4928 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 51, NO. 10, OCTOBER 2013 Real-Time RFI Detection and Mitigation System for Microwave Radiometers Giuseppe F. Forte, Member, IEEE, Jorge Querol, Adriano Camps, Fellow, IEEE, and Mercè Vall-llossera Abstract—Microwave radiometers are very sensitive passive sensors that measure the power of the thermal noise within a deter- mined bandwidth. Therefore, any other signal present in the band modifies the value of the measured power, and the corresponding estimated antenna temperature, from which the geophysical pa- rameters are retrieved. Due to the high sensitivity and accuracy required for these instruments, radio frequency interference (RFI) is becoming more and more a serious problem. On one hand, ground-based or global RFI surveys are helping to understand the occurrence and types of RFI sources. If RFI does not necessarily affect the whole bandwidth, or it is not present during the whole integration time, the application of either frequency blanking, time blanking or signal spectrogram techniques can be applied. However, it would be desirable to apply techniques to estimate the RFI signal so that it can be subtracted from the received signal itself so that some useful measurements are still possible. Such a real-time system is currently being developed for RFI detection and mitigation. This work focuses however in the description and performance of a wavelet-based RFI-mitigation technique imple- mented in a FPGA hardware back-end. The interfering signal is estimated by using the powerful denoising capabilities of the wavelet transform, and it is then subtracted from the total received signal to obtain a RFI-mitigated noise signal. Index Terms—Denoising, detection, microwave radiometry, mitigation, radio frequency interference (RFI), wavelet. I. I NTRODUCTION M ICROWAVE radiometry is routinely used today to ob- tain a number of geophysical parameters. Since it mea- sures the thermal noise power, microwave radiometers are highly sensitive and accurate passive instruments. Radio fre- quency interference (RFI) does not only concern microwave radiometers, although they are more prone to suffer from RFI, since they are passive sensors. Due to its geographical and spectral extensions, RFI is becoming an increasing problem and efficient RFI detection and mitigation techniques are required [1]–[5]. RFI can come from spurious signals and harmonics from lower frequency bands, from spread-spectrum signals overlap- ping the “protected” bands of operation, or from out-of-band emissions not properly rejected by the pre-detection filters due Manuscript received August 30, 2012; revised March 24, 2013; accepted April 27, 2013. Date of publication July 23, 2013; date of current version September 27, 2013. This work was supported in part by funds from the Spanish projects AYA2010-22062-C05-05 and AYA2011-29183-C02-01 from the Spanish Ministry of Science and Innovation and EU Feder. The authors are with Remote Sensing Laboratory, Departament del Teoria del Senyal i Comunicacions, Universitat Politècnica de Catalunya-Barcelona Tech and IEEC/UPC, 08034 Barcelona, Spain (e-mail: giuseppe.forte@tsc. upc.edu; jquerol87@gmail.com; camps@tsc.upc.edu; merce@tsc.upc.edu). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TGRS.2013.2267595 to the finite rejection. In fact, it has also to be considered that the number of sources candidate to become unintentional interferers is large and growing. RFI sources have different signal structures, but the main effect is that they modify the detected power and the corresponding antenna temperature from which geophysical parameters are retrieved. The RFI problem is especially important in populated areas because most of the RFI comes from human activity, especially at low frequencies (e.g., L-band) due to the presence of wireless devices, computers, etc. Furthermore, RFI may be present even during the calibration, producing a systematic error in the whole data set. Several techniques to detect the presence of RFI in radiometric measurements have been developed: time and frequency domain analyses (e.g., [4]–[6]), statistical analysis of the received signal (e.g., [7]–[9]), and spectrogram analysis (e.g., [10], [11]). Some of these techniques have already been implemented in real systems. However, in all these techniques, when RFI is detected, and the signal is blanked either in the time or the frequency domain, the capability to detect the power of the signal is lost. This work is focused in the hardware implementation of a new technique for RFI detection and mitigation [12]. Its main advantage, as compared to the previous ones, is that it always produces an output result, together with an estimation of the interfering power. The most similar system is the adaptive noise cancelling [13] that consists of a separate, dedicated reference channel used to obtain an independent estimate of the RFI signal. There are two data channels: a main channel pointing to the source and containing the RFI signal; and a reference channel (separated antenna pointing off source) that contains also the RFI signal. Both channels contain the RFI signal, which are different due to the different propagation paths, but correlated as they come from the same source, making it possible to eliminate the RFI from the received signal [14]. The problem with this method compared to wavelet denoise is that RFI coming from different places complicates the system much more. In Section II, the proposed technique is described and also the implementation of the concept. The actual hardware implementation is described in Section III. Finally, hardware performance results for different input test signals are shown in Section IV, and the results and the main conclusions are summarized in Section V. II. METHODOLOGY In [12] a technique to mitigate RFI present in radiometric signals was proposed and analyzed in detail. It is based on 0196-2892 © 2013 IEEE