Relative modes of phase manipulation.
An occasion to writing of this article became discussion at our forum, from which follows, that there is a certain vagueness in definitions of relative modes of phase-manipulation.
Let's try to understand more about this subject within the limits of the review. The question is not so difficult, although it demands only attention and patience, besides some historical moments will be interesting.
We will start from definition of two very different concepts, because a rule, and there are various misunderstandings in definition of relative modes.
There is a relative coding, and relative manipulation. These are absolutely different two mechanisms, which sometimes can be confused.
Relative coding has the very interesting story. For the first time it was offered and tested/used in practice, by domestic (Russian) researches. It was very difficult and hard to extort general recognition of the relative coding.
Only now, it all seems natural, logical and clear, while in past relative coding faced a lot of contractions against it.
Only an absolute coding, had been used before relative coding. In absolute coding this or that concrete symbol/character is corresponded to an absolute phase value. For PSK-2 it is 0 or 1. For PSK-4 it is the character, which consist of two bits, di-bits 00, 01, 10 and 11, and triple-bits for PSK-8 etc.
From the present-day positions the narrow moments are well visible at once. It is enough to "lose" a correct binding to an initial phase in the demodulator, and we can forget about any correct demodulation of speech.
In past days PSK-2 has been very widespread, and there was even a term "returned operation", as in this case bits 1 and 0 changed places. Special test sequences were inserted into messages to remove "returned operation" phenomenon, the long procedure of synchronization and phase binding had been preceded the main operation.
In PSK-2 it was possible to remove "returned operation" by simple inversion of the accepted sequence, while in PSK-4 or PSK-8 it was already not simple. The one of reasons is the very difficult recognition of phase manipulation in general. For very long time phase manipulation had been considered as an absolutely unpromising and useless tool for an information transfer.
A bolt from the blue(~in the middle 1950 years), had been sounded the sentence of domestic scientist Petrovich N. T. He offered to encode (at phase manipulation) the character/symbol not as an absolute value, but as the difference between values of adjacent characters.
Solution was so genius, graceful and unexpected that many people simply denied it. No one wanted to hear any mathematical substantiations (!). The real fight for recognition of relative coding as method basically was developed. In present days it looks ridiculous and strange, while in those times the author of the method had been under greatest pressure of the negative public opinion of majority people. No one wanted to listen him, motivating it is a time waste.
To honor of the author, despite the most powerful, almost complete counteraction to the method, He could realize comparative trials and tests almost only on his own resources, where it was brilliantly represented that the relative coding not only removes strongest "headache" in the form of the problem of absolute phasing in the demodulator, but also opens huge perspectives for phase manipulation/modulation as a whole.
So thus, well-known today to anyone method of the relative coding, in some years (the end of 50th years) after absolutely not deserved persecutions, has received a residence permit in a life. Here is such small historical essay. :-)
Coming back to our questions, please pay attention, it is important, it is a question only of relative coding!
The order of phase manipulation remains former! There is a replacement of a bit stream from absolute values of bits on the relative ones. It allows as a result, to ignore concrete phase of the current character/symbol, and to be oriented on the difference of phases of adjacent characters/symbols. As a result, it makes the demodulator invariant to concrete values of phases, and removes the main problem of the absolute coding, the problem of "returned operation".
In this sense relative coding, by the cost of only one-two character/symbol (!) guarantees, as a whole, the faultless demodulation at any initial phase angles in the demodulator. Besides, relative coding uses all possibilities of the current dimension of phase manipulation completely.
After a residence permit and recognition of the relative coding, the very intensive researches on phase manipulation/modulation had been started, and it is possible to consider that Petrovich N. T's well-deserved reward - was that phase manipulation/modulation received the status of the extremely perspective mode for an information transfer.
Both recent skeptics, and already friends, have rushed to searches of even more interesting ideas and solutions in this direction.
Quickly enough there were offered the methods, of phase-different manipulations, not codings! But it was already inevitable, and expected.
Differences between relative coding and relative manipulation (in common enough case):
1) in the relative coding, the bit stream varies, while there is no the law/order of change of a phase in the modulator
2) in the relative manipulation, the bit stream remains former, but the law of change of the phase in the modulator varies.
It may seem that both cases will turn out into same result. But no, these are absolutely different things as a result. Though, in both cases a binding in the demodulator is realized to not to concrete values of sendings/impulses, but to their changes. It is an important point, because usually here hides the problem of all confuses.
In relative manipulation, there is always a change of a position of a point of constellation. It solves two problems:
1) absolutely guaranteed selection of clock frequency of manipulation (as the point is never stay in one place)
2) and because the character/symbol is encoded by an angle of transfer from one point into another, the same invariance to an initial phase is reached, as well as in case of relative coding.
Distinctions of these two different approaches to relativity are well visible on constellations.
Absolute phase manipulation and-or relative coding.
This mode of phase manipulation has at least three names.
PSK-4 Or QPSK - that means all same - an absolute phase manipulation.
DQPSK - usually so designated the mode, with usage of relative coding, but also with an absolute phase manipulation.
As we see there is no ways to tell precisely, by the form constellations, is relative coding used or not.
It is not surprising, because the relative coding mentions a bit stream and does not change angles of the constellation, constellation reaches relativity from the stream, and does not demand any special measures and solutions.
Relative phase manipulation (RPM), di-bits are shown conditionally.
This mode is named pi/4 DQPSK, and concerns to phase-different manipulation. The relativity is already demand efforts, the modulator should select an angle, on which the current point of constellation is turned by di-bit. It is well visible that in an absolute mode the constellation looks like PSK-8, and in relative as PSK-4, turned on the angle in 45 degrees (a prefix pi/4 in the name).
The phase-different manipulation mode can be gotten, by having on hands both absolute and relative constellations, and transitions/transfers in them.
These are all main nuances of relative modes of phase manipulation, in more details these questions are well shined in the literature. Of course relative coding can be transferred on the modulator, carrying out code conversion of the bit stream in it, through angles at once. That is frequently done, but it does not change general state of things, because it is still the relative coding, which is simply realized by modern method.
It would be desirable to mark that PRM possesses several more pluses.
As we can see on the absolute constellation, the trajectory of movement of the point does not intersect zero area (transitions on 180 are not used), but avoids it, it sharply reduces requirements to linearity of a transferring path, and reduces amplitude downfalls of the resulting signal after filters. Although, the price for all that ( including good selection of clock rate) is decreasing of transition/transferring speed.
In this example, the speed is equivalent to the resulting PSK-4 mode, though, the false impression of more high-speed PSK-8 mode can appear, according to the form of absolute constellation.
Добавлять комментарии могут только зарегистрированные, активировавшие регистрацию и не ограниченные в доступе участники сайта!
|Файл создан: 10 Jun 2010 16:00, посл. исправление: 10 Jun 2010 17:28|