Spread spectrum is where, functionally equivalent to the to the randomizer, a pseudo random, or even truly random bits are added at a higher rate than the information bits. In a typical randomizer one bit is produced for each bit in. In the case of spreading, usually a significant number of “extra” bits are inserted at this point. These bits are not predicted by the input data. Instead, they are random in the sense they are nor correlated to the user data.
These extra spreading bits serve to reduce the probability that the transmit energy (its power spectral density) will be observed at any given time in any given bandwidth. These extra bits serve only to reduce the power spectral density over a bandwidth (narrow with respect to the transmit spectral density) but otherwise do not increase the efficiency of the end to end circuit (with one exception I will address shortly). By efficiency, I mean the amount of energy required to get one bit of the input information, prior to any coding, modulation etc., to the users output on the receiver with a particular error rate. (Typically characterized by a performance curve of Eb/No vs B.E.R.) The critical point is, with an end to end link of some particular source coding FEC coding and modulation, its end to end performance can be characterized. in a perfect word, if you then “spread that system” by adding extra bits unrelated to the input information, and at receiving side, you knew how to despread, or remove the spreading bits, the link will have the same end to end performance. That is, adding a spread spectrum system around a communications link does not make it work better, and in most real world will actually degrade the end to end performance. This is because the processes used to despread are never perfect. So in a spread system, you have a transmit signal that covers a wider spectrum than the original link, but because the same energy is used, the power spectral density, the amount of energy per unit bandwidth, is reduced. This reduced density has some advantages if you are trying to hide the fact that you are communicating by making your transmit spectral power less than the noise level. Of course if the fellow you are trying to hide from can get close to your transmitter, you pop up from under the noise, and game over. Two other reasons to use spread spectrum, one very real (but not for typical hams) and one a bit illusionary. The reason for most of the spread spectrum in real use is called CDMA, Code Division Multiple Access. Most hams use FDMA, frequency Division Multiple access. For an FDMA example, a great many of us access the 20m band at once, but the multiple accesses to the band are done by each user being on his own frequency (Frequency Division). Of course in this case the stronger user on a given frequency and given path effectively has the access to the channel. In CDMA each user uses the full band, at the same “carrier” frequency, but each transmitter has a spreading code that is unique. At the receiver, the desired link is “tuned” by dispreading with the same extra bit sequence as was used at the transmitter to spread. Signals in the bandwidth having different codes will appear to the despread process as random noise, once the wide band signal desired is despread to a narrow band link. On the receive end of a spread link, the rejection of other spreading codes is also applied to any other signal. The dispreading process will spread the energy of an interfering signal over the spread bandwidth. An example: Assume a unspread link of an occupied bandwidth of 10 kHz and a power of 10 watts. This will have a power spectral density of 10 watts per 10 kHz, or 1 watt per kHz, or 1 mw per Hz. Assume this is spread with a 1 Mchip/s digital signal using BPSK modulation. The 10w/10 kHz watts is now effectively spread over 1 MHz of bandwidth reducing the power spectral density to 10 watts per 1000 kHz, or 1 watt per 100 KHz, or 1 mW per 1 kHz or .001 uw per Hz. The spreading factor is the 10 Log(Spread BW/unspread BW) = 10*log(1000/10) = 20 dB. Thus 1 mW/Hz (-30 dBW/Hz) is reduced to .001 mW/Hz (-50 dBW/Hz) At the receive end the despread process restores the -50 dBW/Hz to -30 dBW/Hz, assuming for argument, no path loss. However, assume that on the path there is a co-frequency narrow band signal interfering also at 10W/10KHz. The dispreading process, which is identical to the spreading process, will spread that 10 watts of interfering signal over 1 MHz, reducing its power spectral density to – 50 dBw/Hz. So at the exit of the despreader, you have a desired signal of -30 dBW/Hz and an interfering signal of -50 dBW/Hz. Thus what, without spreading would have been a zero dB Signal to Noise(QRM) is now a +30 dB S/N. The gotcha for SS in HF is that you need a wide bandwidth to channel bandwidth ratio to get a significant (15-20 dB) gain in interference rejection. Most HF bands do not have the bandwidth available for any but the very low rate user rates, hence the banishment to UHF. The other gotcha is that every signal in the spread bandwidth gets “despread” to contribute to the despread noise floor. Consider the previous example. If, within the spread bandwidth there was another signal, not within the 10 KHz bandwidth of the desired signal, but 20 dB higher in power, the result would be at the despread point, you are now back to 0 dB S/N. Sane result for ten carrier only 10 dB stronger. Lots of strong carriers that you the narrow band carrier are trying to squeeze through sure sounds like a description of a typical HF band. Instead of spread spectrum, what is the much more powerful technique in the real HP world is source code and modulation code so that you transmit your encoded information as energy over a number of frequencies (Frequency Diversity), simultaneously over a number of time slots, both coding with a powerful FEC code, such as rate ½ or 1/3, so that there is a high probability of recovering the information, un corrupted, from a channel with high interference for short periods of time over narrow, and with frequency selective fading (QSB) also in a narrow band instantaneous occurrence. The optimum solution is to use the available energy divided into multiple small carriers on a range of adjacent frequencies or in a single hopping carrier over a similar bandwidth, coded to provide FEC and time diversity so that if a hops information is lost when it gets stepped on by QRM, but recovered when multiple other hops are received and the result decoded. Any data transmission system will increase the occupied bandwidth in some fashion, as a function of the encoding steps used to develop the modulated waveform. This increase in bandwidth is used to improve the end to end performance, i.e., to improve the Eb/No vs BER curve. This FEC/Coding/Modulation choice is optimized by the ingenuity of the designer, but limited by the allowed (regulated) bandwidth for the situation. There is a point of diminishing returns in all of this. You must be able to maintain synchronization of frequency and bit timing from one end to the other end of a link in order to keep the FEC decode process working. This becomes harder and harder to do when the power spectra density becomes close to the power spectral density of the noise (QRN&QRM). Further on multi hop HF links and on moon bounce, you get multipath effects that make both the received frequency and timing change, in some cases changing faster than the transmit symbol rate. This makes recover of the right bit in the right time location a problem. So, if bits are added to the transmit waveform that are not performing a function of helping to re-create an error free replication of the input data, it meets my test as spread spectrum. If the symbols in the transmit waveform cannot be predicted by the previous sequence of bits over time at the input, it also would meet my test as spread spectrum. To reiterate on this point, just because the symbols of the transmit waveform are changing during an unchanging input, does not imply spread spectrum. Instead, they may well be the result of a defined randomizer process followed by multiple layers of FEC and modulation coding. Excuse please my long winded response. If my intelligence has descended into the noise floor at some point or points, please nail me to the wall and I will try to clarify. Lester B Veenstra MØYCM K1YCM <mailto:les...@veenstras.com> les...@veenstras.com <mailto:m0...@veenstras.com> m0...@veenstras.com <mailto:k1...@veenstras.com> k1...@veenstras.com US Postal Address: PSC 45 Box 781 APO AE 09468 USA UK Postal Address: Dawn Cottage Norwood, Harrogate HG3 1SD, UK Telephones: Office: +44-(0)1423-846-385 Home: +44-(0)1943-880-963 Guam Cell: +1-671-788-5654 UK Cell: +44-(0)7716-298-224 US Cell: +1-240-425-7335 Jamaica: +1-876-352-7504 This e-mail and any documents attached hereto contain confidential or privileged information. The information is intended to be for use only by the individual or entity to whom they are addressed. If you are not the intended recipient or the person responsible for delivering the e-mail to the intended recipient, be aware that any disclosure, copying, distribution or use of the contents of this e-mail or any documents attached hereto is prohibited. From: digitalradio@yahoogroups.com [mailto:digitalra...@yahoogroups.com] On Behalf Of rein...@ix.netcom.com Sent: Monday, July 12, 2010 8:33 PM To: digitalradio@yahoogroups.com Subject: Re: [digitalradio] Random data vs Spread Spectrum Hi W2XJ, Could you tell me please ( I am believe to be the only person in the group of 4000 seriously interested in this subject as a potential user ) the exact definition of SS in this connection.