| |
PA2AGA > HDDIG 11.06.00 02:09l 167 Lines 7027 Bytes #-9448 (0) @ EU
BID : HD_2000_161B
Read: GUEST
Subj: HamDigitalDigest 2000/161B
Path: DB0AAB<DB0SL<DB0RGB<DB0MRW<DB0ERF<DB0FBB<DB0GOS<DB0PKE<DB0ACH<PI8JOP<
PI8ZAA<PI8GCB<PI8HGL
Sent: 000610/1553Z @:PI8HGL.#ZH1.NLD.EU #:49834 [Den Haag] FBB $:HD_2000_161B
From: PA2AGA@PI8HGL.#ZH1.NLD.EU
To : HDDIG@EU
Date: Sat, 10 Jun 00 15:32:53 MET
Message-Id: <hd_2000_161B>
From: pa2aga@pe1mvx.ampr.org
To: hd_broadcast@pa2aga.ampr.org
X-BBS-Msg-Type: B
to be modest.
The system is to offer symbol rates related to the tone spacing at
(tone spacing / n) baud steps where n = 2 to 8.
Transmission tones (and receiver as well) should be adjustable to suit
different transceiver IF filters. Low (~1kHz) and high (~2 kHz)
alternatives to be accommodated without change of actual tone spacing.
Bit stream oriented transmission, allowing convolutional code FEC,
Varicode and binary transfer options. This approach would also allow
data block transmission over sequential tones, such as is used in
Piccolo (two sequential tones per character).
The actual buad rate and coding technique in use should preferably
be obvious from the nature of the signal, or indicated by the receiver
software. I guess a fallback position would be CWID with mode. The fewer
the modes, the lesser the problem of course.
C. Receiver
===========
Receiver to use non-coherent demodulation, probably using an FFT filter
and demodulator, integrating the signal over the whole symbol tone
period by sampling the complete period. Decoder to use bit-wise soft
decisions and offer various data and FEC decoding options.
Decoder to provide tuning indication, AGC and AFC as well as signal
performance metering. This mode will be very sensitive, so consideration
needs to be given to accurately tuning inaudible signals.
The symbol rate (and FFT solution rate) should be settable at
(tone spacing / n) baud steps in concert with the transmitter, where
n = 2 to 8 to ensure orthogonality.
D. Other Requirements
=====================
Symbol synchronisation using detected bitstream metrics or other
techniques not reliant on signal amplitude. A user interface with
the now-expected features such as waterfall display, point and click
tuning, signal strength and quality meters, separate TX and RX windows,
type-ahead buffer and simple macros.
Mode menu offering the different symbol rates and coding options.
CWID and a transmitter tuning signal. Transmit control via VOX or
serial port. Receiver squelch option.
E. Platform
===========
Pentium class PC with SB16 sound card or equivalent and VGA display.
Serial port PTT control as per GNR. Operating system Win95, 98 or NT.
Should operate on at least a Pentium 60. Design, specifications and
source code to be open. Parallel development of a Linux alternative
would be ideal. How about the Motorola EVM or the PTC-II?
User software to be copyright, all rights retained by the various
authors, but for free use by all Amateur users. No commercial or
military use permitted and no responsibility for damages or lack of
performance assumed.
F. Project Plan
===============
Following acceptance of this broad specification by those volunteers
involved in the project, a system block diagram would be issued and
agreed. At this point system software writer(s) and integrator(s) would
be selected or volunteer, and other jobs apportioned accordingly. The
integrator(s) would probably issue requests for suitable algorithms,
and a certain amount of testing and research would take place.
Also at this point a detailed specification for Phase 1 would be issued.
Specifications for Phase 2 and Phase 3 would follow discussion of
results of the previous stages.
Phase 1: a test platform with selectable baud rates, simple ASCII
coding and no error correction. Intended to optimise symbol
synchronisation and demodulator performance. User interface can be
incomplete.
Phase 2: a test platform for FEC and Interleave options. User interface
should be reasonably complete, with creature features and tuning display.
Metering (especially decoder metrics) should be operational. At this
point a beta version might be released.
Phase 3: a released product for general Amateur use. This will be
thoroughly tested and debugged, have optimised performance and a
limited set of the best FEC, data code and baud rate options. This
version will be fully documented prior to release. Hopefully performance
comparisons with other codes will have been completed prior to release.
========================================================================
Notes
1.
MFSK has very good performance on fading channels when using
non-coherent demodulation. The performance improves as the number
of channels (tones) increases. Because no assumptions of signal
phase are made, provided the phase errors introduced by the path
do not lead to excessive cross talk in the receiver filters, the
technique is essentially immune to "Ionospheric Doppler" effects.
The sensitivity and crud rejection of MFSK is also very good, since
(a) the bandwidth per detected tone is low, (b) the signal is
integrated over the whole symbol duration, and (c) the soft decision
bitwise decoder provides best possible S/N of the result.
Typing speed of around 25 WPM would be adequate. Faster options
would be preferable when conditions allowed. Slow alternatives for
the QRP/LF/VHF folk.
2.
8 tone MFSK allows transmission of three bits per symbol.
Tone spacing of 31.25 Hz would allow baud rates as high as 1/2T
(15.625 baud - 47 bps) and a transmission bandwidth of about 300 Hz.
47 bps with a 5 bit data code and R=1/2 FEC coding would provide
4.7 characters / sec (47 WPM). Mere mortals would use lower baud
rates to achieve better performance at more modest typing speeds.
Although not part of this specification, the transmitter, the
receiver technique and the soft decoder would all permit n-of-8
tone operation, with improved throughput at a modest power penalty.
For example 2-of-8 would allow at least 4 bits/symbol and 64 bps.
Also not part of this specification, addition of an ARQ layer
would permit a similar mode to be used for very effective automatic
link applications.
Speed suggestions are based on an 8 kHz sound card sample rate.
Modest linearity requirements imply a certain tolerance of poor
transmitter adjustment by amateurs who know no better (c.f. the
problems with PSK31). Use of continuous phase tones with no AM
would help in this regard.
3.
A bit stream receiver with soft decoding will allow very good
FEC to be deployed. A R=1/2 K = 7 Viterbi system with very fast
decoder is offered by Phil KA9Q. Interleaving is recommended.
Lower baud rates could be simply achieved by summing sequential
results from the decoder prior to the soft decision point. This
technique would of course require logic to decide which samples
belonged together (correct samples will have high cross
correlation).
4.
A "Costas Signal" synchronising system is suggested by Phil Karn
KA9Q. This consists of a unique symbol sequence interleaved with the
data that has a strong correlation response. Such a system is very
robust in multi-tone orthogonal systems, and does not have any
transmission symbol envelope shaping requirements. It also does not
To be continued in digest: hd_2000_161C
Read previous mail | Read next mail
| |
