Consistent Optical and Electrical Noise Figure

The optimum noise figure of an electrical amplifier is
F_e
= 1 and the optimum traditional noise figure of an optical amplifier is
F_pnf
= 2. This irresolvable conflict is due to the fact that
F_e
measurement requires electrical powers, proportional to squared amplitudes (voltages), while
F_pnf
measurement requires squares and variances of photocurrents, proportional to 4th powers of amplitudes (fields). In line with this an electrical receiver can receive I&Q parts of an electric carrier while a direct-detection photoreceiver can detect only power and not phase of an optical carrier. Optical amplifiers cause Gaussian field noise. Photodetection causes shot noise. Spontaneous-spontaneous beat noise in direct detection is taken into account by negative binomial or chi-squared photoelectron distributions, without errors of a Gaussian approximation. Coherent receivers linearly sense the Gaussian field noise. The sensitivity of an ideal coherent I&Q receiver is not degraded if it gets an ideal optical preamplifier, while the corresponding
F_pnf
= 2 suggests degradation. Coherent I&Q or heterodyne receivers have electrical output powers proportional to squared amplitudes (fields). This way one has the same metric in electrical and optical domain. One gets an optical noise figure
F_o
,
IQ
. For large amplifier gain it is
F_pnf
/2. In an ideal amplifier,
F_o
,
IQ
= 1. For true optical homodyne receivers and for direct detection receivers with Gaussian approximation it can be converted into
F_pnf
and vice versa. Phase-sensitive amplifiers are also covered. With
F_e
and the I&Q optical noise figure
F_o
,
IQ
a consistent unified noise figure is derived, valid and usable in electrical and optical domain.