*IEEE Global Communications Conference*,
Dec. 9-12, Atlanta, GA USA.
# Non-parametric Mitigation of Periodic Impulsive
Noise in Narrowband Powerline Communications

*Jing Lin* and
*Brian L. Evans*
Department of Electrical
and Computer Engineering,
Wireless Networking and Communications Group,
The University of Texas at Austin,
Austin, TX 78712 USA

jing.lin08@gmail.com -
bevans@ece.utexas.edu

Paper -
Slides

Smart Grid Communications Research at UT Austin

## Abstract

Periodic impulsive noise synchronous to the main powerline frequency
is the dominant noise component in OFDM-based narrowband (NB)
powerline communications (PLC).
Such noise occurs in periodic bursts, where a single burst could
corrupt multiple OFDM symbols.
Standardized NB PLC systems use frequency-domain interleaving (FDI)
in combination with forward error correction to combat impulsive noise.
Alternate designs adopt time-domain block interleaving (TDI) in which
the receiver deinterleaver scatters an impulsive noise burst into
short impulses over a large number OFDM symbols.
In bursty impulsive noise, TDI-OFDM (FDI-OFDM) works better at high
(low) SNR.
In this paper, we develop non-parametric methods for periodic
impulsive noise mitigation in coded TDI-OFDM systems.
We exploit the sparse structure of the time-domain noise after the
deinterleaver, and propose sparse Bayesian learning based algorithms
that estimate and remove the noise impulses by observing the null
and pilot tones of received signal and using decision feedback from
the decoder.
The proposed methods do not assume any statistical noise model and
hence do not require any training.
In simulations, the proposed methods in TDI-OFDM systems achieve up
to 6 dB SNR gain over FDI-OFDM systems at typical NB PLC SNR values.
## Discussion

**Question 1:** Does it improve the BER performance if a larger
interleaver is adopted?

**Answer 1:** Yes. In our simulations, using a larger interleaver
improves the performance, but not significantly.
A larger interleavler does make the noise samples more independent.
However the performance of the sparse Bayesian learning algorithm
is mainly affected by the sparsity of the noise.
As long as the interleaver size is multiples of the noise period,
the sparsity of the deinterleaved noise is the same, regardless of
the interleaver sizes.
**Question 2:** How did you generate the periodic impulsive
noise in your simulation?

**Answer 2:** We used the linear periodically time-varying noise
model in the IEEE P1901.2 narrowband PLC standard.
We shape the noise spectrum according to the field measurements as
shown in slide 2.

**Question 3:** In your formulation of the compressed sensing
problem, is the sensing matrix random or fixed?
It's not exactly compressed sensing if the sensing matrix is fixed.

**Answer 3**: The sensing matrix is fixed. It's the sub-DFT matrix.
I think compressed sensing doesn't require the sensing matrix to be random.

One participant mentioned that a smaller interleaver could also break a
long burst into several shorter bursts.
Then one could exploit the block sparsity of the noise to recover it.

COPYRIGHT NOTICE: All the documents on this server
have been submitted by their authors to scholarly journals or conferences
as indicated, for the purpose of non-commercial dissemination of
scientific work.
The manuscripts are put on-line to facilitate this purpose.
These manuscripts are copyrighted by the authors or the journals in which
they were published.
You may copy a manuscript for scholarly, non-commercial purposes, such
as research or instruction, provided that you agree to respect these
copyrights.

Last Updated 01/03/14.