On dual optical detection: homodyne and transmitted-reference heterodyne reception

The authors provide mathematical justification, using classical detection theory, for the observed superior performance of coherent dual optical detection over single-diode detection and establish a rationale for determining the intensity of the local oscillator. Viewing the dual detector as an optical channel with one input and two outputs, they show that in the limit of large local oscillator intensity, the optimal processor uses the difference of the two detector outputs, and consequently no bias subtraction needs to be carried out. Since phase is a major problem in applications of coherent optical communication, the authors propose modulation and detection formats that can possibly trade the effects of phase noise for the often less objectionable additive noise