The electrocardiogram (ECG) is the recording of the heart’s electrical potential versus time. Three types of predominant noise that commonly contaminate the signal are baseline wander (BW) noise, electromyographic (EMG) interference, and 50 or 60 Hz power line interference. Among them power line interference is a significant source of noise. Cables carrying ECG signals from the patients to the monitoring equipment are susceptible to electromagnetic interference (EMI) from the 50/60Hz power line noise.

Many methods were proposed in the past for the removal of power line interference in the ECG. They can be categorised into non-adaptive and adaptive filtering. The non-adaptive filtering approach employs a sharp notch filter and is easy to implement at low cost. However its performance depends on the frequency stability of the power line. Adaptive filtering, on the other hand, is able to effectively remove time-varying power line frequency, but requires considerable computational power which is not suitable for portable battery powered ECG devices. To lower power consumption we opt for non-adaptive filtering based upon a multiplication-free linear phase FIR filter. The main advantage of using an FIR filter is that it minimises waveform distortion. The multiplication-free filter is able to achieve low cost and low power consumption.

The filter was designed to have the following specifications: passband from 0.5 to 100 Hz; notch frequency is selectable at either 50 or 60 Hz; and the sampling frequency is 600Hz. The schematic of the proposed filter is shown in Figure 1, whilst the frequency responses of its subfilters are shown in Figure 2. The new structure makes use of an interpolated recursive running sum (RRS) filter, H_RRS(z^M), to alternate the notch frequency. By replacing each delay element in a RRS by M delays, a multi-band filter is formed whose transition bandwidth is M times narrower than that of H_RRS(z). By setting the interpolation factor (M=6 or M=5), the proposed filter produces a notch at either 50Hz or 60 Hz. A lowpass filter, F(z), is used to remove the unwanted passbands of H_RRS(z^M) beyond 100 Hz. F(z) needs to be very sharp and the complexity of such a filter can be very high if the conventional filter design method is used, we chose to use the RRS filter of order 4.

The resulting filter structure involved the use of 8 switches, 37 adders and 2248 delays and was stimulated using Matlab Simulink. In practice, the 8 switches can be controlled by one main switch. The results show that the proposed filter effectively removes the power line noise from the ECG signal.