Post on 18-Jan-2016
BIEN425 – Lecture 15
• By the end of this lecture, you should be able to:– Design and implement integer decimators and
interpolators– Design and implement a narrow band filter using
interpolation and decimation techniques.
To resample a digital signal
• The simplest way:
• It will introduce additional quantization noise and aliasing noise
• Computationally intensive
DAC ADC
Decimation
• Decreasing sampling rate
• fM = fs/M
• Or simply taking every M samples (decimation)
• However, we will need to consider an anti-aliasing filter
)()( MkxkxM
Digital anti-aliasing filter• We can consider this as FIR filter
• Where HM(f) =
• y(k) =
HM(z) M
Intepolation
• Increasing sampling rate
• fL = Lfs
• Observe here we are using zero-padding
otherwise
LLkLkxkxL ,0
...2,,0),/()(
Effect in frequency domain
• Observe XL(f) = X(Lf)
• This means that frequency content is the power of xL(k) is 1/L times the original x(k)
• Need to compensate for the effect of 1/L in the anti-imaging filter
• Where HL(f) =
• y(k) =
L HL(z)
Example
• Lecture15.m
Rational sampling rate converter
L HL(z) HM(z) M
• fnew = (L/M)fs
• Combine HL and HM to form H0
• Since HL and HM are both LP
• H0(f) =
• y(k)=
Narrow band filter
• Definition: sharp filter whose passband or stopband is small in comparison with sampling frequency
• Usually need high-order FIR filters
• Example: Ideal response
• Reduce sampling rate by M
• Then create new filter G(f)
• Then interpolate by M again
G(z) MMHM(z) HM(z)