1
Getting RFI Mitigation into the Workplace
J.R. Fisher (NRAO) rfisher@gb.nrao.edu

Over the past few years we have seen some impressive demonstrations of excision of RFI from radio telescope data. The techniques used include parametric signal synthesis and subtraction, adaptive filtering, post correlation subtraction, and pulse blanking. However, relatively little of this work has made its way into astronomical use. The distance between a demonstrated algorithm and devices that are useful to routine observing may be greater than we tend to imply.

The signal processing power required for RFI excision will be roughly in the range of one tenth to ten times the power required to do signal processing without excision, depending on the complexity of the RFI environment and the types of signals being removed. Hence, we will be directly competing for resources with astronomers' desire for wider bandwidths and higher time resolution. This calls for a careful match between the state of our art and the RFI problems that astronomers want solved. Thus far, there is not much overlap. Full implementation of RFI mitigation means that it must be usable by the non-expert. We tend to forget how much hand tweaking has gone into the optimization of our algorithms.

In this presentation I suggest that we take a very science-driven approach to RFI mitigation starting with simple solutions that have the highest ratio of scientific impact to effort required. Even the simplest algorithms can require a lot of work to make useful to the average observer. Experience in the entire implementation process will be a valuable guide to more sophisticated development. Continued funding for RFI mitigation research must and will depend on scientifically useful products. Designers of new radio telescopes are faced with a conundrum: systems are not easily retrofitted to RFI mitigation, but a lot of experience is still needed to confidently design excision algorithms into the instruments currently on the drawing boards. Some caution is required to avoid wasting resources.