Polyphase Filter: What It Is and How It Works

A polyphase filter is a signal processing technique that splits a filter into multiple smaller filters called phases to efficiently process signals, especially for changing sample rates. It reduces computation by handling parts of the signal separately and combining results, often used in multirate filtering and sample rate conversion.

How It Works

Imagine you want to filter a long stream of data but only need to keep every few samples or change the speed of the data. Instead of filtering the whole stream at once, a polyphase filter breaks the filter into smaller parts called phases. Each phase handles a portion of the input data, like a team of workers each doing a small job.

This splitting lets the filter work more efficiently because it avoids repeating calculations on samples that will be discarded later. The results from each phase are then combined to produce the final filtered output. This is especially useful when changing the sample rate, like speeding up or slowing down audio or sensor data.

Example

This example shows how to create a simple polyphase filter for downsampling a signal by a factor of 2 using Python and SciPy.

python

import numpy as np
from scipy.signal import firwin, upfirdn

# Create a sample signal: a sine wave plus noise
fs = 1000  # original sample rate
f = 50     # signal frequency
t = np.arange(0, 1.0, 1/fs)
signal = np.sin(2 * np.pi * f * t) + 0.1 * np.random.randn(len(t))

# Design a lowpass FIR filter for downsampling
numtaps = 64
cutoff = 0.4  # normalized cutoff frequency (0.5 is Nyquist)
fir_coeff = firwin(numtaps, cutoff)

# Use polyphase filtering to downsample by 2
downsample_factor = 2
filtered_downsampled = upfirdn(fir_coeff, signal, down=downsample_factor)

print(filtered_downsampled[:10])

Output

[ 0.00426343  0.02762407  0.06613627  0.10239352  0.13425194  0.15748449
  0.16992754  0.17219945  0.16713446  0.15740106]

When to Use

Use polyphase filters when you need to change the sample rate of a signal efficiently, such as in audio processing, telecommunications, or sensor data analysis. They are ideal for downsampling (reducing sample rate) or upsampling (increasing sample rate) without losing important signal information.

For example, streaming music services use polyphase filters to convert audio between different sample rates smoothly. Also, in wireless communication, polyphase filters help process signals at different speeds to save computing power and battery life.

Key Points

• Polyphase filters split a filter into smaller parts to reduce computation.
• They are commonly used in sample rate conversion tasks.
• They improve efficiency by avoiding unnecessary calculations on discarded samples.
• Widely applied in audio, communications, and sensor signal processing.

Key Takeaways

Polyphase filters improve efficiency by splitting filtering into smaller phases.

They are essential for efficient sample rate conversion in signal processing.

Use them when upsampling or downsampling signals to save computation.

Common in audio processing, telecommunications, and sensor data applications.