Adaptive cancellation of geomagnetic background noise using a sign-error normalized LMS algorithm

Magnetic anomaly detection (MAD) systems are often used aboard aircraft to detect transient magnetic fields produced by submarines. In this operating environment, the earth´s magnetic field produces a large undersirable noise component at the detector. An adaptive filtering scheme for canceling geomagnetic background noise in the desired signal using a second MAD sensor aboard a separate aircraft as a reference is demonstrated. It is shown that conventional adaptive algorithms such as the normalized least-mean-square (NLMS) algorithm partially cancel the desired transient magnetic anomalies, thus reducing detection performance. The sign-error NLMS algorithm, an algorithm that is capable of preserving significant transients in the desired response, is presented. Using actual data measured aboard twin airborne MAD systems, the noise cancellation performances of these two algorithms are compared. Receiver operating characteristics indicate that the sign-error NLMS algorithm provides a higher detection likelihood and a lower false alarm rate than the NLMS algorithm.<>