Sample Reduction
Downsampling operations remove redundant samples from a digitized waveform to lower the data throughput. The decimation ratio defines the integer or fractional factor by which the original sampling frequency is divided. It dictates the new Nyquist frequency for the system.
Signal Resolution
Digital filtering before the sample removal increases the effective number of bits by spreading the quantization noise over a wider band. If the decimation ratio is high, the resulting process gains significant signal to noise ratio improvements. This improvement assumes the input signal contains white noise.
Bandwidth Conservation
Narrowing the observation window allows a processor to handle data from multiple sensors simultaneously without exceeding the bus capacity. A specific decimation ratio allows a high speed analog converter to interface with a slower microcontroller. The hardware must apply a low pass filter to prevent frequencies above the new folding frequency from contaminating the baseband.
This filtering step ensures that the decimation ratio does not introduce artifacts into the measurement.
Hardware Impact
Memory requirements for buffer storage decrease proportionally with the reduction in the sample count. Processing cycles are saved because the downstream algorithms operate on fewer data points per second. The decimation ratio directly influences the latency of the digital signal path.
