June 23, 2013
June 23, 2013
June 26, 2013
Computers in Education
23.172.1 - 23.172.10
An Inexpensive Approach for Teaching Adaptive Filters Using Real-Time DSP on a New Hardware PlatformDigital signal processing (DSP) has become one of the “must know” topics that many employersexpect of new electrical and computer engineering graduates. It has been found that a trueunderstanding of many fundamental DSP topics can be more fully realized by a student whenthey attempt to implement various DSP algorithms in real-time (typically in C), when comparedto non-real-time implementations with tools such as MATLAB or LabVIEW . In order tohelp students successfully transition from theory to real-time practice, there needs to be both apedagogical method and an infrastructure in place to support them and target as many modes oflearning as are reasonably possible.The authors have long advocated a three-step method of teaching DSP [1–3]. First, we teach thetheory along with interesting and motivating real-time demonstrations. We then have studentsimplement a particular concept using MATLAB, until they are comfortable with the basic topic.Finally, we have them “de-vectorize” their MATLAB code and convert it to C, in order tocompile and run it in real-time on high-performance DSP hardware.There are several choices of hardware the professor may make if using this approach. For thevery popular Texas Instruments (TI) processors, the Spectrum Digital C6713 DSK, the Logic PDZoom OMAP-L138 Experimenters Kit (ZEK), and the relatively new Texas Instruments OMAP-L138 Low Cost Development Kit (LCDK) are all highly capable platforms currently availablefor real-time DSP and all can be used effectively with the recommended pedagogical method.The LCDK is the newest of the three, having been introduced in the latter part of 2012. Itscapabilities are equal to or greater than the other two platforms, and it is the lowest in cost (only$195 as of this writing). One capability of particular interest is the access to not only “line in”audio but also to “microphone in” audio that can support generic stereo microphones or dualmonaural microphones.One application of this now-available stereo microphone capability, that is useful for teachingDSP to students, is adaptive noise cancelation using an adaptive digital filter [4,5]. The theory isstraightforward. As shown in the figure below, two signals are provided to the system. Theupper signal contains “signal plus noise” while the lower signal contains “correlated noise.” Theadaptive filter (the box labeled with the H k ( z ) transfer function) uses the error signal to adjust itstransfer function in real-time so as to optimally cancel out the noise at the output. signal plus noise + output correlated – noise Hk (z) error signalIn the scenario we present to our students, we are trying to duplicate the noisy environment of afirefighter at the scene of an emergency. The “signal plus noise” represents the combined signalsfrom the firefighter’s helmet-mounted microphone (where the “signal” is his/her voice and the“noise” is a chainsaw running in the background). The “correlated noise" signal represents justthe chainsaw’s signal, as detected by a second microphone located on the firefighter, butnowhere near his/her mouth. The adaptive filter’s purpose is to enhance the voice signal so thatthe firefighter may communicate effectively, for example, when using a radio.The full paper will include a brief review of the basic concepts of adaptive noise cancelation, theimportant DSP concepts which it reinforces in student minds, and the method used for thisexample to enhance student learning, along with student results. A helpful comparison of thethree currently available real-time DSP platforms using TI processors will also be provided.During our presentation at the Annual ASEE Conference in Atlanta, we will provide a real-timedemonstration of the effectiveness of this example system. author info removed for blind review, “Teaching DSP: Bridging the gap from theory to real-time hardware,” ASEE Comput. Educ. J., pp. 14–26, July- September 2003. author info removed for blind review, “Old Tricks for a New Dog: An Innovative Software Tool for Teaching Real-Time DSP on a New Hardware Platform,” ASEE Comput. Educ. J., pp. 64–69, October-December 2011. author info removed for blind review, Real-time digital signal processing: from MATLAB to C with TMS320C6x DSPs, 2nd ed., CRC Press, 2012. S. D. Stern and R. A. David, Signal Processing Algorithms in MATLAB, Prentice Hall PTR, Upper Saddle River, NJ, 1996. S. Haykin, Adaptive Filter Theory, 3rd ed., Prentice Hall, Upper Saddle River, NJ, 1996.
Morrow, M. G., & Wright, C. H. G., & Welch, T. B. (2013, June), An Inexpensive Approach for Teaching Adaptive Filters Using Real-Time DSP on a New Hardware Platform Paper presented at 2013 ASEE Annual Conference & Exposition, Atlanta, Georgia. https://peer.asee.org/19186
ASEE holds the copyright on this document. It may be read by the public free of charge. Authors may archive their work on personal websites or in institutional repositories with the following citation: © 2013 American Society for Engineering Education. Other scholars may excerpt or quote from these materials with the same citation. When excerpting or quoting from Conference Proceedings, authors should, in addition to noting the ASEE copyright, list all the original authors and their institutions and name the host city of the conference. - Last updated April 1, 2015