I am using an FIR filter with LMS to achieve adaptive noise cancellation, and hence I need to find the step size of the algorithm. How can I approach this to find the step size?
Thanks
The step size should satisfy the condition:
0 < step-size < 1/(max eigen value of the input autocorrection matrix)
For fast convergence step-size is set close to its maximum allowed value.
If you choose very big step-size, convergence is fast, but final error is also large.
I would suggest, first use maximum allowed value and once abs(change in error) is stabilized, keep reducing step-size to reach the best result.
You can put 'if' loop indicating this condition in your program.
You may find the discussion in the following chapter useful:
http://www.ece.utah.edu/~mathews/ece6550/chapter5.pdf
Initially start at maximum step-size, and keep reducing step size to get the exact result.
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Process