Talk:Symbol rate

Discussion of Suggested Merges
This article should not be merged with Symbol_(data) until the article bit rate is merged with the article bit.--GrahamDavies (talk) 21:50, 5 May 2008 (UTC)

This article should not be merged with baud until the article length is merged with metre.--GrahamDavies (talk) 21:50, 5 May 2008 (UTC)


 * Actually bit/s and bps are forwarded to the bitrate article. Mange01 (talk) 08:34, 6 May 2008 (UTC)
 * Actually bit/s and bps are forwarded to the bitrate article. Mange01 (talk) 08:34, 6 May 2008 (UTC)

I re-added the merge prop tag, as it seems fairly straightforward to me that this is the most basic example of Wikipedia's merging policy. Wikipedia's policy on article mergers specifically states that Wikipedia is NOT a dictionary and should not be treated as such. Graham's suggestion that length and metre should then be merged is a nonsequitur, because there are many distinct units of length measurement in addition to metre (and even then a case could still be made for a merge), whereas the only unit for measuring symbol rate is the Baud.

Furthermore, look at the two articles, and tell me if you see any content contained in Baud that is not contained in this article. It's all just duplicate content! There's absolutely no need whatsoever for there to be a separate article for baud. None. The baud article is virtually devoid of content, and about 90% of the content it does contain is already contained in this article. 67.168.49.198 (talk) 22:25, 1 December 2008 (UTC)

I agree they should be merged 220.233.24.123 (talk) 11:40, 16 August 2010 (UTC)

History
The "Baud rate" article was renamed to "Baud" in 24 January 2007, but "Baud rate" is still forwarded to "Baud". Merge suggestion templates were placed in May/June 2007 on the Modulation rate, Symbol (data) and Baud articles. Merge suggestion templates were placed in December 2007 on the Significant condition and Symbol (data) articles.

Yesterday I merged the "modulation rate" and "Symbol rate" articles into the "baud rate/Baud" article, since there has not been any protests to the merge suggestion templates. However, the forwarding of the Symbol rate article was reverted. To resolve the dispute, I suppose we need to vote. Mange01 (talk) 08:34, 6 May 2008 (UTC)

Survey 1: Merge with Baud? Article name?
Should Symbol rate be merged with Baud, which already is merged with Baud rate and Modulation rate? And what should the name be of the article?
 * Strongly support. The main article title may either be Symbol rate or Baud. Argument: These rates are synonyms. Mange01 (talk) 08:34, 6 May 2008 (UTC)


 * Strongly object. First, I also undid the merge of Modulation rate. This now re-directs to this article.  The reason is that, as originally described, modulation rate is synonymous with symbol rate in the special circumstance of a modulated channel.  I hope this is OK.  Second, I don't see the sense in merging an article about a physical property with the article about its unit of measure.  To cite another parallel, this would be like merging the temperature article with the kelvin article.  Third, the term baud is subject to technically dubious usage.  In fact, it is more frequency used incorrectly, as "baud rate", than correctly.  But, correct me if I'm wrong, Wikipedia is, like an English language dictionary, descriptive, rather than prescriptive.  It is not our place to exclude a common usage of a term just because we think it is wrong.  So, the article on baud should address both usages, one of which is closely associated with this article and the other of which isn't.  Therefore, baud and symbol rate should be separate articles.


 * Strongly object. Baud is still--even if I'm being oxymoronic--a historically current term. In other words, many living people have experience of baud, even in its less correct meanings. Baud needs to have an article of its own. If, at some time in the future, baud becomes a totally obsolescent term which no-one remembers, then it's time to have it refer to a more comprehensive article. Until "symbol rate" is an industry standard term and "bits per second" (bps) or "bytes per second" (Bps) is not, the Baud article is needed. I understand the logic behind symbols per second, but until bps and Bps are displaced and I see sps (Symbols Per Second) quoted in industry specification sheets, Strongly Object.
 * Agree with the tone of the previous comment. ("First, I also undid the merge of Laguna CA (talk) 05:26, 20 April 2011 (UTC)

Survey 2: Merge with Symbol?
Should Symbol rate be merged with Symbol (data)?
 * Support. The articles discuss the same issues, and one of them is reduntant. I can not think of any issue that should be discussed in one of the articles but not the other. Mange01 (talk) 08:34, 6 May 2008 (UTC)


 * Support for the same reasons as Mange01.--GrahamDavies (talk) 13:48, 6 May 2008 (UTC)

Discussion continued
If the answer is "no" on survey 1, I suppose Baud rate should point at Symbol rate instead of the Baud article for the same reason?

For now the "Symbol rate" article is not very well written, and I don't understand what difference you had in mind. I have already merged text from "Symbol rate" and "Modulation rate" into the "Baud rate" article. Can we copy all the text from the current Baud article to the Symbol rate article? After that, perhaps some text may be removed from the Baud unit article? At least temporarily while waiting for the survey result, in view to clarify how two separate articles would look like? As a third step, I suggest that the Symbol (data) text is merged into the Symbol rate article.

I agree that the term Baud rate is problematic. Not so much because it has been confused with bit rate, it was long time since I saw that. I teach using a course literature that claims that 1 baud = 1 symbol, which is an understandable mistake, since "rate" in "baud rate" can be understood as "per time unit". But, as you pointed out, we have to mention baud rate in the articles since it is a widespread term.

There are some units that are forwarded to the measure article. Example: (bit/s)/Hz -> Spectral efficiency. Is it important to always create separate unit articles? Why?

(You and I Graham had a discussion a few days ago, but then you gave me the impression that you supported all the suggested merges. Sorry for misunderstanding you.)

Mange01 (talk) 22:09, 6 May 2008 (UTC)

Reply to the above discussion
I'm sorry that we appear to have different approaches to improving this area of Wikipedia, but it's good that we both want to do it. I am not, like you, a "wikiholic", so it would be unfair of me to stand in your way if you want to push changes through. I will reposition myself as a helper. I have a long history of work with both passband and baseband communication systems, I have designed symbol constellations, channel equalizers, etc., so I think I understand many of the issues at a deep level. I will make a couple of points and try to explain my position fully and then I will let you take the lead. I don't know how long we should wait to see if anyone else contributes to this discussion. I have the impression I showed up at the eleventh hour myself.

The parallel of (bit/s)/Hz redirecting to Spectral efficiency is not appropriate here because (bit/s)/Hz is a combination of fundamental quantities and units of measure. By contrast, baud is a single word assigned as the unit of measure of symbol rate to honor a major contributor to the field. It deserves its own page to explain the derivation from Emile Baudot's name and to link to that article. Please also consider that the (mis)use of the term needs to be explained and this should not be done in an article about symbols.

If we agree that articles are not well written, we should consider a gradual re-write as an alternative to copying existing text from one place to another. This would be facilitated by having in mind where we want this set of articles to end up. I will now explain where I think that should be and leave the final structure to you.

The Symbol (data) article is the most fundamental, should be the longest and most thorough. It should make free, but only correct, use of the other terms.

The Symbol rate article should be very brief and point the reader to Symbol (data). However, there should be hooks here for related articles. Modulation rate should redirect to this article and the equivalence of these terms should be stated clearly in the Symbol rate article. It should also be stated that baud is the unit of measure of symbol rate, thus referencing that article. The non-equivalence of baud rate should be noted, thus referencing that article as well. Whether someone is looking for the correct or incorrect use of terms, when they reach this article they should know where to go next.

The baud article should also be very brief and point the reader to symbol rate. This is where Emile Baudot gets a mention. Again, it should be pointed out that baud rate is used in contexts unrelated to symbols, allowing a reader to click over to that article.

The baud rate article needs sections on each possible misuse of the term. As I see it, these are 1) the same as baud, with "rate" erroneously appended, 2) the use where bit rate would be the correct term and 3) the use to mean gross bit rate, as distinct from payload bit rate, where communication involves a protocol overhead.

That makes one large, one medium/short and two very brief articles with modulation rate redirected as at present.

--GrahamDavies (talk) 14:25, 7 May 2008 (UTC)


 * Discussion: I theoretically agree with the idea, but "symbol" does not have a definite number of bits. As such, "Symbol rate" is meaningless. Example: If the a symbol rate of 2 with the symbols "0" and "1" is equivalent to a symbol rate of 1 with the symbols "0", "1", "2" and "3" is equivalent to a symbol rate of 1 with the symbols "00", "01", "10" and "11"; how does symbol rate have anything other than a wishy-washy, theoretical meaning? I totally agree that "symbol rate" has a place in theoretic discussions of communications, but until some hard definition is placed on "symbol", symbol can be redefined to be any bit rate one desires: as such, "Symbol Rate", while definitely a useful term for theoretical communications discussions, has no concrete meaning. Thus, "symbol" and especially "symbol rate" (which is doubly vague) should be treated as the theoretical construct they are. Even if X=8 commonly, X and 8 are vastly different terms. Laguna CA (talk) 05:40, 20 April 2011 (UTC)


 * Symbol rate is a well-defined concept used in data communication literature, and that also is of practical use in modem specifications. See my answer below. I think the article needs illustrations that explain things better, for example a time diagram showing the signal from an 8QAM modulator or a 4PSK modulator. An analogy: You could say that modems play music to each other - they represent bits by tones. The symbol rate is the number of tones that are played per second. An 8 QAM modem is like a music instrument with 8 keys, each generating a tone with different amplitude and phase. Each key represents 3 bits. So, if the musician (the modem) plays 100 tones per second (or symbols per second) the data rate is 300 bit/s. Mange01 (talk) 06:52, 20 April 2011 (UTC)

another stone in the choice of the links with bit rate, symbol rate or Baud rate
In fact use of Baud (instead of Baud rate) is of overwhelming importance instead of symbol rate because when considering transmission as a chain of processing, we can distinguish : * the source rate (either bit if binary or symbols if non binary) * the rate at the output of the data compressor (if used) which can be also either a compressed bit rate or a compressed symbol rate * the rate at the output of the(single or if concatenation plural) channel encoder(s) can be aither (whether the code is binary or not) a channel coded bit rate or a channel coded symbol rate * the rate at the output of the constellation mapper which you called Symbol there taht depends on the order of the chosen modulation (binary, quaternary, 8-ary,...). This is in fact a constellation symbol rate.

Due to the different processings, there is in fact a link between the source bit rate and this constellation symbol rate but you can see that it is a combination of different ratio that gives the final one.

To be honest, a french man Emile Baudot has already made this trip and understand it was quite ambigouus to only use the term symbol rate when only telegraph signalling was used, let's use the unity Baud to define a speed of variation of constellation symbols inside the channel for our far more complex transmission chains. We try to use it in our telecommunications projects at CNES (French Space Agency) and in European and world cooperation programs so as to avoid all those ambiguïties coming from the use of the imprecise word symbol rate...

And remember that when you have chosen a waveform shaping associated to your modulation, you are the able to derive directly from the Baud your spectral occupancy which is of course of main importance. Going back to your source bit rate you can then derive your real spectral efficiency in b/s/Hz, the ratio in Baud/Hz (bandwidth in Hz defined at 99% power for example) giving you the shaping penalty. Too often you can get some comments talking about spectral efficiency at Nyquist in b/s/Hz, use of Baud can there also avoid this quite curious and theoretic definition by talking about a new "rate efficiency" in b/s/Baud qualifying then all the processes (from source coding to modulation) before the shaping choice.

So let's emphasize on the use of Baud for the constellation symbol rate instead of talking only about Symbol rate

GuyLesthievent (talk) 16:55, 8 August 2008 (UTC)GuyLesthievent (engineering and teaching in Telecommunications since 25 years)


 * I did not understand. Please give an example when Symbol rate may be interpreted differently from Baud rate. The article should i.m.o. present both terms, and as synonyms. Okay, "Baud rate" gives more hits in Google and "Symbol rate", but Baud rate is often confused with bit rate.
 * I suppose what you call b/s/Baud, I suppose is often called bit/symbol. Mange01 (talk) 11:15, 11 August 2008 (UTC)

Merge of Symbol rate with Symbol (Data)
Symbol rate was recently merged into Symbol (data). However, the merge resulted in overlapping text and in several double redirects, and in the above discussion it was suggested to keep them separate. See Special:WhatLinksHere/Symbol_(data).

(I have already merged Symbol rate with Modulation rate and Baud rate, and moved some of the material from Baud to Baud rate, according to the discussion above. Baud is still a separate article).

The old symbol rate text was placed in the end of the merged article, and the article was named Symbol (Data). However, the symbol rate text was i.m.o. more wellwritten and easy to understand. Symbol rate is more important than Symbol (Data), since many more articles are linking to Symbol rate. See Special:WhatLinksHere/Symbol_(data). I have therefore moved the symbol rate text to the beginning of the merged article.

Question: Should the merge of the articles be kept? And what should it be named? Symbol rate, Baud rate or Symbol (Data)? Mange01 (talk) 10:53, 11 August 2008 (UTC)


 * Since I got no response on the above questions, I renamed the merged "Symbol (Data)" article myself to "Symbol rate" according to my own suggestion.
 * Note that still there is overlap between the merged texts. In case no one fixes the overlap within reasonably long time, I suggest that the merge is reverted and the article is split into two articles. Mange01 (talk) 20:54, 14 August 2008 (UTC)

Baud, bits and symbol
In this article and in the article for "baud", the term "symbol" may be a better term to use the the term "bit" only if the term "symbol" has more general usage that its use in telecom. IEEE usage, for example, interchangeably refers to "Baud" and to "bit rate". If "Baud" is used outside of telecom (outside of electronics, actually) then "symbol rate" might be a better definition than "bit rate". However, it is possible too that use of the term "symbol" is due to an error in translation.

As examples, consider that "symbol" can appropriately refer to certain ASCII characters. The binary form of these symbols requires 7 bits or more (depending on the character set). The metric (aka the parameter) used to compare different devices and connections that represents the speed at which data travels is "bit rate" or "baud", not "symbol rate". Because "baud" refers to rate by definition it is not expressed as "baud rate".

Incorrect usage was not unusual in the previous century. This century it almost never heard because the rate is typically in units of millions (Mega-) or billions (Giga-).

Kernel.package (talk) 01:46, 14 July 2010 (UTC)


 * I changed the article now to further clarify that in this context a symbol is a waveform, e.g. a pulse or a modulated sinewave. It has nothing to do with characters. In today's modems and Ethernet transmision, the bit rates are much higher than the symbol rates. For example a telephone modem may have a symbol rate of 2400 baud upstreams and 8000 baud downstreams, while the bit rate is 48000 and 56000 bit/s respectively. So it is important to not confuse bit rate with baud. In old modems there were only two symbols, one representing 0 and one representing 1, meaning that the symbol rate and gross bit rate were the same. Also on most serial interfaces (EIA232 and similar), there are only two voltage levels of the pulses, meaning that gross bit rate and baud is the same in that context. But in all modern modems (faster than 2400 bit/s) and high-speed Ethernet (100 Mbps over copper or faster), there are more than two alternative symbols, meaning that the Ethernet pulses have more than two levels, or the modem sinewaves have more than two phases and amplitudes. This implies that each symbol represents more than one bit. IEEE people would never confuse the concepts. The bit rate article also expains this. Maybe someone can provide an illustration that further clarifies this common misunderstanding. Mange01 (talk) 08:32, 14 July 2010 (UTC)