Frequency tracking and variable bandwidth for line noise filtering without a reference

This paper presents a method for filtering line noise using an adaptive noise canceling (ANC) technique. This method effectively eliminates the sinusoidal contamination while achieving a narrower bandwidth than typical notch filters and without relying on the availability of a noise reference signal as ANC methods normally do. A sinusoidal reference is instead digitally generated and the filter efficiently tracks the power line frequency, which drifts around a known value. The filter´s learning rate is also automatically adjusted to achieve faster and more accurate convergence and to control the filter´s bandwidth. In this paper the focus of the discussion and the data will be electrocorticographic (ECoG) neural signals, but the presented technique is applicable to other recordings.