Skip to content

PULP implementation of adaptive filter algorithms: NLMS, RLS, BlockNLMS, BlockRLS.

Notifications You must be signed in to change notification settings

gaoyiyeah/AdaptiveFilters

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

59 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Adaptive Filters

This is a porting of the following adaptive filter algorithms:

  • Normalized Least Mean Squares
  • Recursive Least Squares
  • Block form of LMS
  • Block form of RLS

Implementation taken from

Pyroomacoustics

and ported to C, more precisely for Parallel Ultra-Low Power platform

PULP

AdaptiveFiltersPy

In this folder, you can find the Python implementation of the adaptive filters algorithms. The golden model results are written to results.txt, for checksum with the C code porting.

AdaptiveFiltersC

In this folder instead, you can find the C of the adaptive filter algorithms: for each single algorithm a .c file is present, while in PerformanceMeasure.md for each single algorithm a fine grained perofrmane measure is recorded, including the number of cycles and instrctions (this allows to appreciate and understand gains of optimization and parallelization).

Run instruction

Running the code could be a bit difficult: first is necessary to get the PULP virtual machine and then run it (I used VMWare, free version). Secondly, I prefered to produce the python golden model on my Windows python virtual environemnt, run adaptive_filters.py to write on results.txt and transfer them to the PULP VM thanks to git. Then the C code can run int he PULP VM with the make clean all run command.

The data_visualization.py scripts is inteded to both study the behaviour of the algorithms while changing their parameters and analyzing the performances of the C code (output in graphs directories).

Third Party Code

As said, Adaptive Filters code is taken from PyRoomAcoustics repository, while the invert matrix function in the C code is taken from:

https://github.com/pulp-platform/pulp-dsp

About

PULP implementation of adaptive filter algorithms: NLMS, RLS, BlockNLMS, BlockRLS.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C 75.1%
  • Python 24.8%
  • Makefile 0.1%