[net.ham-radio] Oregon Packet Proposal

rick (12/13/82)

An open letter to Dave, WB7RDI,

     Saw your draft proposal for a packet radio network on
the net.  Here are a few comments/criticisms/suggestions:

	o   you might consider using delay modulation coding
	    instead Manchester.  Delay modulation has a minimum
	    at dc (as opposed to a null for manchester), yet 
	    with some minor waveform shaping it should pass
 	    the ac coupled radios without much distortion.

	o   Should you decide continue using Manchester, take a
	    look at the Zilog 8530 SCC.  It has a Manchester 
	    encoder/decoder on chip.  The receiver is not
	    quite optimal in the sense that it uses digital
	    techniques (digital pll) to regenerate the clock,
	    however, it should provide adequate results in your
	    application.  As you might expect, the chip provides
	    on board baud rate generators, bit stuffing, selectable
	    error detection polynomials, and universal microprocessor
	    interface (cost: ~$38).

	o   I was somewhat puzzled by your trying to evaluate
	    the CRC polynomials in software.  Most SCC's provide
	    CRC generation.  Even if your application requires
	    software generated error detection (cost reasons?),
	    there are checksums which are better suited to the
	    sort of arithmetic available on microprocessors (CRC
	    polynomial evaluation is long division, ie a serial process).
	    Take a look at IEEE Trans. on Communications,
	    January, 1982 pp 247-251 for an article describing
	    ones complement checksums.  The claimed fraction
	    of errors which pass undetected is .001538%, versus
	    .001526%for CRC's.  The ones complement technique is
	    not as good with double bit errors, however.

       o    I would strongly suggest that you consider using AX.25
	    packet format instead of your home grown format.  You might 
	    consider building a standard AX.25 network FIRST (something
	    stable for others to use), then set up a gateway to your
	    experimental network.

       o    Some other papers in Transactions on Communications which
	    I found useful are:

	    - L Ehrman, PF Maloney, "A Multimode Modem for Digital 
	      Radio", IEEE Trans. on Comm., Vol COM-27, pp 1786-1793,
	      Dec 1979.

	      This paper addresses upgrading Military FM Radios to
	      digital service.  The radios are similar in characteristics
	      to existing Amateur FM equipment.  They investigate various
	      modulation schemes including MSK, Manchester, and Delay
	      Modulation.

	    - JS Ma, "On the Impact of HDLC Zero Insertion and Deletion 
	      Link Utilization and Reliability", IEEE Trans. on Comm.,
  	      VOL COM-30, pp 375-381, Feb 1982.

	      This paper discusses the impact of bit stuffing on
	      Link error performance.

	o   Take a look at MSK!  The references are too numerous to
	    list here, just take a look a T Comm over the last few
	    years.  You will still need a baseband coding scheme which
	    inserts clocking info, however, it will buy you even more
	    spectral efficiency.

			Rick Spanbauer WB2CFV
			..!allegra!sbcs!rick