OFDM based for WIFI Ziqi SONG Zhuo ZHANG Supervisor: Prabin

Overview Introduction –Problem of Multipath transmission –OFDM solution Realization in Simulink –System overview –Channel Model Fixed FIR Rayleigh –Channel Estimation Block-Type Comb-Type Conlusion

Multipath Transmission Multipath is the unique character of wireless system More than one transmission path between transmitter and receiver Received signal is the sum of many versions of the transmitted signal with varying delay and attenuation

Effect of Multipath on Received Signal ISI (Intersymbol Interference) –Received signal at any time depends on a number of transmitted bits

Effect of Multipath on Received Signal Frequency Selective Fading –Multipath fading causes some frequencies to be attenuated –Causes changes both in Amplitude and Phase –Approximately constant over narrow band

How OFDM Solves Problem Multi-Carrier Approach –Data transmitted in parallel –Cyclic prefix –Channel estimation and equalization

Cyclic Prefix If multipath delay is less than the cyclic prefix –no intersymbol or intercarrier interference –amplitude may increase or decrease

CP Interval vs. Multipath Delay If multipath delay is equal to or smaller than CP interval-perfect equalization

CP Interval vs. Multipath Delay If multipath delay is larger than CP interval, e.g. 1% larger –ISI emerging

OFDM System Overview

Channel Estimation Block-type (Training Symbol) Estimation symbols are transmitted periodically, all subcarriers are used as pilots Uses the estimated channel condition until next pilot symbol arrives Estimation can be based on LS, MMSE

Channel Estimation Comb-Type Pilots are uniformly inserted into subcarriers apart from each other Efficient interpolation technique is needed Based on linear, second-order, low- pass, spline cubic interpolation and so on.

Channel Model-FIR Ten taps and fixed complex coefficients

Without Equalizer BER=0.66 mainly caused by phase shift.

After Equalized Left :block-type training every 3 symbols Right: comb-type a pilot every 3 subcarriers

Channel Model-Rayleigh Impulse response

Channel Model-rayleigh Frequency response

Without Equalizer Rayleigh Channel plus additive Gaussian noise BER=0.795

After Equalized Using block-type estimation At the condition of 1)The pilot to data ratio is 1:2 2)The signal to noise is 20DB Get BER=0.005

After Equalized Using comb-type estimation At the condition of 1)The pilot to data ratio is 1:2 2)The signal to noise is 20DB Get BER=0.117

Block-Type Simulation Result

Comb-Type Simulation Result

Conclusion In our channel condition, block-type estimation is better The weightier pilots, the better estimation But trade-off between BER performance and efficiency