Adaptive Multipath Interference Cancellation in Wi-Fi CSI-Based Cardiorespiratory Monitoring: A Kalman Filter Approach with Dynamic Subcarrier Selection

Wi-Fi Channel State Information (CSI) enables contactless vital sign extraction but suffers from severe multi-path interference in indoor environments. This paperproposes an adaptive framework combining Extended Kalman Filtering (EKF) with dynamic subcarrier selection (DSS) to enhance cardiac signal recovery. We formulate multipath propagation as a time-varying state-space model and derive optimal subcarrier weighting coefficients via constrained convex optimization. Theoretical analysis establishes Cram´er–Rao Lower Bounds (CRLB) for inter-beat interval (IBI) estimation under additive complex Gaussian noise and multipath fading with Rice factor K ∈ [2, 15] dB. Simulation results demonstrate 34% reduction in IBI estimation error (RMSE: 8.2 ms vs. 12.5 ms baseline) and 28% improvement in Signal-to-Interference-plus-Noise Ratio (SINR) across 10,000 Monte Carlo trials with realistic IEEE 802.11n channel models. The proposed algorithm achieves real-time performance (latency < 120 ms per 1-second window) on commodity ESP32 hardware, enabling robust deployment in cluttered indoor environments without human subject experimentation.