machael
Member level 2
Hi guys:
I met some research trouble, and the problem is: I need to estimate the impulse response (CIR) time of a certain channel. Precisly, I need to get the time from the impulse response beginning to its peak, i.e. the rising time, by transmit an OFDM preamble signal between a pair of transceivers, the original preamble signal is known for both sides.
A general thought is: the receiver gets N samples of received preamble, and calculate its FFT. After synchronizing by its phase information, divide it by the FFT of standard preamble, so I get the channel transfer function(CTF) H(f). And after an IFFT to H(f), I got CIR h
. i.e. ifft(fft(rcv)./fft(preamble))
But the problem is: as there is a band-pass filter on the receiver, the premble signal is not a Full-Frequncey signal, but a partial subcarrier signal. If I perform the calculation as above, what I got is not the real CIR h, but a noncausal signal with huge ripple, and I can't get the rising time of CIR. :-(
Another thought is: if assuming the number of effective(non-zero) samples of CIR sequence h is far less than N, I can build an linear equations set by the CTF points, and by solving the equations set, I can got the h. But such algorithm also brings problem: 1. the calculation is too complex, and is not suited for embedded software implementation; 2. the equations set often has a near-singular matrix to solve, so the h' is often wrong. :-?
Because what I need is not a precise CIR, but only the rising timing of CIR from initial to peak, is there any better way to get it?
Does anyone can give me some advice or recommend some research paper related? I will appreciate it very very much!
Thank you!
I met some research trouble, and the problem is: I need to estimate the impulse response (CIR) time of a certain channel. Precisly, I need to get the time from the impulse response beginning to its peak, i.e. the rising time, by transmit an OFDM preamble signal between a pair of transceivers, the original preamble signal is known for both sides.
A general thought is: the receiver gets N samples of received preamble, and calculate its FFT. After synchronizing by its phase information, divide it by the FFT of standard preamble, so I get the channel transfer function(CTF) H(f). And after an IFFT to H(f), I got CIR h
. i.e. ifft(fft(rcv)./fft(preamble))But the problem is: as there is a band-pass filter on the receiver, the premble signal is not a Full-Frequncey signal, but a partial subcarrier signal. If I perform the calculation as above, what I got is not the real CIR h
, but a noncausal signal with huge ripple, and I can't get the rising time of CIR. :-(Another thought is: if assuming the number of effective(non-zero) samples of CIR sequence h
is far less than N, I can build an linear equations set by the CTF points, and by solving the equations set, I can got the h. But such algorithm also brings problem: 1. the calculation is too complex, and is not suited for embedded software implementation; 2. the equations set often has a near-singular matrix to solve, so the h' is often wrong. :-?Because what I need is not a precise CIR, but only the rising timing of CIR from initial to peak, is there any better way to get it?
Does anyone can give me some advice or recommend some research paper related? I will appreciate it very very much!
Thank you!
