[Laser] fundmentals

[Glenn Thomas]

   Hi James!
   Interesting  post. My comments interspersed. Much text deleted to save
   bandwidth.
   At 03:29 PM 2/12/2007, James Whitfield wrote:

     The fundamental goal of optic     stream of photons and decoding the information that it contains.

   I  agree.  In  fact,  this  is  also  the  fundamental  goal  of radio
   communications  as  well.  The  one  and  only  difference at the most
   fundamental  level  is  that  for  radio  the wavelengths are somewhat
   longer.
   <snip>

     To  simplify  further,  c     versus  "off"  states  of the photon stream instead of extracting a
     tone  impressed on the photon stream.  If we did have an OOK stream
     of  bits,  would  we  need  to impress FEC to deal with weak signal
     issues?   How would that compare to encoding the data on tones?

   This  is  the  basic  problem for communications in general. The first
   basic  principle  we  need to be dealing with is that the narrower the
   bandwidth  we need to examine for a possible signal, the less noise we
   have  to  deal with. Thus the optimum receiver bandwidth for any given
   signal is exactly the same bandwidth as the transmitted signal itself,
   no  more  and  no less. Less receiver bandwidth will eliminate some of
   the  signal thus producing additional noise in the form of distortion.
   More  receiver  bandwidth  will  admit more noise and thus degrade the
   signal  to noise ratio (SNR). Part of the design of any communications
   system,  radio,  optical,  gravitic  or  whatever,  is  to  match  the
   transmitter and receiver bandwidths.
   Once  that issue is dealt with, the next issue is determining if there
   is  a  signal present or not at any particular time. This is a process
   where  mistakes will be made - sometimes noise will look like a signal
   when  there  is  none and sometimes noise will make a signal look like
   it's  not  there.  Once the decision has been made that there is or is
   not  a  signal  present,  it  is critically important to know what the
   probability  is  that  the correct decision has been made. In general,
   higher  SNR  results  in  a  higher probability of a correct detection
   decision and lower SNR results in a lower probability.
   This  probability  must be more than 0.5 - the detector is more likely
   than  not  to  make  the correct decision. If it's so, then redundancy
   (aka FEC) can provide better signal recovery. For example, if each bit
   is sent three times and the result taken as a "vote" of the three, the
   probability  p  of  detecting the correct bit by voting is [3*(p**2) -
   3*(p**3)].  A  little  algebra  will demonstrate that if p < .5, we're
   more  likely  to  make  the  wrong decision and that voting makes this
   effect worse. If p=.25, then the probability of the vote getting the
   correct decision is less than 0.15. At the same time, if p=.75, then
   the  probability  of the vote getting the correct decision is 0.844, a
   12.5%  improvement.  Thus we have a measure of the effectiveness of 3x
   redundancy,  aka  FEC,  and the probability of correctly detecting the
   signal  in  the  first  place,  aka SNR, are related. Lesser levels of
   redundancy  will  be less effective and greater levels more effective,
   but  it's  clear  that the probability of the basic detection decision
   must be greater than 0.5.
   Speaking  of redundancy, aka FEC, it's clear that the higher p is, the
   better  FEC  will  work.  Put  another  way, the less you need it, the
   better  it works and the more you need it, the less it works. BTW, the
   method  of  taking  many  frames from a CCD detector (or a single long
   exposure) used by astronomers can be viewed as either a voting scheme,
   where  p  has to be >.5 for it to work at all, or as severely limiting
   the receiver bandwidth.
   That's  it.  The fundamental design process is to match the signal and
   receiver  bandwidths  and  then  maximize  the  probability  that  the
   detector  will  correctly  detect the presence or absence of a signal.
   Laser communication systems, like all communication system, need to be
   designed, however you do it, with these principles in mind.
   Imposing  an  800 Hz square wave on a laser, keyed with Morse code, is
   one  way  among many of doing this. An optical bandpass filter can get
   rid  of  much  of  the  light  noise, but it's bandwidth is still many
   orders  of  magnitude  wider  than a CW signal. Converting the photons
   into  an  electrical  signal  and  then  using a DSP filter to further
   narrow  the bandwidth will provide a filter that is closely matched to
   the signal bandwidth, i.e. a hundred Hz or so for a Morse code signal.
   Note  that  the  photon  detector will add noise of it's own that will
   degrade  the  probability  of being able to detect the signal, so this
   and  other  noise sources need to be identified and considered as part
   of the system design.
   <snip>

     James N5GUI

   73 de Glenn WB6W