Fixed Point Realization Of Channel Estimation Algorithm Based On Fft For Ofdm Systems

Orthogonal frequency division multiplexing(OFDM)is a special case of multi-carriertransmission that can accommodate high data rate requirement of multimedia based wirelesssystems[1]. Channel estimation is an integral part of OFDM systems, also is a challengingproblem in all wireless systems.Unlike other guided media, the radio channel is highly dynamic. Transmitted signals aretypically reflected and scattered, arriving at receivers along multiple paths. In addition, due to themobility of transmitters, receivers, or scattering objects, the channel response can change rapidlyover time. Multi-path propagation, mobility, and local scattering cause the signal to be spread infrequency, time, and angle. These spreads, which are related to the selectivity of the channel,have significant implications on the received signal.Two FFT based channel estimation techniques employed in OFDM systems are presented.In addition, the advantages, drawbacks, and relationship of these estimation techniques with eachother are analyzed and discussed. Although the existing techniques differ in terms ofcomputational complexity and their mean squared error(MSE)performance, a suitablemethod among the presented techniques can be applied. based on a given system's resources andspecifications,Four schemes of the fixed point counterpart of the channel estimation algorithm are realized.obtained several results through simulation. First the fixed point DIF(Decimation InFrequency, DIF)algorithm has one bit gain than the fixed point DIT(Decimation In Time,DIT)algorithm. Second when the overflow are tested we can save 1~2 bit than directly rightshiftone bit before each addition. Three the fixed point algorithm performance decreases as theFFT(Fast Fourier Transform, FFT)point increases. The mean squared error (MSE)performances are compared between the float point channel estimation algorithm and its fixedpoint counterpart. The simulation result shows that these two performances are almost the samewith 16bit data quantization.