How does multipath propagation affect the accuracy of direction finding at microwave frequencies?

Multipath propagation is the most significant source of bearing error in practical direction finding systems at microwave frequencies. It occurs when the signal arrives at the DF array via multiple paths (direct and reflected), creating a composite wavefront that distorts the measured phase and amplitude: (1) Mechanism: the direct signal arrives at angle θ_direct from the emitter. One or more reflected signals arrive at different angles θ_reflected (from buildings, terrain, vehicles, ship structures). The DF system measures the combined wavefront (the vector sum of direct and reflected signals). The measured AOA is displaced from the true AOA by an amount that depends on: the reflected signal amplitude relative to the direct signal, the angle difference between direct and reflected paths, and the phase relationship between the signals (which varies with frequency). (2) Bearing error: for a single reflection with amplitude ratio ρ (reflected/direct) and angle separation Δθ: the maximum bearing error ≈ arctan(ρ × sin(Δθ) / (1 + ρ × cos(Δθ))). For ρ = 0.5 (-6 dB reflection) and Δθ = 30°: max error ≈ arctan(0.25/1.43) ≈ 10°. For ρ = 0.1 (-20 dB reflection): max error ≈ 3°. Even weak reflections can cause significant bearing errors. (3) Frequency dependence: the phase between the direct and reflected signals varies with frequency (because the path length difference translates to a frequency-dependent phase shift). The bearing error oscillates as the frequency changes. At one frequency: the error may be +5°. At a slightly different frequency: -5°. This means: wideband signals experience less multipath error than narrowband signals (the errors average out across the bandwidth). Frequency diversity: measuring the bearing at multiple frequencies and averaging reduces the multipath-induced error. (4) Mitigation techniques: site selection (mount the DF array on a clear elevated position with minimal nearby reflectors), absorber material (place RF absorber around the array to reduce local reflections), spatial smoothing (use multiple sub-arrays and average the results; multipath-induced errors decorrelate between sub-arrays), wideband processing (compute the bearing across the full signal bandwidth and average), and super-resolution algorithms (MUSIC, ESPRIT can separate the direct and reflected signal components and estimate the bearing of each).