Multi-cell processing architectures for modelling and impairment compensation in multi-input multi-output systems
A method for predistortion including receiving a plurality of input signals forming a multiple-input signal in a multiple-input multiple-output system, generating a pre-distorted multiple-input signal from the received multiple-input signal, generating a multiple-output signal by feeding the pre-distorted multiple-input signal into a multiple-input and multiple-output transmitter, estimating impairments generated by the multiple-input and multiple-output transmitter, the impairments including nonlinear crosstalk between distinct ones of the plurality of input signals; and adjusting the pre-distorted multiple-input signal to compensate for the estimated impairments.
1. A method for predistortion, the method comprising:
receiving a plurality of input signals forming received multiple input (MI) signals;
generating a pre-distorted MI signal from the received MI signals;
generating a multiple-output (MO) signal by feeding the pre-distorted MI signal into a multiple-input and multiple-output (MIMO) transmitter;
estimating a MIMO digital predistortion (DPD) behavioral model of the MIMO transmitter by comparing the pre-distorted MI signals and the MO signal of the MIMO transmitter; and
adjusting at least one signal in the received MI signal to generate the pre-distorted MI signal to compensate for impairments in the MIMO transmitter by applying a processing function identified by the behavioral model to the at least one signal.
2. A method for predistortion, the method comprising:
receiving along a plurality of signal paths respective input signals forming a received multiple-input (MI) signal in a multiple-input multiple-output (MIMO) system;
generating a pre-distorted MI signal from the received MI signal;
generating a multiple-output (MO) signal by feeding the pre-distorted MI signal into a MIMO transmitter;
estimating a MIMO digital predistortion (DPD) behavioral model by comparing the pre-distorted MI signals and the MO signals of the MIMO transmitter; and
for each signal path, adjusting at least one of the received MI signals to compensate for the estimated impairments, wherein generating the pre-distorted MI signals includes pre-processing, based on the behavioral model, of the at least one of the received MI signals, using:
first linear and nonlinear processing to compensate for nonlinear distortions of the MIMO transmitter; and
second linear processing to compensate for linear distortions of the MIMO transmitter.
3. The method of claim 2 , wherein the first linear and nonlinear processing is based on analytic processing functions.
4. The method of claim 2 , wherein the first linear and nonlinear processing is based on neural networks.
5. The method of claim 2 , wherein the first linear and nonlinear processing is based on look up tables.
6. The method of claim 2 , wherein the adjusting includes introducing linear and nonlinear distortions in the signal path of the MI signal.
7. The method of claim 2 , wherein the adjusting includes introducing interference between signal paths of the MI signal.
8. The method of claim 2 , wherein the adjusting includes introducing interference between each signal path of the MI signal.
9. The method of claim 2 , wherein nonlinear processing includes:
processing the MI signal and the MO signal to determine a desired MO signal that pre-compensates for the nonlinear distortions; and
estimating, based on the desired MO signal, nonlinear processing functions.
10. The method of claim 2 , wherein linear processing includes:
processing the MI signal and the MO signal to determine a desired MO signal that pre-compensates for the linear distortions; and
estimating, based on the desired MO signal, linear processing functions.
11. The method of claim 2 , wherein the non-linear and linear processing includes:
processing the MI signal and the MO signal to determine a desired MO signal that pre-compensates for the non-linear and linear distortions;
estimating, based on the desired MO signal, linear processing functions; and
estimating, based on the desired MO signal, nonlinear processing functions.
12. The method of claim 2 , wherein said DPD behavioral model is estimated for combinations of each of the MI signals and each of the MO signals to forming matrix of preprocessing cells, wherein each element of the matrix of pre-processing cells models a behavior of the MIMO system.
13. A predistorter for a transmitter, comprising:
multiple transmit signal paths forming a multiple-input MI signal received from MI signals for feeding to a MIMO transmitter; and
a preprocessor having:
a MI for receiving the MI signal and for generating a pre-distorted MI signal from the received MI signal; and
a multiple-output (MO) for feeding the pre-distorted MI signal to the MIMO transmitter, and wherein the preprocessor is configured to estimate impairments generated by MIMO transmitter and adjust the pre-distorted MI signal to compensate for the estimated impairments, the estimates based on a MIMO digital predistortion (DPD) behavioral model of the MIMO transmitter by comparing the pre-distorted MI signals and the MO signals of the MIMO transmitter.
14. The predistorter of claim 13 , the preprocessor further including a matrix of pre-processing cells wherein each of the pre-processing cells of the matrix includes:
nonlinear processing blocks compensating for MIMO nonlinear distortions and an effect of interference between signal paths of the MI signal and signal paths of the MO signal; and
linear processing blocks compensating for MIMO linear distortions and the effect of interference between the signal paths of the MI signal and the signal paths of the MO signal.
15. The predistorter of claim 14 , wherein the nonlinear processing blocks are configured to: process the MI signal and the MO signal to determine a desired MO signal that pre-compensates for the nonlinear distortions; and
estimate a nonlinear function for each nonlinear processing block.
16. The predistorter of claim 14 , wherein the linear processing blocks are configured to:
process the multiple-input signal and the multiple-output signal to determine a desired multiple-output signal that pre-compensates for the linear distortions; and
estimate a linear function for each linear processing block.
17. The predistorter of claim 14 , wherein the non-linear and linear processing blocks are configured to:
process the multiple-input signal and the multiple-output signal to determine a desired multiple-output signal that pre-compensates for the non-linear and linear distortions, respectively;
for the non-linear processing blocks, estimate a non-linear function for each nonlinear processing block; and
for the linear processing blocks, estimate a linear function for each linear processing block.
18. The predistorter of claim 14 , wherein each of the pre-processing cells of the matrix models a behavior of a multi-input multi-output system.