[By Subject] [By Date] [By Sender [By Thread]
Previous In Time

Next in Time

From: leif.asbrink@mbox300.swipnet.se
Subject: Spread Spectrum and weak signals
Date: Sat, 7 Dec 96 02:04:23 PST
Hi all,

There seems to be an idea that spread spectrum might be a good idea to reduce power requirements for weak signal communications such as EME.

Fundamentally this has to be incorrect. Spread spectrum may average out QSB if the spectrum is wide enough, but it will not improve the average S/N as long as the noise floor is white noise. (And if the noise is not white, the best strategy is to select the frequency where the noise is lowest.)

Numbers like 20 dB for spread spectrum systems over simple morse code have been claimed, but I seriously doubt that number comes from a realistic evaluation of how the morse code is actually used by the weak signal community today. If you assume 10 dB S/N in a 200Hz bandwidth to be necessary, the start assumption is wrong by 20 dB !!

The real signal to noise ratio of the practical, commonly used EME signal is about 0 dB in a 20 Hz bandwidth. This is the average S/N, but the actual QSO is based on copying a few letters now and then on the QSB peaks, when the signal is something like 7 dB above the noise, still in a 20 Hz bandwidth. (Most operators do not use such narrow filters, but they have a BIG DSP between their ears doing the equivalent).

Now, if QSB is averaged out by synchronous jumping in frequency with both transmitter and receiver, the detect problem will not change. We (our computer) still would need to find out if the total amount of energy within a narrow pass band is noise only or if it is signal plus noise. The detect problem is well known, and a synchronous detector - frequency mixing down to DC - followed by averaging is the best way, but it requires knowledge of the phase.

I do not suggest that the detect process should be done by analog circuits, I just give the analog version of what the DSP would do. If using FFT, it is not meaningful to sample over longer times than the coherence time of the signal we look for. If spread spectrum is not used, the coherence time is about 0.25 seconds, corresponding to about 4 Hz bandwidth on 144 MHz.

To what extent the use of spread spectrum will average out the phase modulation in the EME path is unknown as far as I know, but I doubt it is going to allow a very significant reduction of the bandwidth. The only possible advantage of spread spectrum would have to rely on that. Averaging the QSB out is not an advantage. A clever computerised EME system would use the QSB to an advantage in analogue to what we already do when using morse code.

Of course frequency shift keying will improve simply by increasing the average power, but by shifting between several frequencies rather than two, it may also increase the data rate. Such methods, trading power against time squared, are possible ways to make EME between two oscar class stations, but they have nothing to do with spread spectrum.

The current state of the art in 144MHz EME is well represented by the UNKN422 challenge on the AF9Y home page. I regularly work stations on random at this level during contests, but it usually takes me something like four to six one minute periods to become certain on a new and unknown call sign. You may check with FFTDSP on this signal. In order to have some certainty in detect, one would need something like 10 seconds. For a frequency shift method, 5 seconds would be enough due to the doubled average power. To gain 20 dB on this, an averaging time of 50000 seconds will be needed and that is clearly unrealistic regardless of how many frequencies we use to improve the information rate.


What I try to say with this posting is : MORSE CODE IS REALLY EFFICIENT, AND VERY HARD TO BEAT !!!!!

When I work EME, I regularly use 17Hz bandwidth for weak signals. At such bandwidths, I have found it useful to use an expander, a variable gain device that amplifies stronger signals more and weaker signals less. I am well aware that many weak signal operators dislike narrow filters because of their ringing and prefer to do all the filtering "biologically". On my home page, there is a demo of a weak EME station recorded with various receivers. I would be interested to know from which of these recordings you are able to hear who is "XX", the station in this demo.


73 Leif / SM5BSZ

------ Submissions: vhf@w6yx.stanford.edu Subscription/removal requests: vhf-request@w6yx.stanford.edu Human list administrator: vhf-approval@w6yx.stanford.edu