Upsampling

Upsampling by an integer factor

Rate increase by an integer factor ${\displaystyle L}$ can be explained as a 2-step process, with an equivalent implementation that is more efficient:

1. Expansion: Create a sequence, ${\displaystyle x_L[n],}$ comprising the original samples, ${\displaystyle x[n],}$ separated by ${\displaystyle L-1}$ zeros. A notation for this operation is: ${\displaystyle x_L[n]=x[n{]}_{\uparrow L}.}$
2. Interpolation: Smooth out the discontinuities using a lowpass filter, which replaces the zeros.

In this application, the filter is called an interpolation filter, and its design is discussed below. When the interpolation filter is an FIR type, its efficiency can be improved, because the zeros contribute nothing to its dot product calculations. It is an easy matter to omit them from both the data stream and the calculations. The calculation performed by a multirate interpolating FIR filter for each output sample is a dot product:

where the ${\displaystyle h}$ sequence is the impulse response of the interpolation filter, and ${\displaystyle K}$ is the largest value of ${\displaystyle k}$ for which ${\displaystyle h[j+kL]}$ is non-zero.

In the case ${\displaystyle L=2,}$ function ${\displaystyle h}$ can be designed as a half-band filter, where almost half of the coefficients are zero and need not be included in the dot products. Impulse response coefficients taken at intervals of ${\displaystyle L}$ form a subsequence, and there are ${\displaystyle L}$ such subsequences (called phases) multiplexed together. Each of ${\displaystyle L}$ phases of the impulse response is filtering the same sequential values of the ${\displaystyle x}$ data stream and producing one of ${\displaystyle L}$ sequential output values. In some multi-processor architectures, these dot products are performed simultaneously, in which case it is called a polyphase filter.

For completeness, we now mention that a possible, but unlikely, implementation of each phase is to replace the coefficients of the other phases with zeros in a copy of the ${\displaystyle h}$ array, and process the ${\displaystyle x_L[n]}$ sequence at ${\displaystyle L}$ times faster than the original input rate. Then ${\displaystyle L-1}$ of every ${\displaystyle L}$ outputs are zero. The desired ${\displaystyle y}$ sequence is the sum of the phases, where ${\displaystyle L-1}$ terms of the each sum are identically zero. Computing ${\displaystyle L-1}$ zeros between the useful outputs of a phase and adding them to a sum is effectively decimation. It's the same result as not computing them at all. That equivalence is known as the second Noble identity. It is sometimes used in derivations of the polyphase method.