00.0
                  In Progress   00.0
                  Complete1338.0
                  Total1338.0

                  Dates

                    Created:
                    Updated:
                    Resolved:

                    Structure Helper Panel