Analysis of efficiency and quality of filters suppressing RFI in radio detection of cosmic rays

Abstract Radio stations of the Auger Engineering Radio Array (AERA) investigates radio signals from coherent emissions due to geomagnetic radiation and charge excess processes. The frequency band is limited to 30 - 80 MHz. In order to improve the signal to noise ratio, RFI filters have to be used to suppress this contamination. The paper compares several (currently in use and proposed) RFI filters. We analyzed: the non-adaptive filter (IIR notch) as well as adaptive filters: IIR notch supported by the virtual NIOS® processor, FIR based on the linear prediction supported by the virtual NIOS® processor and finally the Least Mean Square filter. The LMS filter proposed for the RFI suppression should use correct learning factor, crucial for a stability of the filter, which could be determined by eigenvalues of the auto-correlation matrix built for ADC samples. For typical AERA configuration (4 beacons with fixed frequencies) the best filter seems to be the IIR as introducing very high suppression efficiency with relatively low distortions and low energy consumption. However, when unexpected RFI appears it is worth consider a combination of fixed frequency IIR + IIR filter supported with NIOS processor identifying additional RFI source as well as FIR filters. LMS filters can be used when very fast response on new sources is needed.