-
COST 286 - 25/03/2004 - TELICE 1
Transmission on indoor power lines: from a stochastic channel
model to the optimization and performances evaluation of
multicarrier system
By V. Dégardin, M. Liénard, P. DegauqueUniversity of Lille, IEMN
/ DHS
TELICE GroupTélécommunications, Interférences et
Compatibilité Electromagnétique
-
COST 286 - 25/03/2004 - TELICE 2
Objective
Analysis of the Bit Error Ratein the presence of narrow band
noise &in time varying selective channelin the presence of
impulsive noise
Influence of channel estimation methods⇒ Optimization of
semi-blind techniquesOptimization of noise processing and channel
coding
-
COST 286 - 25/03/2004 - TELICE 3
Outline
I. Indoor Powerline channel : Measurements & modelII. System
optimization in the presence of narrow band noise & in time
varying selective channel
Transmission processSemi-blind channel estimationSimulation
results
III. System optimization in the presence of impulsive noiseNoise
processing and channel codingSimulation results
Conclusion : software and demonstrator
-
COST 286 - 25/03/2004 - TELICE 4
I. Indoor powerline channel model
Two classes of NoiseThe stationary Noise : colored and narrow
band noiseThe impulsive noise
The transfer function with 2 main characteristicsMultipath
environment : frequency selective channelTime varying channel
-
COST 286 - 25/03/2004 - TELICE 5
The stationary noise [0-30] MHz
⇒ Modelled by filtered gaussian noise and sinusoids
characterized by their frequency, amplitude and phase (randomly
distributed)
Transmission Band
I. Indoor powerline channel model
-
COST 286 - 25/03/2004 - TELICE 6
I. Indoor powerline channel modelIdea : mobile platform for
measuring impulsive noiseDifficulties :* Programming (Measurement
of interarrival time and data acquisition in real time)* System
protection against over voltages (attenuators et couplers ) ⇒
frequency response must be flat in the band
Noise acquisition
acquisitionTIA
coupler
resetBattery 12V
PC
CH AExt trigger
Port //Trig out
-
COST 286 - 25/03/2004 - TELICE 7
⇒ Damped sinusoid
⇒ Succession of damped sinusoids :higher pseudo-frequency higher
amplitude
⇒ Bursts may degrade PLT systems
I. Indoor powerline channel model
Impulsive noise :Measurements have been performed in a house
during 40 hours. Two types of impulses are given : Single transient
& Burst impulse
-
COST 286 - 25/03/2004 - TELICE 8
Characteristic parameters of single transient :
- peak amplitude - pseudo-frequency- damped coefficient -
duration- interarrival time
(b) Model of burst impulse
(a) Model of single transient
I. Indoor powerline channel model
-
COST 286 - 25/03/2004 - TELICE 9
Class 5 (18 %)Class 4 (11 %)Class 3 (3 % )Burst pulses Class 2
(20 %)Class 1 (48 %)Single transient
fo>3 MHz0.5 MHz < fo < 3MHzfo
-
COST 286 - 25/03/2004 - TELICE 10
I. Indoor powerline channel modelModel validation : Comparison
of mean spectral density of measured and generated pulses
Good agreement Measurement/Model ⇒ Validation of stochastic
model of impulsive noise
-
COST 286 - 25/03/2004 - TELICE 11
I. Indoor powerline channel model
The measured transfer function :intensive measurements were
carried out on a specific installation comprising a line, 30 m
long, with wall plugs
⇒ Fading in each frequency band⇒ Sudden channel variations :
succession of stationary states⇒ Modeled by a tap delay line
filter, the amplitude being determined by a Rayleigh distribution
and phase angle being uniformly distributed.
-
COST 286 - 25/03/2004 - TELICE 12
II. System optimization in the presence of narrow band noise
& in time varying selective channel
The OFDM process & semi-blind (SB) channel estimation⇒
Simulation resultsThe DMT process & extension of SB channel
estimation⇒ Simulation comparison
-
COST 286 - 25/03/2004 - TELICE 13
II. The OFDM process
Bits stream
S/P
IFFT
PrefixAdd.
P/S
S/P
Removeprefix
FFT
P/S
EqualizerQAM
demapper
ReceivedBits
streamQAM
Mapper
Parameters :256 subcarriers modulated in 4-QAM, prefix = 32
samplesBit rate = 10 Mbit/s, Bandwidth = 5.5 MHzSynchronization
assumed to be ideal
Channel responseand noise
ZF criteria :inverse of the channel estimation
kk HC ˆ
1=
-
COST 286 - 25/03/2004 - TELICE 14
New algorithm based on a semi-blind (SB) estimationPerformances
comparison of different estimation methods
Canal 2Canal 1
Eb/No = 14 dB
⇒ efficient results for SB estimation & detection
⇒ extension to the DMT ?
II. Simulation results of OFDM processes :Influence of the
channel estimation
-
COST 286 - 25/03/2004 - TELICE 15
II. DMT process & extension of SB estimation
The DMT principle :OFDM link + Algorithm which optimizes the
number of bits/symbol (size of constellation) versus the SNR in
each subband
Extension of SB estimation not straightforward !⇒ SB estimation
not accurate when M
-
COST 286 - 25/03/2004 - TELICE 16
III. System optimization in the presence of impulsive noise
Noise processingChannel codingSimulation results
-
COST 286 - 25/03/2004 - TELICE 17
Preprocessing :Remove noise>
threshold(As=3.4 V)
{r}Matsuo iterative process
(M, number i of iterations){X}
Iteration n° 1 Iteration n° i > 1
time
M=3 largest amplitudes Optimization : threshold As
III. Noise processingMatsuo iterative process: consist in
determining at reception the largest amplitudes of bruit and
subtract them
Optimization: amplitude (noise) >> amplitude (signal )⇒
Possibility of determining a threshold As
-
COST 286 - 25/03/2004 - TELICE 18
Informative word of K bytes
Reed-Solomoncode
Code word of 255 bytes
bytes
III. Channel codingThe Reed-Solomon code:This code adds (255-K)
parity bytes and correct up to t = (N-K)/2 bytes. (N=255, K=239, t
= 8 bytes)
The interleaving:
1 2 3 255
256257258 510
511 765
766
Bytes are read on 255 columns
Bytes are written on D rows
2 ≤ D ≤ 64
-
COST 286 - 25/03/2004 - TELICE 19
⇒ P (BER
-
COST 286 - 25/03/2004 - TELICE 20
Conclusion of the study
Simulation toolDemonstrator
Perspectives
-
COST 286 - 25/03/2004 - TELICE 21
Software based on MATLAB (The Mathworks)
Conclusion : Simulation Tool
-
COST 286 - 25/03/2004 - TELICE 22
Characteristics :256 subcarriersP=-50 dBm/HzBit rate: 2,5
Mb/s
Adaptative aspects of the system :Transmission bandwidth [10-30]
MHzInjection PowerNumber of subcarriersvariable rate (following
phase)
The difficulties of the OFDM process:Synchronization of the
received symbolsSynchronization of the OFDM framesOL
synchronization Peak power of the received signal
Conclusion : Demonstrator realization
-
COST 286 - 25/03/2004 - TELICE 23
emission cardCPLD (e/s)DSP (IFFT)
ModulationDemodulation
TimeSynchro
Horlogecard
Conclusion : Demonstrator realization
Impulsive noise generator
-
COST 286 - 25/03/2004 - TELICE 24
Perspectives
Transmission on powerline in subway stationsProblem : low SNR,
long lines (few hundred meters) ⇒ channel measurement and
modeling
PLC in cars (high bit rate)
-
COST 286 - 25/03/2004 - TELICE 25
Transmission on indoor power lines: from a stochastic channel
model to the optimization and performances evaluation of
multicarrier system
By V. Dégardin, M. Liénard, P. DegauqueUniversity of Lille, IEMN
/ DHS
TELICE GroupTélécommunications, Interférences et
Compatibilité Electromagnétique