Pilot-Aided and Blind Equalization in FBMC Modulation for PMR … · 2015-02-13 · Pilot-Aided and Blind Equalization in FBMC Modulation for PMR Networks Séminaire SCEE CentraleSupélec,

Pilot-Aided and Blind Equalization

in FBMC Modulation for PMR Networks

Séminaire SCEE CentraleSupélec,

Campus de Rennes

12 Février 2015

Vincent Savaux

Outline

• PMR Networks

• FBMC Modulation Scheme

• OFDM/OQAM• OFDM/OQAM

• Pilot-Aided Estimation/Equalization

• Blind Equalization

Foreword …

What is the PROFIL project ?

PROFIL: Evolution de la PROfessional Mobile Radio large bande basée sur lamodulation FILter Bank MultiCarrier

Evolution of the Wideband Pro fessional Mobile Radio Based on the Fil ter Bank MultiCarrier Modulation

• ANR Project with Cassidian, CEA-Leti, TeamCast and CentraleSupélec

• CentraleSupélec’s part: Blind equalization in FBMC for PMR Networks

PMR NetworksPMR : Professional Mobile Radio

• Emergency Networks for police, firefighters, ambulances …

• Transmission of 2G-like data

TV GSM WiFi

Currently:

frequency380 - 395 MHz

radio

Standard TETRA -TETRAPOL

TV GSM WiFi

PMR Networks

• Specificities

UL

DL

P to P

� mobility� Narrowband channels

PMR Networks

Goal: transmission of broadband signal in the 380 – 395 MHz band

OFDM

Low ratePMR

Broadband PMR

frequency

Applications

FBMC Modulation Scheme

x0,n

SFB

x1,n

xm,n

Transmissionchannel

H

w

… +

+

IFF PPN

……

……

P/S

AFB

FFT

PPN

……

……

S/P

…

y0,n

y1,n

ym,n

- OFDM/OQAM Modem

xm,n

xM-1,n

H

…

+T … … T… …

…

ym,n

yM-1,n

FBMC Modulation Scheme

Bellanger’s filter frequency response [1]:

The subcarriers are orthogonal only in the real field

FBMC Modulation Scheme

m

m-1

m+1

n-1 n n+1 time

m+1

Surrounding interference part

xm,n

frequency

The channel induces complex interferences from neighboring subcarriers and symbols

FBMC Modulation Scheme

Comparison with OFDM:

Advantages Drawbacks

• Modem more complex

• Intrinsic interferences

• No cyclic prefix higher spectral efficiency

• Low out-of-band interferencesbetter coexistence

• Specific processes at receiverbetter coexistence properties

Pilot-Aided Equalization

Reminder : Channel equalization in OFDM/OQAM

Transmitted signal x Received signal y

Equalizer F

Equalized signal a

Pilot-Aided EqualizationFirst solution: pilots are multiplexed in the data stream (1/2)

t

pilot, help pilot

data

Loss of spectralefficiency

Transmitted data knownat the receiver

f

efficiency

BUT

Simple Equalizationprocess

Pilot-Aided Equalization

Pilot positions

First solution: pilots are multiplexed in the data stream (2/2)

2 steps : 1) The channel H is estimated on pilot tones2) The signal is equalized with a one-tap per-carrier channel inversion

F = 1/Ĥ

Unknownchannel

Pilot-Aided Equalization

Different pilot allocation methods in OFDM/OQAM [2], [3], [4] (1/4)

p 0m

n

Simple implementation

POP

Pair of pilots (POP) [3]

Simple implementation

Good spectral efficiency (same as OFDM)

Weak performance [c1]

noisevariance

interferencevariance

channelvariance

Pilot-Aided Equalization

p a

Different pilot allocation methods in OFDM/OQAM [2], [3], [4] (2/4)

m

nAP

Auxiliary pilot (AP) [4]

Good spectral efficiency

Cancellation of the interference(elegantly)

High complexity at transmitter

Pilot-Aided Equalization

p ba

p-

Different pilot allocation methods in OFDM/OQAM [2], [3], [4] (3/4)

m

nIAMIAM-R [2]: p- = ±1, p = ±1, p+ = -p- and a = 0,b = 0.

p+

Interference approximation

method (IAM) [2]

Pilot-Aided Equalization

p ba

p-

Different pilot allocation methods in OFDM/OQAM [2], [3], [4] (4/4)

m

nIAM

Reduction of the noise and interference energyp+

Interference approximation

method (IAM) [2]

Reduction of the noise and interference energy

Simple implementation

Loss of spectral efficiency

Good performance [c1]

Pilot-Aided Equalization

LMMSE in OFDM/OQAM [c2], [c3]

? ? ?

?

OFDM-like

Pilot-Aided Equalization

Some results: Typical urban (TU) channel, 3 km.h-1, no channel coding

Pilot-Aided EqualizationApplication to PMR band: some changes in the LTE PHY parameters [c4] (1/3)

From LTE parameters …

… to proposed LTE-like parameters

More flexibility

Pilot-Aided Equalization

time

freq

uenc

y

Nt

Nf

…

…

1 RBin LTE

…

…

1 RB in proposed

scheme

Application to PMR band: some changes in the LTE PHY parameters [c4] (2/3)

… …

2τ0

…

…

… … …

…

…

TOFDM

15 KHz 5 KHz

addedsubcarrier

a. LTE pilot distribution b. Proposed rectangular distribution

Pilot-Aided EqualizationApplication to PMR band: some changes in the LTE PHY parameters [c4] (3/3)

1 RB

Channel estimationalong the whole band

pilot

time

freq

uenc

y

frequency

PMR signals

BB BB

PMR signal

added carriers

frequency

More flexibility

Good channel estimationover each sub-band

Pilot-Aided Equalization

Some results (1/2): achieved bit rate

- 4-QAM symbols, 3 MHz bandwidth- reference in OFDM in LTE: 4.8 Mbits.s-1

- Proposed scheme:

Similar to LTE

Enable high bit rate applications, as video streaming

Pilot-Aided Equalization

Some results (1/2): achieved BER

- 4-QAM, no channel coding

Blind Equalization

Second solution: blind equalization

Iterative processus, without pilot

Transmitted signal

Receivedsignal Recovered

signal

H F ��

x y

w

z a+AFB

Channel Blind equalizer

- Goal: achieve H.F =1

- Constraint : limited knowledge of the signal feature

H F ��+

AFB

Blind Equalization

Second solution: blind equalization

H.F H.F H.F

Itération 0 Initial state Transient Steady state

Gain in terms of bit rate

Requires a convergence delay

state

Blind Equalization

Basics (1/2), [5], [6], [7]: AFB

FFT

PPN

……

……

y0,n

y1,n

ym,n

yM-1,n

INP

UT

Equal.am,nFm,n

- Goal: Solve the optimizationproblem

yM-1,n

- How? Use the stochastic gradient

with

- µ is the step-size parameter- J a given cost function

2 degrees of freedom

Blind Equalization

Basics (2/2), [5], [6], [7]:

Cost function: Constant modulus algorithm [5], [6]

with p: another degree of freedom

Transient state = 1000 OFDM/OQAMsymbols !

Blind Equalization

Some results: blind equalization in mobile environment (1/2)

MSE = (1/M).Σ(|x|-|a|)²

SNR = 20 dB

Fast increase

Blind Equalization

Some results: blind equalization in mobile environment (2/2)

Blind equalization does not track the channel variations

Blind Equalization

Any solutions ? … Yes !

Several ways are investigated:

Can be reconsidered [c5]

Can be adapted to OFDM/OQAM [c6], [c7]

New blind receiver designAFB

FFT

PPN

……

……

y0,n

y1,n

ym,n

yM-1,n

INP

UT

Equal.am,nFm,n

Bibliography[1] M. Bellanger , “Specification and Design of a Prototype Filter for Filter Bank Based Multicarrier Transmission,” in proc. Of ICASSP, vol. 4, Salt Lake City, UT, May 2001, pp. 2417 – 2420.[2] E. Kofidis, D. Katselis, A.Rontogiannis, and S. Theodoridis , “Preamble-based channel estimation in OFDM/OQAM systems: A review,” Signal Processing, Elsevier, vol. 93, pp. 2038 – 2054, January 2013. [3] C. Lélé, J.-P. Javaudin, R. Legouable, A. Skrzyp czak, and P. Siohan , “Channel estimation methods for preamble-based OFDM/OQAM modulations,” European Transactions on Telecommunications, vol. 19, no. 7, pp. 741 –750, November 2008. Transactions on Telecommunications, vol. 19, no. 7, pp. 741 –750, November 2008. [4] J.-P. Javaudin, D. Lacroix, and A. Rouxel , “Pilot-Aided Channel Estimation for OFDM/OQAM,” in VTC’03 - Spring, vol. 3, Jeju, Korea, April 2003, pp. 1581 – 1585. [5] D. N. Godard , “Self-Recovering Equalization and Carrier Tracking in Two-Dimensional Data Communication Systems,” IEEE Transactions on Communications, vol. com-28, no. 11, pp. 1867 – 1875, November 1980.[6] B. Farhang-Boroujeny , “Multicarrier Modulation With Blind Detection Capability Using Cosine Modulated Filter Banks,” IEEE Transactions on Communications, vol. 51, no. 12, pp. 2057 – 2070, December 2003.[7] A. Goupil and J. Palicot , “New Algorithms for Blind Equalization: The Constant Norm Algorithm Family,” IEEE Transactions on Signal Processing, vol. 55, no. 4, pp. 1436 – 1444, April 2007.

BibliographyContributions:

[c1] V. Savaux, F. Bader , “Mean Square Error Analysis and LMMSE Application for Preamble-Based Channel Estimation in OFDM/OQAM Systems,”[c2] V. Savaux, F. Bader, Y. Louët “A Joint MMSE Channel and Noise VarianceEstimation for OFDM/OQAM Modulation,” [c3] L. Caro, V. Savaux, D. Boiteau, M. Djoko-Kouam , Y. Louët “Preamble-Based LMMSE Channel Estimation in OFDM/OQAM Modulation,” in proc of VTC’15 Spring, Glasgow, May 2015Glasgow, May 2015[c4] V. Savaux, F. Bader “Pilot Adaptation for Broadband LTE-Like FBMC System in PMR Band,” in proc of VTC’15 Spring, Glasgow, May 2015[c5] V. Savaux, F. Bader, J. Palicot “Frequency Blind Equalization Using Constant Norm Algorithm for OFDM Systems,”[c6] V. Savaux, F. Bader “Sub-Optimal Initialization for Blind Equalization with Fast Convergence in OFDM/OQAM Modulation,”[c7] V. Savaux, F. Bader “Enhancing the Constant Modulus Algorithm for Blind Equalization in OFDM/OQAM Modulation,”

Thanks for your attention !

Merci de votre attention !

Des questions ?