Wideband Real-Time Adaptive Full-Duplex Radios
Abstract
Full-duplex (FD) wireless enables simultaneous transmission and reception on the same frequency channel, enhancing spectrum and power efficiency provided self-interference (SI) is reliably suppressed. Despite significant research into SI cancellation (SIC), real-world FD deployment remains an open challenge. The highly dynamic nature of SI, especially at higher bandwidths, requires system-level designs capable of adapting to a changing electromagnetic environment in order to sustain cancellation. We present a wideband, adaptive FD radio that combines a highly configurable RFIC canceller with a novel, device-specific, sparsity-based configuration algorithm for real-time SIC tracking. The proposed data-driven algorithm exploits measured tap responses of the RFIC canceller to account for hardware imperfections and is implemented on an FPGA as part of an SDR-based FD radio. In VNA-based validation, this algorithm outperforms theory-based and literature-inspired baselines across bandwidths up to 100 MHz. In the SDR-based FD radio, the FPGA-controlled RFIC canceller with linear digital SIC achieves 72 dB of overall cancellation across 100 MHz, reducing SI to the noise floor. Under mobility, dynamic optimization outperforms static routines and maintains noise-floor-level cancellation despite variations in the SI channel. Finally, an FD link between two SDRs exhibits <0.8 dB EVM degradation relative to an equivalent half-duplex link, confirming that a bidirectional FD link can approximately double throughput.