Powerline Communications for Smart Grids Prof. Brian L. Evans Department of Electrical & Computer Engineering Wireless Networking & Communications Group.

Powerline Communications forSmart Grids

Prof. Brian L. EvansDepartment of Electrical & Computer Engineering

Wireless Networking & Communications GroupThe University of Texas at Austin

17 July 2012

Seminar at the American University of BeirutCo-sponsored by the local IEEE chapter

In collaboration with PhD students Ms. Jing Lin, Mr. Marcel Nassar andMr. Yousof Mortazavi and TI R&D engineers Dr. Anand Dabak and Dr. Il Han Kim

2

Outline

• Research group overview• Smart power grids• Powerline noise

CyclostationaryGaussian mixture

• Testbed• Conclusion

2004200520062007

20082009

20102011

2012

My visits to Lebanon

(x2)

Research Group

• Present: 9 PhD, 1 MS, 5 BS• Alumni: 20 PhD, 9 MS, 140 BS• Communication systems

Powerline communication systems (design tradeoffs)Cellular, Wimax & Wi-Fi (interference modeling & mitigation)Mixed-signal IC design (mostly digital ADCs and synthesizers)Underwater acoustic communications (large receiver arrays)

• Video processing (rolling shutter artifact reduction)

• Electronic design automation (EDA) tools/methods

• Part of Wireless Networking & Communications Group160 graduate students,18 faculty members, 12 affiliate companies

3

wncg.org

Research Group – Completed Projects

System Contribution SW release

Prototype Companies

ADSL equalization Matlab DSP/C Freescale, TI

MIMO testbed LabVIEW LabVIEW/PXI Oil&Gas

Wimax/LTE resource allocation LabVIEW DSP/C Freescale, TI

Camera image acquisition Matlab DSP/C Intel, Ricoh

Display image halftoning Matlab C HP, Xerox

video halftoning Matlab Qualcomm

EDA tools fixed point conv. Matlab FPGA Intel, NI

distributed comp. Linux/C++ Navy sonar Navy, NI

DSP Digital Signal Processor LTE Long-Term Evolution (cellular) MIMO Multi-Input Multi-Output PXI PCI Extensions for Instrumentation

20 PhD and 9 MS alumni

4

Research Group – Current Projects

System Contributions SW release Prototype Companies

Powerline comm.

noise reduction; MIMO testbed

LabVIEW LabVIEW / PXI chassis

Freescale, IBM, TI

Wimax, LTE & WiFi

interference reduction

Matlab FPGA Intel, NI

time-based ADC IBM 45nm

Underwater comm.

space-time methods; MIMO testbed

Matlab Lake Travis testbed

Navy

Cell phone camera

reducing rolling shutter artifacts

Matlab Android TI

EDA Tools reliability patterns NI

6 PhD and 4 MS students

MIMO Multi-Input Multi-Output PXI PCI Extensions for Instrumentation

5

6

Smart Grid Goals

• Accommodate all generation typesRenewable energy sourcesEnergy storage options

• Enable new products, service and markets• Improve asset utilization and operating efficiencies

Scale voltage with energy demandGeneration cost 30x higher during peak times vs. normal load (USA)Plug-in vehicles create unpredictability in residential power load

• Improve system reliability including power quality• Enable informed customer participation

Source: Jerry Melcher, IEEE Smart Grid Short Course, 22 Oct. 2011, Austin TX USA

7

A Smart Grid

Electric car charging & smart billing

Communication to isolated area

Power generation optimization

Disturbance monitoring

Smart metering

Integrating alternative energy

sources

Source: ETSI

Load balancing

8

• Built for unidirectional energy flow• Bidirectional information flow

throughout smart grid will occur

Power Lines

Medium Voltage (MV)1 kV – 33 kV

Low Voltage (LV)under 1 kV

High Voltage (HV)33 kV – 765 kV

Source: ERDFTransformer

9

Today’s Situation in USA

• 7 large-scale power grids each managed by a regional utility companyWestern US, Eastern US, Texas, and others700 GW generation capacity in total for long-haul high-voltage power transmissionSynchronized independently, and exchange power via DC transfer

• 130+ medium-scale power grids each managed by a local utilityLocal power distribution to residential, commercial and industrial customers

• Heavy penalties in US for blackouts (2003 legislation)Utilities generate expected energy demand plus 12%

• Traditional ways to increase capacity to meet peak demand increaseBuild new large-scale power generation plant at cost of $1-10B if permit issuedBuild new transmission line at $0.6M/km which will take 5-10 years to complete

Source: Jerry Melcher, IEEE Smart Grid Short Course, 22 Oct. 2011, Austin TX USA

10

Smart Power Meters at Customer Site

• Enable local utilities to improveOperating efficiencySystem reliabilityCustomer participation

• Automatic metering infrastructure functionsInterval reads (every 1/15/30/60 minutes) and on-demand reads and pingsTransmit customer load profiles and system load snapshotsPower quality monitoringRemote disconnect/reconnect and outage/restoration event notification

• Need low-delay highly-reliable communication link to local utility• 75M smart meters sold in 2011 (20% increase vs. 2010)

Source: Jerry Melcher, IEEE Smart Grid Short Course, 22 Oct. 2011, Austin TX USA

11

Local Utility Powerline Communications (PLC)

• PLC modems (PRIME, etc.) use carrier sensed multiple access to determine when the medium is available for transmission• MV router plays similar role as a Wi-Fi access point

12

PLC In Different Frequency Bands

Category Band Bit Rate Applications Standards

Ultra Narrowband

0.3 – 3 kHz ~100 bps

• Automatic meter reading• Outage detection• Load control

N/A

Narrowband 3 – 500 kHz ~500 kbps

• Smart metering• Real-time energy

management

• PRIME, G3• ITU-T G.hnem• IEEE P1901.2

Broadband 1.8 – 250 MHz ~200 Mbps • Home area networks

• HomePlug• ITU-T G.hn• IEEE P1901

All of the above standards are based on multicarrier communications using orthogonal frequency division multiplexing (OFDM).

13

Comparison Between Wireless and PLC SystemsWireless Communications Narrowband PLC (3-500 kHz)

Time selectivity Due to node mobility From random load variationsdue to switching activity

Time-varying stochastic model

Doppler spectrum Periodic with period of half AC main freq. plus lognormal time-selective fading

Power loss vs. distance d

d –n/2 where n is propagation constant

e – a(f) d plus additional attenuation when passing through transformers

Additive noise Assumed stationaryand Gaussian

non-Gaussian and impulsive with dominant cyclostationary component

Propagation Dynamically changing Determinism from fixed grid topology

Interference limited

In Wi-Fi deployments and increasing in cellular

Increasing due to uncoordinated users using different standards

MIMO Standardized forWi-Fi and cellular

Order of #wires minus 1;G.9964 standard for broadband PLC

Synchronization Difficult across network AC main frequency makes simpler

14

Phys

ical

Lay

er P

aram

eter

s fo

r O

FDM

Nar

row

band

PLC

Sta

ndar

ds

CENELEC A band is from 3 to 95 kHz. FCC band is from 34.375 to 487.5 kHz.PRIME and G3 use real-valued baseband OFDM. Others are complex-valued.

15

Sources of Powerline Noise

Electronic devices

Uncoordinatedtransmission

Power linedisturbance

Taken froma local utility point of view

16

Types of Powerline Noise

Background Noise Cyclostationary Noise Impulsive Noise

Spectrally shaped noise with 1/f spectral decay

Period synchronous to half of the AC cycle Random impulsive bursts

Superposition of low intensity noise sources

Switching power supplies and rectifiers

Circuit transient noise and uncoordinated interference

Present in all PLC Dominant inNarrowband PLC

Dominant inBroadband PLC

0 100 200 300 400 500-150

-100

-50

Frequency (kHz)time

17

• Performance of conventional communication system degrades in non-AWGN environment

• Statistical modeling of powerline noise

• Noise mitigation exploiting the noise model or structure

Non-Gaussian Noise: Challenge to PLC

Listen to the environment

Estimate noise model

Use model or structure to mitigate noise

18

Cyclostationary Noise Modeling in Narrowband PLC (3-500 kHz)

1. M. Nassar, A. Dabak, I. H. Kim, T. Pande and B. L. Evans, “Cyclostationary Noise Modeling In Narrowband Powerline Communication For Smart Grid Applications”, Proc. IEEE Int. Conf. on Acoustics, Speech, and Signal Proc., Mar. 25-30, 2012, Kyoto, Japan.

2. M. Nassar, J. Lin, Y. Mortazavi, A. Dabak, I. H. Kim and B. L. Evans, “Local Utility Powerline Communications in the 3-500 kHz Band: Channel Impairments, Noise, and Standards”, IEEE Signal Processing Magazine, Special Issue on Signal Processing Techniques for the Smart Grid, Sep. 2012, 14 pages.

19

Cyclostationary Noise: Field Measurement

Medium Voltage Site Low Voltage Site

Data collected jointly with Aclara and Texas Instruments near St. Louis, MO USA

20

• Linear periodically time-varying (LPTV) system model

o A period is partitioned into M segmentso Noise within each segment is stationary, i.e. modeled by an LTI system

Noise Modeling

1H

2H

MH

…

Nv R

Hi - Linear time invariant filterN - Period in samples

Nn R

Segment: 1 2 3

21

• LPTV model (M = 3) captures temporal-spectral cyclostationarity

Model Fitting

Measurement data Noise synthesized from model

The proposed TI-Aclara-UT model was adopted in the IEEE P1901.2 narrowband PLC standard

22

A Lebanese Interlude

JezzineSidon

Beiteddine

Jbeil/Byblos

Baruk Cedars

Ghine

Notshown:Baalbek,Beirut,Tripoli,Tyre,Zahle,and other great places

23

Impulsive Noise in Broadband PLC: Modeling and Mitigation

3. M. Nassar, K. Gulati, Y. Mortazavi, and B. L. Evans, “Statistical Modeling of Asynchronous Impulsive Noise in Powerline Communication Networks”, Proc. IEEE Int. Global Communications Conf., Dec. 5-9, 2011, Houston, TX USA.

4. J. Lin, M. Nassar and B. L. Evans, “Non-Parametric Impulsive Noise Mitigation in OFDM Systems Using Sparse Bayesian Learning”, Proc. IEEE Int. Global Communications Conf., Dec. 5-9, 2011, Houston, TX USA.

24

Sources of Impulsive Noise

Wireless Emissions

In-home PLC Switching Transients

Uncoordinated Meters

(coexistence)

Total interference at receiver: Interference from source i

25

• Interference from a single source

Emission duration: geometrically distributed with mean μPulse arrivals: homogeneous Poisson point process with rate λAssuming channel between interference source and receiver has flat fading

Statistical-Physical Modeling

t=0

Noise

env

elop

e

(k) (j) (2) (1)

k pulses in a window of duration T

Tk

Pulse emission duration

τj

Pulse arrival time

26

• Aggregate interference from multiple sources

Statistical-Physical Modeling (cont.)

Ex. Rural areas, industrial areas with heavy machinery

Dominant interference source Middleton class A

2 2( ) [ ]

2

A

A d E h B

Impulse rate lImpulse duration m

Homogeneous networkEx. Semi-urban areas, apartment complexes

Middleton class A

2 2[ ]

2

A M

A E h B

M

li = , l mi = m, (g di) = g

General (heterogeneous) network

Ex. Dense urban and commercial settings

Gaussian mixture model

π and σ2 in [3]

li, mi, (g di) = gi

27

Model Fitting: Tail Probability

Homogeneous PLC Network General PLC Network

Middleton Class A model is a special case of the Gaussian mixture model (GMM)

28

• FFT spreads out impulsive energy across all tones

SNR in each tone is decreasedReceiver performance degrades

OFDM Systems in Impulsive Noise

29

• A linear system with Gaussian disturbance

Estimate the impulsive noise and remove it from the received signal

Apply standard OFDM decoder as if only AWGN were present

Impulsive Noise Mitigation in OFDM Systems

* 2, ~ ( , )y Fe FHF x Fn Fe v v CN x I

g

v

ˆ ˆy y Fe x g

30

• Noise in different PLC networks has different statistical models• Mitigation algorithms need to be robust in different noise scenarios

Parametric Vs. Non-Parametric Methods

Parametric Methods Non-Parametric Methods

Assume parameterized noise statistics Yes No

Performance degradation due to model mismatch Yes No

Training needed Yes No

31

• A compressed sensing problemExploiting the sparse structure of the time-domain impulsive noise

• Sparse Bayesian learning (SBL)Proposed initially by M. L. Tipping

A Bayesian inference framework with sparsity promoting prior

Non-Parametric Mitigation Using Null Tones

J : Index set of null tonesFJ : DFT sub-matrixe: Impulsive noise in time domaing: AWGN with unknown variance

32

• Bayesian inferenceSparsity promoting prior: Likelihood:Posterior probability:

• Iterative algorithmStep 1: Maximum likelihood estimation of hyper-parameters (γ, σ2) Solved by expectation maximization (EM) algorithm (e is latent variable)Step 2: Estimate e from the mean of the posterior probability, go to Step 1

Sparse Bayesian Learning

| ~ (0, ), ( )e CN diag 2 * 2| , ~ (0, )Jy CN F F I

2| ; , ~ ( , )J ee y CN

33

• Joint estimation of data and noise

Treat the received signal in data tones as additional hyper-parameters

Estimate of is sent to standard OFDM equalizer and symbol detector

Non-Parametric Mitigation Using All Tones

: Index set of data tonesz : Received signal in frequency domainJ

Jz

-10 -5 0 5 10

10-5

10-4

10-3

10-2

10-1

SNR (dB)

Sym

bol E

rror

Rat

e

No cancellationSBL w/ null tonesSBL w/ all tones

• Interference in time domain

• Learn statistical modelUse sparse Bayesian learning

Exploit sparsity in time domain

• SNR gain of 6-10 dBIncreases 2-3 bits per tone for

same error rate - OR -

Decreases bit error rate by 10-100x for same SNR

Simulated Communication Performance

34

~10dB

~6dBtime

Transmission places 0-3 bits at each tone (frequency). At receiver, null tone carries 0 bits

and only contains impulsive noise.

Our PLC Testbed

• Quantify application performance vs. complexity tradeoffsExtend our real-time DSL testbed (deployed in field)Integrate ideas from multiple narrowband PLC standardsProvide suite of user-configurable algorithms and system settingsDisplay statistics of communication performance

• 1x1 PLC testbed (completed)Adaptive signal processing algorithmsImproved communication performance 2-3x on indoor power lines

• 2x2 PLC testbed (on-going)Use one phase, neutral and groundGoal: Improve communication performance by another 2x

35

Our PLC Testbed

36

Hardware Software

• National Instruments (NI) controllers stream data

• NI cards generates/receives analog signals• Texas Instruments (TI) analog front end

couples to power line

• NI LabVIEW Real-Time system runs transceiver algorithms

• Desktop PC running LabVIEW is used as an input and visualization tool to display important system parameters.

1x1 Testbed

37

Conclusion

• Communication performance of PLC systems

Primarily limited by non-Gaussian noise

• Proposed statistical models for

Cyclostationary noise in narrowband PLC systems

Impulsive noise in broadband PLC systems (also useful in narrowband PLC)

• Proposed non-parametric impulsive noise mitigation algorithms

OFDM PLC systems (G3, IEEE P1901.2, ITU G.hnem, etc.)

Robust in noise scenarios tested

6-10 dB SNR gain over conventional OFDM receivers

Thank you …

38

39

• Symbol error rate in different noise scenarios

Simulated Performance

Middleton class A modelGaussian mixture model

~6dB

~8dB

~4dB

~6dB

~10dB

• MMSE w/ (w/o) CSI: Parametric estimator assuming known (unknown) statistical parameters of noise• CS+LS: A compressed sensing and least squares based algorithm