Noise balanced QAM detection

In a radio receiver in a communication system, a method of demodulating digital data, using M'ary QAM, comprising the steps of detecting a complex symbol vector D, establishing within which reference symbol boundaries the detected complex symbol vector D falls, the given reference symbol boundaries being associated with a complex reference vector, establishing quadrature components, which comprise a real quadrature component and an imaginary component, of an error vector, constituting a difference between the detected complex symbol vector D and the associated complex reference vector, and seeking to approximate an error control signal as a feedback signal in a demodulation stage, whereby when the detected complex symbol vector D falls within a first sector in a complex plane surrounding an imaginary axis (Q), the first sector being delimited by at least two lines crossing origin, the first sector being symmetrical with regard to the imaginary axis, approximating the error control signal by the imaginary quadrature component of the error vector, whereby when the detected complex symbol vector D falls within a second sector in the complex plane surrounding a real axis (I), the second sector being delimited by at least two lines crossing origin, the second sector being symmetrical with regard to the real axis, approximating the error control signal by the real quadrature component of the error vector;

• and when the detected complex symbol vector D belongs neither to the first sector nor to the second sector, approximating the error control signal by the mean value of the real quadrature component and the imaginary quadrature component.