A Novel one-tap frequency domain RLS equalizer combined with Viterbi decoder using channel state information in OFDM systems Advisor: Yung-an Kao Student: Chian Young

outline  Introduction  System overview  A Novel 1-tap frequency domain RLS equalizer  Viterbi Decoding with CSI  Simulation result

Introduction(1/3)  The advantage of OFDM ’ s parallel transmission scheme: frequency selective channel  multiple flat fading sub-channels  the sub-channel equalization in frequency is simple Transmit Spectrum Receive Spectrum Channel Training Tone Data Tone Channel Spectrum

Introduction(2/3)  The main factors that distort the signal in OFDM system: Channel Noise Frequency offset: CFO (carrier frequency offset), SFO (sampling frequency offset)

Introduction(3/3)  The benefit of the proposed equalizer structure: Compensate signal distorted by the channel, CFO and SFO at the same time The division operation is not required Combine with CSI to improve system performance

System overview (receiver) In the proposed FEQ structure, the constellation must be adjusted Signal affect by channel SFO, CFO, noise, etc.. CSI is obtain from 1-tap FEQ A modified RLS algorithm is used or … or

A Novel 1-tap frequency domain RLS equalizer(1/4)  Filtering equation: Definition of :  rewrite RLS filtering equation: magnitude phase Y k,l : EQ output X k,l :EQ input k : k-th sub-carrier l : l-th OFDM symbol λ k : forgetting factor 0<λ <1. Φ k,l :correlation matrix w k,l : equalizer weighting

A Novel 1-tap frequency domain RLS equalizer(2/4)  The update equation for  The update equation for Φ k,l  Error signal is not used  No division

A Novel 1-tap frequency domain RLS equalizer(3/4)  From,the magnitude is not compensated: 1-tap FEQ input 1-tap FEQ output

A Novel 1-tap frequency domain RLS equalizer(4/4)  We have to adjust the constellation of the equalized signal by Φ k,l times on each sub- carrier : Equalized signal constellation size multiply by Φ k,l times

Viterbi Decoding with CSI(1/3)  CSI (channel state information) Each sub-carrier experiences different flat fading (SNR) SNR high  high reliability SNR low  low reliability Those reliabilities can be collected as the CSI

Viterbi Decoding with CSI(2/3)  VITERBI DECODING WITH CSI We use CSI to reflect different sub- channel fading Viterbi decoding: select the path on code trellis with the minimum Euclidean distance Adding CSI when calculating the Euclidean distance  improve reliability on calculating the Euclidean distance

Viterbi Decoding with CSI(3/3)  The calculation of the Euclidean distance:  When SNR is high enough:  Adding CSI to D : the adjusted signal constellation : possible transmitted signal n : n-th de-interleaver output

Simulation environment & parameters  Follows IEEE a standard  CFO=3.125kHz,SFO=800Hz  Indoor Rayleigh fading multipath channel with Trms=50ns,Ts=50ns  1000 packets, 256bytes/packet

Simulation result PER performance for IEEE a (no CSI aided) PER performance for IEEE a (CSI aided)