Fixed-Point Signal Processing Systems

David V. Anderson      Pole Positions        Wayne T. Padgett    

When

Contact David Anderson or Wayne Padgett to reserve your place today!  On-site course offerings can be arranged.

Click here to register online for the Georgia Tech offering.

Who Should Attend

The course is designed for engineers, scientists, and technicians who need to do signal processing on fixed-point processors.  Some exposure to signal processing is assumed but general concepts will be briefly reviewed the first day. 

Course Objectives

Over 90% of signal processing systems use finite-precision (fixed-point) arithmetic.  This four-day course presents an in-depth look into designing fixed-point signal processing systems.  The course includes extensive laboratory time so that participants can explore concepts as they are taught.  The course begins with a brief review of pertinent DSP concepts and then continues to cover notation and fixed-point filtering fundamentals, noise analysis, scaling issues, multi-rate systems, and system performance analysis.

Laboratories

Each day’s instruction will include laboratory time designed to supplement the lectures.  These laboratories are either interactive system design exercises or system analysis projects.  The instructors will be available to work with participants.

Course Outline

Day 1

  1. DSP concepts review – Z transforms, frequency response, convolution
  2. DSP concepts review – FIR, IIR, DF1, DF2, cascade of sos filter structures
  3. Lab time – MATLAB tools, zplane, freqz, butter, pmfir, tf2sos
  4. Random processes – autocorrelation, power spectrum, filtering noise, sum(h.^2)*sigma identity

 Day 2

  1. Fixed point – Q notation, quantization noise, multiplication, rounding
  2. Fixed point – coefficient quantization, roundoff error, overflow/scaling
  3. Lab time – quantizing coefficients in direct form, in cascaded sos
  4. Fixed point – Analyzing roundoff error, noise power, spectrum at output

 Day 3

  1. Lab time – comparing roundoff noise model to simulated fixed point filter
  2. Fixed point – Analyzing overflow / scaling, computing scaling values
  3. Lab time – detecting overflow, computing scale factors
  4. Lab time – observing effects of scaling on noise power

 Day 4

  1. Multirate – polyphase FIR beats IIR
  2. Multirate – staged interpolation advantages
  3. Lab time – comparison of IIR to staged polyphase (order, noise power, overflow)
  4. Wrap up, summary

Course Administrators

Dr. David V. Anderson

(404) 385-4979

david.anderson@ece.gatech.edu

Dr. Wayne Padgett

(812) 877-8185

wayne.padgett@rose-hulman.edu