Topic outline

  • General

  • Topic 1: Finite Impulse Response Filters

    Finite impulse response (FIR) filters are widely used to suppress noise, to control phase and add special effects to audio. As their name suggests, they comprise a limited number of coefficients which are then convolved with the input signal to generate the output. Finite impulse response filters are inherently stable since they do not employ feedback. They are also known as feed-forward filters.

    In this topic we will explore their characteristics, how they are designed and how they are implemented.

    Work package: Section 4.1

    • Topic 2: Infinite Impulse Response Filters

      Like FIR filters, infinite impulse response filters are also widely used to suppress noise, to control phase and add special effects to audio. Since they employ feedback they require fewer coefficients to produce a given response, and so are more efficient than their FIR counterparts. However, if poorly designed they risk instability, so care is needed, especially in relation to word length and filter order.

      In this topic we will explore their characteristics, how they are designed and how they are implemented.

      Work package: Section 4.2 and 4.4

    • Topic 3: Adaptive filters

      Unlike FIR and IIR filters, adaptive filters alter their characteristics in response to the statistics of the incoming signal (and noise). Typically, they dynamically modify the values of the coefficients according to an error signal which arises by comparing the actual filtered signal to some ideal reference.

      Adaptive filters are very powerful, since (unlike linear filters) they allow the signal to be recovered even if the noise occupies the same bandwidth of the signal.

      In this topic we will how to design one very popular form of adaptive filter, known as the Least Mean Square (LMS) FIR adaptive. We will also examine how it is used in practice to suppress broadband interference.

      Work package: Section 4.3

      • Course Materials