Frequency Domain Processing
As an example, we will take a close look at the content of an SSB signal, as shown in Figure 1. The bandwidth needed for a voice channel is around 3 kHz. During every fraction of a second however, only a part of that channel is used. There are always only a limited number of frequency components present.
Figure 1 — Limited number of frequency components per unit of time of a single source.
Depending on the speech content in the signal, the channel will contain different frequency components every following moment. Every interfering source, whether local or not, will also contain its own frequency components at any moment. The chance that frequency components of different sources will overlap one another is very small. See Figure 2.
Figure 2 — Non-overlapping frequency components from two sources.
If we are able to select only the desired frequency components, this is comparable with a reduction in bandwidth. Those parts of the spectrum in which we are not interested are cut off, just as with a CW filter, but within the channel. This assumes that each frequency component always comes from just one source. This is of course, not entirely true, as there is noise from the antenna preamplifiers and atmospheric noise over the whole band in variable amplitude. There is always a small chance that more than one source will contribute to a frequency component. We will deal with this later in more detail.
Distinguishing Frequency Components Using TDOA
How can we distinguish the desired frequency components from the interfering frequency components for the purpose of selection in the frequency domain? How do the various signal sources differ?
We can look for a difference in the typical characteristics per type of signal source. CW,
SSB and ambient noise have different characteristics. The assumptions one must make for this do not always hold true. An interfering source can, for example, have the characteristics of speech or can even be speech.
A practical and more useable difference is seen in the direction (azimuth and elevation) from which the signals arrive. Direction translates, when using two or more antennas, into time difference of arrival (TDOA) or phase differences and in ratios of amplitudes. We see the same effective difference whether or not signals are circularly polarized, and the sense of this circular polarization. We will also see this as a phase difference later on.
We can now choose what we are going to do with each frequency component based on phase difference and/or amplitude ratio. We limit ourselves to discrimination based on phase difference alone. There are two reasons for this. First, because the phase difference mainly provides the information on the direction, and second, because the processing is carried out at the audio, and not at the intermediate frequency level. As a consequence of the AGC, it is not possible to determine the real amplitude ratio. It is mainly due to the AGC that this sort of processing must actually be carried out at the intermediate frequency level.
Last update: September 24, 2006