FMT Modulation for Wireless Communication

Santosh V Jadhav(Advisor: Prof V M Gadre)

Multicarrier System:

Spectral partitioning: -sinc(f) overlapping (OFDM) -Non-overlapping (FMT)OFDM: FMT :

2( ) 1( ) , 0,1, , 1, 0,1, , 1ij ni Mh n e n M and i M

M

2( ) 1( ) ( ). , 0,1, , 1, 0,1, , 1ij ni Mh n h n e n M and i M

M

1( ) ( )

0

( ) ( ). ( )M

i i

i k

n A k h n kMx

( ) : , :h n LP prototype filter overlap

Spectral characteristics:OFDM

Spectral characteristics:FMT

OFDM Communication System

OFDM v/s FMT:OFDM: -perfect detection(zero ISI, zero ICI) under ideal channel and absence of noise. -simple equalization scheme with cyclic prefix -high spectral overlap.

FMT: -nearly rectangular frequency-domain amplitude characteristic -High spectral containment, negligible ICI -The absence of the cyclic prefix, together with

the reduced no of virtual carriers -Equalizer is required even under ideal channel condition and absence of noise

Why to prefer FMT over OFDM?

-Cyclic Extension in OFDM -> Loss of transmission efficiency

-High spectral overlap in OFDM -> Frequency offset sensitivity

-In OFDM side lobes are just 13dB lower than the main lobe

-> Significant power leakage into adjacent band

FMT System: Direct implementation

2( ) 1, ( ) ( 1). , 1,2, ,

j nii Mwhere g n h n e n MM

Efficient implementation: DFT based FMT system - FMT Transmitter-

( ), ( ) ( ) ( )thkwhere h m h mM k is the k phase of h n

Contd…

FMT Receiver-

Prototype Design:Prototype filter h(n) completely defines the

system. Allows tradeoff between- - No. of subcarriers - Level of spectral containment - Complexity of implementation and latency

Can be designed by any standard FIR filter design method. Typical design parameters are- - fcutoff : determines the ISI per subchannel - Stopband energy : determines ICI introduced by

adjacent subchannels, so should as low as possible

Contd…

e.g. -

Hamming Window=14, fcutoff = 0.38/T

Blackman Window=14, fcutoff = 0.3/T

Equalization in FMT System:Due to negligible ICI, each subchannel will

be independent of adjacent channel(s) as shown -

Simplified subchannel model is –

1( ) ( ) ( )

1 0

( ) ( ) ( ) ( )M L

i i ioverall

n p

h k g n c p h kM n p

Contd…Subchannel impulse response ( = 14):

So, ISI and per subchannel equalization is must.

Conventional Schemes and FMT specific equalizer design

-Channel is equalized by one-tap equalizer-ISI due to transmit and receive filter need to

be compensated.

Linear equalizer DFE equalizer

Contd…Here estimated error –

Optimality criterias – - Zero forcing(ZF) - MMSE

(i) (i) (i)(k) = d (k) - y (k)e

Simulation results with MMSEQPSK, M=64, =14:

Contd…8-PSK, TCM encoder, M=64, =14

Problems with conventional method and solution

Problems-The process of making hard decisions on the channel symbols

actually destroys information pertaining to how likely each of the possible channel symbols might have been.

Solution-1. Additional soft information can be converted into probabilities

that each of the received symbol takes on and can be exploited by a BER optimal decoding algorithm.

2. Let decoder generate its own soft information based on the soft information received from equalizer and use it in the process of equalization. So let soft information flow in both direction between decoder and equalizer for no of iteration -> Turbo Equalizer.

Turbo FMT System Model:

Contd…Turbo Equalizer- -Channel equalized by 1-tap filter -SISI equalizer neutralizes the effect of transmit and receive filters.

Performance of Turbo equalization based schemesMMSE-DFE(BPSK, Hamming, =10):

Contd…MMSE-LE(BPSK, Hamming, =10):

Future Scope:-Turbo equalization based scheme to be extended

to other higher order constellations (like QPSK, QAM)

-Turbo equalizer for STBC-FMT

-Multi-user detection in FMT Systems

-Adapting the System if flat fading assumption fails