HYBRID EVENT: You can participate in person at Rome, Italy or Virtually from your home or work.
ELOS 2023

Correct optical and Universal noise figures

Reinhold Noe, Speaker at Optic Conferences
Paderborn University, Germany
Title : Correct optical and Universal noise figures


Would one insert squarer + lowpass filter into the linear signal path when measuring the electrical noise figure (NF) Fe? Of course not. In line with this, one should no longer insert a direct detection photodiode into the linear signal path when measuring the optical NF. But the traditional optical noise figure Fpnf (so far called F) of E. Desurvire takes as powers the squares and variances of direct detection photocurrents. These “powers” are proportional to 4th powers of amplitudes (fields). This is in conflict with ~150 years of science and with Fe. which is defined for a linear system with 2 quadratures and power proportional to squared amplitudes. The optical in-phase and quadrature noise figure Fo,IQ avoids such contradictions. It is measured with a coherent I&Q receiver which has 2 available quadratures, like an electrical receiver.  For an amplifier with high gain one gets the optimum noise figure Fo,IQ = 1, while all prior optical noise figures are equal to 2 in this case. Fe and Fo,IQ have been combined into one universal noise figure FIQ for all frequencies. FIQ is the noise figure. Its limits in optical and electrical domains are Fo,IQ and Fe, respectively. For homodyne receivers with 1 quadrature, the optical homodyne noise figure Fo,I has been derived. It can also combined with the electrical homodyne noise figure Fe,I into a universal homodyne noise figure FI. While Fo,IQ and Fo,I differ, Fe and Fe,I are identical. All this and the generalized noise figures Ffas and Fase of H. Haus are discussed.

Audience Take Away Notes:

  • The traditional optical noise figure Fpnf is not a noise figure. It is in conflict with ~150 years of science. If it were a noise figure then power would be proportional to voltage taken to the 4th power. Its minimum value 2 does not reflect the fact that the sensitivity of an ideal optical I&Q receiver remains unchanged when it gets an ideal optical preamplifier. I have derived the correct optical I&Q noise figure Fo,IQ which is conceptually identical with the electrical noise figure Fe. Its minimum value equals 1, corresponding to 0 dB. I have also combined the two into one I&Q noise figure FIQ or simply F, the noise figure. This is important in the beginning convergence of electrical-THz-optical
  • In the convergence of electrical-THz-optical, conflicting Fpnf and Fe would be opposed. The correct Fo,IQ and FIQ = F reconcile optical with electrical. The sooner the practitioners understand that Fpnf is conceptually faulty and the sooner they use the correct Fo,IQ, the better.
  • The wrong optical noise figure Fpnf is in ~10000 heads. The sooner the correct optical noise figure Fo,IQ is taught, the better
  • It eliminates the coming quarrel between Fpnf and Fe. In this sense the job becomes more efficient
  • It will, in cases where there are optical and electrical effects.


Reinhold Noe obtained Dipl.-Ing. (1984) and Dr.-Ing. (1987) degrees in electrical engineering from Technical University of Munich, Germany. 1987-1988 he was with Bellcore, NJ, USA. 1988-1992 he was with Siemens Central R&D in Munich, Germany. Since 1992 he is Professor for Optical Communication and High-Frequency Engineering at Paderborn University in Germany.  He has published >300 papers. In 2006 he co-published the first coherent synchronous QPSK transmission with standard DFB lasers, in 2007 with polarization multiplex and electronic polarization control. In 2008 he co-received the Innovation Prize of the Land North-Rhine Westphalia. In 2010 he co-founded Novoptel GmbH.