Method and device for signal processing and applied to microwave communication system

A signal processing method applied to a microwave communication system, the method comprising:

• performing phase noise estimation on each input signal to obtain a phase noise estimation value of each input signal;
• performing channel estimation on each input signal by using the phase noise estimation value to obtain a channel estimation value, comprising,
• multiplying each input signal by a constant matrix A by using a first matrix multiplier to obtain a first multiplication result,
• multiplying the first multiplication result by a conjugate of the phase noise estimation value by using a second matrix multiplier to obtain a second multiplication result, and
• multiplying the second multiplication result by a constant matrix B by using a third matrix multiplier to obtain the channel estimation value of each input signal, wherein the constant matrix A=pinv(S), the constant matrix B=pinv(delta*SH);
• pinv represents an inverse operation;
• and delta*SH represents selecting the first L lines of SH, L is the number of input signals, delta is an L×Np matrix, the first L lines of the L×Np matrix are an I matrix and the rest part is 0, and the S is a training sequence formed by the pilots;
• sending the channel estimation value into an equalizer to enable the equalizer to perform fast convergence by using the channel estimation value;
• performing equalization processing on each input signal by using the equalizer to obtain an equalized signal corresponding to each input signal;
• suppressing phase noise in the equalized signal by using a phase-locked loop to output an error signal and a received signal;
• and
• performing an iterative update on a filter coefficient of the equalizer by using the error signal.

A signal processing device applied to a microwave communication system, the device comprising:

• a phase noise estimator, configured to perform phase noise estimation on each input signal to obtain a phase noise estimation value of each input signal;
• a multi-path channel estimator, configured to perform channel estimation on each input signal by using the phase noise estimation value of each input signal to obtain a channel estimation value and send the channel estimation value into an equalizer;
• the equalizer, configured to perform equalization processing on each input signal to obtain an equalized signal corresponding to each input signal, and output the equalized signal to a phase estimator and a signal rotator, and perform fast convergence by using the channel estimation value;
• the phase estimator, configured to perform phase estimation on phase noise in the equalized signal to obtain an estimated phase of the phase noise, and output the estimated phase to the signal rotator;
• the signal rotator, configured to perform phase rotation on the equalized signal to offset the estimated phase of the phase noise to obtain a phase rotation signal and output the phase rotation signal to a phase-locked loop;
• the phase-locked loop, configured to suppress residual phase noise in the phase rotation signal to output an error signal and a received signal,
• wherein the equalizer is further configured to perform an iterative update on a filter coefficient of the equalizer by using the error signal;
• and
• wherein the multi-path channel estimator comprises:
• a first matrix multiplier, configured to multiply each input signal by a constant matrix A to obtain a first multiplication result,
• a second matrix multiplier, configured to multiply the first multiplication result by a conjugate of the phase noise estimation value to obtain a second multiplication result, and
• a third matrix multiplier, configured to multiply the second multiplication result by a constant matrix B to obtain the channel estimation value of the each input signal, wherein the constant matrix A=pinv(S), the constant matrix B=pinv(delta*SH);
• pinv represents an inverse operation;
• S is a training sequence formed by pilots;
• and delta*SH represents selecting the first L lines of SH, L is the number of input signals, delta is an L×Np matrix, and the first L lines of the L×Np matrix are an I matrix and the rest part is 0.

A signal processing device applied to a microwave communication system, the device comprising:

• a phase noise estimator, configured to perform phase noise estimation on each input signal to obtain a phase noise estimation value of each input signal;
• a multi-path channel estimator, configured to perform channel estimation on each input signal by using the phase noise estimation value to obtain a channel estimation value and send the channel estimation value into an equalizer;
• the equalizer, configured to perform equalization processing on each input signal to obtain an equalized signal corresponding to each input signal and output the equalized signal to a phase-locked loop, and perform fast convergence by using the channel estimation value;
• the phase-locked loop, configured to suppress phase noise in the equalized signal to output an error signal and a received signal,
• wherein the equalizer is further configured to perform an iterative update on a filter coefficient of the equalizer by using the error signal;
• and
• wherein the multi-path channel estimator comprises:
• a first matrix multiplier, configured to multiply each input signal by a constant matrix A to obtain a first multiplication result,
• a second matrix multiplier, configured to multiply the first multiplication result by a conjugate of the phase noise estimation value to obtain a second multiplication result, and
• a third matrix multiplier, configured to multiply the second multiplication result by a constant matrix B to obtain the channel estimation value of each input signal, wherein the constant matrix A=pinv(S), the constant matrix B=pinv(delta*SH);
• pinv represents an inverse operation;
• S is a training sequence foamed by pilots;
• and delta*SH represents selecting the first L lines of SH, L is the number of input signals, delta is an L×Np matrix, and the first L lines of the L×Np matrix are an I matrix and the rest part is 0.