I believe in the KISS principal for product design.

Transmission:
A heavily error corrected 7 bit ASCII subchannel (possibly using 8--14
modulation, with 2 ECC bits) is not hard to implement in software. Turbo
Codes or Viterbi Codes could be used at a later date, if the 2 layer EEC
(DRM + NUMBERS)layered EEC does not work well beyond 4 hops. Plain vanilla
DRM EEC allows DRM data to be error free within a range of 10,000 kms. A 2
layer EEC scheme for data could extend this to 12,000...14,000 kms. I would
assume a data rate no more than 120 baud, 60 baud nominal.

Data layer:
The hidden 'numbers' would just becomes a new DRM packet type. The packet
type is never published, and the packets are merely slipped into the
regional SW transmissions of that country's SW broadcaster. [From relay
stations under its direct control.]

Decoding:
There would need to be some parsing rules for the encrypted data stream to
allow for unit addressing (like pagers have), hard crypto (algorithms
swithable over time) and digital signatures for verification. XML could help
overcome a lot of compatibility issues over time. XML solutions already
exist for the character transmission level of this communications network.
Don't reinvent the wheel!

Decoder:
The decoding issues could be handled by PDA with 'add on' software, like a
fake DRM -- PDA audio converter.
Smart cards are a possibility here too.

Reception:
The DRM receiver would then need only support raw DRM bitstream output.
Using USB, not a challenge -- and not conspicuous either.

Overall:
There would still be 2 pieces of equipment, up from just 1 -- the SW
receiver.
The agent would be expected to remember a unique series of codewords to get
the unit to go into decoder mode, and initialise to decode their messages.
No 'one time pads' would need to be carried. The software would need to be
updated seasonally, to keep its security code fresh. One time pads could be
used as an add in option in the software. Self erasing software can be
written!

Users:
Admittedly, I would only recommend this for usage in Latin America,
Sub-Saharan Africa (populated regions) and Europe.
The concept is still usable in Asia, but not initially.

There is no perfect solution -- station chiefs and section / region
coordinators would test the software 1st.
If the mass of software and hardware finally jells into a workable
transmission system, great.
If it only becomes useable as a fallback transmission system, all is not
lost either.

DRM's waveforms have origins in NATO HF modem research, so the use of DRM
that I am suggesting is not even novel or innovative.


DRM : 7 bit ASCII teletype 'subchannel' : possible elemination of numbers
stations as an end product!

One forgets that DRM technology is improving all the time.
In 3 or 4 years a very portable DRM reciver, with a menu system could be
reprogrammed via a smart card.
A smart card or memory card (possibly with with a keypad and display) has
all the crypto power one needs.

-=-=-=-=-=-=-=-
A one time pad, written on water soluable paper, is no farther than a trip
down the drain or a dunk in a cup of coffee from being gone forever. I
suspect real spys want to stay as far as possible from spy gadgets.

Anyway, there's plenty of encrypted diplomatic and military traffic right
now on RTTY. RTTY can be easily decoded by SWLs, right now. What does
DRM
add to the picture?

Frank Dresser