Blind M-FSK Modulation in the ISI Channel

Abstract

Channel estimation has received considerable attention over the years for its contribution to more reliable signal decoding. General wireless communication environment would cause multi-path fading for signals that propagate through them. Multi-path fading has two major effects on the system; causing inter symbol interference (ISI) and reshaping signal constellation. Estimating the channel would enable us to combat these two effects. Channel estimation can be done either blindly or with the help of training sequences.

In this thesis, we propose a new blind channel estimation technique for M-FSK modulation systems. Our method can decrease the effect of signal reshaping and thus decreasing the probability of error. It also has the ability to track the channel variations in a time-variant environment.

In our method, an initial estimation is assumed as the channel impulse response. Utilizing this channel, received signals are demodulated and decoded. Based on output of the demodulator, a new estimation is generated for the channel. Consequently, a new output can be produced by exploiting the new channel estimate. This process can be done iteratively $n$ times to reach the minimum possible probability of error.