Compensation of I/Q gain mismatch in a communications receiver

A method of compensating I and Q gain mismatch in a sequence of input samples received over a channel, said method comprising the steps of:

• pre-calculating a plurality of training sequence convolution H matrices, each H matrix corresponding to a different channel length L;
• calculating an estimate of said channel and determining a channel length L therefrom;
• selecting a number of uncompensated training sequence input samples y and one of said pre-calculated plurality of H matrices in accordance with said channel length;
• calculating a gain estimate g given by
• ⁢
• g
• =
• Im
• ⁡
• (
• y
• _
• H
• )
• ⁢
• Im
• ⁡
• (
• H
• ⁢
• h
• _
• )
• Im
• ⁡
• (
• h
• _
• H
• ⁢
• H
• H
• )
• ⁢
• Im
• ⁡
• (
• H
• ⁢
• h
• _
• )
• where h is a vector of channel estimation taps;
• splitting said input samples into a real component and an imaginary component;
• and
• multiplying said imaginary component by 1/g and summing said real component therewith to yield a sequence of compensated input samples.

An apparatus for compensating I and Q gain mismatch in a sequence of input samples received over a channel, comprising:

• means for pre-calculating a plurality of training sequence convolution H matrices, each H matrix corresponding to a different channel length L;
• a channel estimation module adapted to generate an estimate of said channel including a channel length L;
• means for selecting a number of uncompensated training sequence input samples y and one of said pre-calculated plurality of H matrices in accordance with said channel length;
• means for splitting said input samples into a real component and an imaginary component;
• means for calculating a gain estimate g given by
• ⁢
• g
• =
• Im
• ⁡
• (
• y
• _
• H
• )
• ⁢
• Im
• ⁡
• (
• H
• ⁢
• h
• _
• )
• Im
• ⁡
• (
• h
• _
• H
• ⁢
• H
• H
• )
• ⁢
• Im
• ⁡
• (
• H
• ⁢
• h
• _
• )
• where h is a vector of channel estimation taps;
• and
• means for multiplying said imaginary component by 1/g and summing said real component therewith to yield a sequence of compensated input samples.

A communications receiver for receiving and decoding an M-ary transmitted signal, comprising:

• a radio frequency (RF) front end circuit for receiving and converting said M-ary transmitted signal to a plurality of baseband input samples;
• a DC offset compensation module comprising processing means programmed to:
• split said input samples into a real component and an imaginary component;
• calculate a gain estimate g given by
• ⁢
• g
• =
• Im
• ⁡
• (
• y
• _
• H
• )
• ⁢
• Im
• ⁡
• (
• H
• ⁢
• h
• _
• )
• Im
• ⁡
• (
• h
• _
• H
• ⁢
• H
• H
• )
• ⁢
• Im
• ⁡
• (
• H
• ⁢
• h
• _
• )
• where y is a vector of uncompensated training sequence input samples, H is a training sequence convolution matrix, h is a vector of channel estimation taps;
• multiply said imaginary component by 1/g and summing said real component therewith to yield a sequence of compensated input samples;
• a demodulator adapted to receive said compensated baseband input samples and to generate a received signal therefrom in accordance with the M-ary modulation scheme used to generate said transmitted signal;
• a first decoder operative to receive said received signal and to generate a sequence of soft symbol decisions therefrom;
• a second decoder adapted to receive said soft symbol decisions and to generate binary received data therefrom;
• and
• wherein M is a positive integer.

A computer readable storage medium having a computer program embodied thereon for causing a suitably programmed system to compensate for I and Q gain mismatch in a sequence of input samples received over a channel by performing the following steps when such program is executed on said system:

• splitting said input samples into a real component and an imaginary component;
• calculating a gain estimate g given by
• ⁢
• g
• =
• Im
• ⁡
• (
• y
• _
• H
• )
• ⁢
• Im
• ⁡
• (
• H
• ⁢
• h
• _
• )
• Im
• ⁡
• (
• h
• _
• H
• ⁢
• H
• H
• )
• ⁢
• Im
• ⁡
• (
• H
• ⁢
• h
• _
• )
• where y is a vector of uncompensated training sequence input samples, H is a training sequence convolution matrix, h is a vector of channel estimation taps;
• and
• multiplying said imaginary component by 1/g and summing said real component therewith to yield a sequence of compensated input samples.