Decision Feedback-Aided Known-Interference Cancellation

Known-interference cancellation (KIC) in combination with cooperative jamming can be used to provide covertness and security to wireless communications at the physical layer. However, since the signal of interest (SI) of a wireless communication system acts as estimation noise, i.e., interference, to KIC, the SI limits the extent to which the known interference (KI) can be canceled and that in turn limits the throughput of the wireless communication system that is being hidden or secured. In this letter, we analyze a decision feedback-aided known-interference cancellation (DF-KIC) structure in which both the KI and SI are canceled iteratively and successively. Measurement results demonstrate that introducing decision feedback to KIC improves its KI cancellation capability and hence increases the wireless communication system’s useful throughput, albeit at the expense of a higher computational load.

💡 Research Summary

This paper addresses a critical limitation in physical-layer security techniques that use cooperative jamming combined with known-interference cancellation (KIC). While KIC allows authorized receivers to cancel out a pre-shared “known interference” (KI) signal, thereby hiding or securing the actual “signal of interest” (SI), its performance is fundamentally capped. The SI itself acts as estimation noise during the adaptive filtering process used to estimate and cancel the KI channel. This residual KI limits the post-cancellation signal-to-interference-and-noise ratio (SINR) and, consequently, the throughput of the covert communication system.

To overcome this, the authors propose a Decision Feedback-aided KIC (DF-KIC) structure. The core innovation is an iterative, successive cancellation scheme that targets both the KI and the SI. The algorithm begins with standard KIC (using a Variable Step Size Frequency-Offset-compensated Least Mean Squares - VSS-FO-LMS - algorithm) to produce an initial estimate of the SI. It then demodulates both the original received signal and this KIC-output signal, selecting the symbol stream with the better quality metric (using non-data-aided Error Vector Magnitude as a proxy for SINR). This selected stream is re-modulated to create an estimate of the transmitted SI waveform. This estimated SI is then fed into another adaptive filter (VSS-FO-LMS) to estimate and subtract the SI component from the original received signal, creating an “SI-reduced” signal. Crucially, with the SI’s interfering effect diminished, the KI channel can be re-estimated much more accurately from this cleaner signal. This new KI estimate is then subtracted, completing one feedback iteration. The process repeats, using warm-restart initializations for the filter parameters, until a quality threshold is met, performance stops improving, or a maximum iteration count is reached.

The proposed method was rigorously evaluated using a laboratory testbed with three USRP-2900 software-defined radios connected via cables, simulating realistic conditions with carrier frequency offsets, sampling frequency offsets, and phase noise. The KI was a band-limited pseudo-random signal, and the SI was an OFDM signal with modulation orders ranging from 4-QAM to 256-QAM.

Key experimental findings include:

1. SI Limits Basic KIC: Measurements confirmed that the presence of a strong SI severely limits the cancellation depth achievable by basic KIC, leaving residual KI well above the noise floor (Figure 3).
2. DF-KIC Enhances Cancellation: DF-KIC significantly outperforms basic KIC, suppressing the residual KI much closer to the noise floor and enabling successful demodulation of the SI even when it was initially buried under interference (Figure 4).
3. Modulation-Dependent Benefit: The advantage of DF-KIC becomes essential for higher-order modulations (e.g., 128-, 256-QAM) that require high SINR. For these cases, basic KIC alone is often insufficient to reach a target symbol error rate (SER) of 10^-3, while DF-KIC achieves it, albeit requiring more feedback iterations on average (Figure 5).
4. Throughput Gain: The improved cancellation directly translates into higher useful throughput (Goodput). DF-KIC can increase the maximum achievable goodput by up to 35% compared to basic KIC in scenarios with powerful KI (Figure 6).

The paper concludes that DF-KIC effectively mitigates the mutual interference problem between KI and SI in cooperative jamming scenarios, leading to superior interference cancellation and higher communication throughput. This performance gain comes at the cost of increased computational complexity due to the iterative processing and multiple runs of the adaptive filtering algorithm. The work provides a validated, practical enhancement to KIC-based physical-layer security systems.