owx/docs/FSK

/* $Id$ */

Audio FSK support in OWX has been added to aid in programming radios in 
the field, using only a smartphone and a simple, microcontroller-based 
programmer, which will be designed at a later time. Using owx-fsk it is 
possible to convert binary files to files that can be played to the 
programmer through a speaker or an audio jack of the smartphone in the 
field. Programmer then demodulates the tones and programs the radio. In 
fact, any binary file (not only Wouxun memory file) can be converted to 
AFSK using owx-fsk tool.

owx-fsk should be used as a filter. It reads binary data from stdin and 
outputs raw data that should be fed to a lossless audio encoder. Example:

owx-fsk < data.bin | sox -t raw -b 16 -c 1 -e unsigned-integer -r 44100 - out.wav
 
Data format is 16-bit, little-endian, unsigned integer, monaural data 
sampled at 44100 Hz.

This file describes how exactly binary data is modulated.

Modulation is a three-layer process. First layer is a Bell 202 modem 
modulation, the same that is used for APRS. It uses 1200 Hz tone for mark 
(binary 1), 2200 Hz tone for space (binary 0) and each of these tones 
lasts for 833 us (1/1200s), giving a bitrate of 1200 bps.

Second layer is the bi-phase (Manchester coding) layer. It encodes each 
input bit into two output bits. Bit '0' is encoded as '10' and bit '1' is 
encoded as '01'. This way it is much easier for a decoder to keep it's 
clock in sync with the signal clock. First bit will always be '0', because 
LSB of the sync byte is '1', which translates to '01'.

Third layer is a packet layer. It defines packets.

In OWX, bytes are encoded LSB first and multi-byte variables are encoded 
in little-endian order. Also after each byte a parity byte follows. Parity 
chosen is an even-parity. It means that if a byte has an odd count of '1' 
bits, then parity byte is set to '1' to make it even, otherwise it is set 
to '0'.

OWX adds the concept of a packet. A packet consists of a header and packet 
payload. Header consists of:

- sync byte (0x53, that is ascii 'S')
- index byte (0x00 - 0xFF, then rolls back to 0x00)
- type byte (0x00 - start, 0x01 - data, 0x02 - end)
- payload size byte (one byte, 0x00 - 0xFF, for packets other than 0x01 it is 0x00)
- crc16 (two bytes; crc16 includes header - for calculation crc is 0x0000)
- payload (if payload size is greater than 0x00)

This way we are able to detect any error during the transmission.

Output sound can be summarized as:

- 1s of lead-in (mark tone)
- bi-phase encoded start packet
- bi-phase encoded data packets
- bi-phase encoded end packet
- 1s of lead-out (mark tone)