A filter is a system that is designed to remove some component or modify some characteristic of a signal.

Several different things are introduced, including: - Finite impulse response (FIR) systems: refered as FIR filters, are systems for which each output value is the sum of a finite number of weighted values of the input sequence. - Difference equation: the basis of the input-output structure of the FIR filter as a time-domain computation. - Unit impulse response - Convolution - Linearity and time invariance - Discrete-time systems

Discrete-time systems

A discrete-time system is a computational process for transforming one sequence into another sequence. - x[n]T{}y[n]=T{x[n]}, where x[n] is the input signal and y[n] is the output signal, both of which are discrete-time signals.

The running-average (moving-average) filter

  • Difference equation, e.g., the general, causal, linear and time invariant difference equation: y[n]=Mk=0bkx[nk]Nl=0aly[nl], where k and l are the "dummy" counting indices for the sum and n denotes the index of the nth sample of the output sequence.
  • Causal and noncausal:
    • Causal filter: a filter that uses only the present and past values of the input.
    • Noncausal filter: a filter that uses future values of the input.
  • Causal running averager or backward averager, similarly, we have the centralized running averager and the forward averager.

The general FIR filter

  • The general causal difference equation y[n]=Mk=0bkx[nk], where the coefficients bk are fixed numbers.
    • M, the order of the FIR filter
    • L=M+1, the number of filter coefficients is the filter length
  • Eq. (2) could be written as y[n]=nl=nMbnlx[l], where l=nk showing the FIR is causal using the input x[l] start from the previous M samples, i.e. l=nM, up to the current one l=n
  • For finite length input signal, i.e., x[l]0 for l[0,N1] and a Mth-order FIR filter (of length M+1, i.e., involving M+1 samples), there would be transient component of the output including M samples running onto and running off session. And the total output length would be N+M (order)=N+L1.

The unit impulse response and convolution

The impulse response provides a complete characterization of the FIR filter. - Three new ideas introduced: - the unit impulse sequence - the unit impulse response - the convolution sum

Unit impulse sequence

  • Unit impulse, or mathematically taken as the Kronecker delta function δ[n]={1n=00n0
  • Express any sequence interm of delta function x[n]=kx[k]δ[nk]\labelx
    • the unit impulse is a sequence
    • x is a summation of infinite impulse sequences δk

Unit impulse response sequence

  • The output from a filter is called the response to the input.
  • Unit impulse response h[n] represents the output when the input is the unit impulse δ[n].