표 1. | Table 1. 기존 및 제안된 필터 비교 | Comparison of EKF-CA and proposed R2-AEKF.

Category	EKF-CA [Conventional]	R2-AEKF [Proposed]
Process noise Q	Fixed, derived from white-jerk PSD, identical for x/y axes	Anisotropic shaping based on maneuver indicator (axial vs. lateral, velocity-based)
Measurement noise R	Fixed variances for range and bearing	Channel-wise adaptive rescaling (innovation EWMA, deadband, rate-limit, robust gating)
Estimation property	Stable in nominal conditions, degraded under abrupt maneuvers	Similar to EKF-CA in nominal conditions, more robust during rapid turn or bursts
Computational cost	Low	Slightly increased (scale adaptation overhead), still real-time feasible
Suitable scenarios	Straight flight, smooth turns	Abrupt turns, bursts, jerk, varying, noise statistics