Source-linked AI summary
Autoencoder-based Radio Frequency Interference Mitigation For SMAP Passive Radiometer
Ali Owfi, Fatemeh Afghah
TL;DR
SMAP’s protected-band passive radiometer measurements face RFI, and current mitigation discards flagged data, risking loss of valuable information. The paper proposes an autoencoder to remove dominant known-source interference from contaminated signals. Simulation results suggest effective dominant-RFI removal and potential preservation of information that would otherwise be discarded.
Problem
SMAP’s current RFI mitigation discards flagged contaminated measurements, causing substantial data loss as active-user coexistence is considered for the protected band.
Source: §1 · pp. 1–2Method
The paper trains a denoising autoencoder on simulated contaminated signals and ground-truth RFI to reconstruct and subtract dominant known-source interference.
Source: §1 · p. 2, §2.2 · p. 2Results
Simulation results suggest the autoencoder can effectively remove dominant RFI from the received signal in a 5G-base-station interference scenario.
Source: §4 · p. 4Takeaways & Limitations
The method potentially preserves information that would otherwise be discarded by SMAP’s interference-mitigation system.
Source: §4 · p. 4
Abstract
from arXiv · showhide
Passive space-borne radiometers operating in the 1400-1427 MHz protected frequency band face radio frequency interference (RFI) from terrestrial sources. With the growth of wireless devices and the appearance of new technologies, the possibility of sharing this spectrum with other technologies would introduce more RFI to these radiometers. This band could be an ideal mid-band frequency for 5G and Beyond, as it offers high capacity and good coverage. Current RFI detection and mitigation techniques at SMAP (Soil Moisture Active Passive) depend on correctly detecting and discarding or filtering the contaminated data leading to the loss of valuable information, especially in severe RFI cases. In this paper, we propose an autoencoder-based RFI mitigation method to remove the dominant RFI caused by potential coexistent terrestrial users (i.e., 5G base station) from the received contaminated signal at the passive receiver side, potentially preserving valuable information and preventing the contaminated data from being discarded.
1. INTRODUCTION
SMAP’s protected L-band radiometer measurements are vulnerable to RFI, while future spectrum sharing with active terrestrial users could increase coexistence challenges. Existing SMAP mitigation discards flagged data, motivating a method to remove known-source interference while preserving contaminated measurements.
Protected-band context
- The 1400–1427 MHz band supports SMAP and other remote-sensing measurements, and surface emissivity at L-band is sensitive to soil moisture.The band was set aside for radio astronomy, space research, and geophysical remote sensing; SMAP and SMOS obtain measurements relevant to soil moisture and salinity.Source: §1 · p. 1
Spectrum-sharing motivation
- Spectrum sharing between active and passive users is under study, but protecting orbiting satellites may require dynamic sharing and radiometer-side RFI mitigation.The paper notes that the 1400–1427 MHz band is of interest for active communications, including for rain resiliency and lower-cost equipment.Source: §1 · p. 1
Current SMAP RFI mitigation
- SMAP uses multiple statistical detectors whose logical OR flags contaminated subsamples, after which flagged data are discarded and the remaining data are averaged.The paper describes this as causing substantial data loss and motivating methods that preserve information from contaminated measurements.Source: §1 · pp. 1–2
Prior 5G coexistence work
- A prior study of 5G downlink coexistence focused on scheduling access for one base station or a close cluster, while future active-user RFI motivates cancellation at SMAP.The authors state that transmitter scheduling alone may not suffice if exclusive band protection ends.Source: §1 · p. 2
Paper approach
- The proposed autoencoder targets dominant RFI from a known source such as a 5G base station, using simulated contaminated signals and ground-truth RFI for training.The authors distinguish this setup from prior SMAP detection work using real data with unknown RFI sources.Source: §1 · p. 2
2. METHODOLOGY
The methodology models a terrestrial 5G cluster’s interference at SMAP and uses a denoising autoencoder to reconstruct the desired signal from contaminated I/Q samples. The encoder compresses the input to a 512-element latent representation, and the decoder reconstructs the samples.
2.1. Communication Model
- With unobstructed line of sight, the communication model treats free-space path loss as the main attenuation and atmospheric attenuation as negligible.The model relates received power to transmitter and receiver antenna gains, carrier frequency, and satellite–base-station distance.Source: §2.1 · p. 2
- Because SMAP’s orbit is fixed at 685 km, the model uses elevation angle to determine satellite–base-station distance for the path-loss calculation.Figure 1 labels the geometry variables d, e, h, and R; the text describes calculating angle a and then distance d.Source: §2.1 · p. 2, Figure 1 · p. 2
2.2. Autoencoder-based Interference Mitigation
- The method reconstructs and subtracts dominant known-source interference rather than detecting and discarding signals contaminated by unknown RFI.The autoencoder maps input to a lower-dimensional latent representation and decodes that representation back into a signal.Source: §2.2 · p. 2
- The contaminated signal is the denoising autoencoder’s input, and the clean signal is its reconstruction target rather than the input itself.This training objective is intended to teach reconstruction of the clean signal from the contaminated one.Source: §2.2 · pp. 2–3
- The encoder maps 2 × 1000 I/Q samples through three convolutional and three dense layers to a 512-element latent representation, which the decoder uses to reconstruct 2 × 1000 samples.The decoder reverses the encoder’s dense and convolutional stages.Source: §2.2 · p. 3, Figure 3 · p. 3
3. SIMULATION AND RESULTS
The simulations model severe 5G interference that dominates thermal emission and evaluate whether the autoencoder can recover the RFI-free signal. The reported reconstructed amplitude is close to the ground truth amplitude.
Simulation setup
- The simulated dataset uses a 5G RFI power of −140 dB for d = 800 km, fc = 1.4 GHz, and Gt = 12 dB, and thermal-emission power of −144 dB at T = 300 K.The paper uses simulated data because training requires both contaminated and clean signals, which are not available together from the collected SMAP measurements.Source: §3 · pp. 3–4
Simulation results
- The reconstructed RFI-free signal’s amplitude is close to the ground truth amplitude, suggesting that the method mitigates dominant RFI.Figure 5 compares the reconstructed signal with ground truth and the contaminated signal.Source: Figure 5 · p. 4, §3 · p. 4
Simulation setup
- In the severe-interference evaluation, the 5G signal power is −50 dB and natural-emission signal power is −54 dB.The method reconstructs the interfering 5G signal from the contaminated input and subtracts it, leaving thermal emission plus minor unknown RFI sources for SMAP’s existing mitigation techniques.Source: §3 · p. 4
4. CONCLUSION
The paper reports simulation-based evaluation of an autoencoder that extracts dominant 5G interference from signals received by SMAP’s passive radiometer. The results suggest effective removal and potential preservation of information otherwise discarded.
Conclusion
- The proposed method reconstructs the received signal by extracting dominant RFI in a scenario where a 5G base station interferes with SMAP’s passive radiometer.The paper reports performance evaluation using simulation data.Source: §4 · p. 4
- The simulation results suggest the autoencoder can effectively remove dominant RFI from the received signal and potentially preserve information that SMAP’s mitigation would otherwise discard.The evaluated scenario has a 5G base station causing dominant interference on SMAP’s passive radiometer.Source: §4 · p. 4
- The authors identify validation with testbeds collecting remote-sensing data with ground truth as a future step, once such testbeds become available.Source: §4 · p. 4