The efficient implementation of connected filters is an important issue in
signal processing. A typical example is the cascade of two filters, e.g., a
n adaptive filter with a time-invariant prefilter, The filtering and adapta
tion is carried out very efficiently in the frequency domain whenever filte
rs with many coefficients are required. This is implemented as a block algo
rithm by using overlap-save or overlap add techniques. However, in many rea
l-time applications also, a short latency time through the system is requir
ed, which leads to a degradation of the computational efficiency, Partition
ed frequency-domain adaptive filters, also known as multidelay adaptive fil
ters, provide an efficient way for the filtering and adaptation with long f
ilters maintaining short professing delays.
This paper shows a computationally efficient way of implementing two or mor
e partitioned frequency-domain filters in cascade or in parallel when their
filter lengths are large. The methods presented require only one fast Four
ier transform (FFT) and one inverse fast Fourier transform per input and ou
tput port, respectively, The FFT size can be even smaller than the length o
f the filters. The filters can be either time invariant or adaptive.