226:
propagate through the channel (by means of the usual linear convolution style), and the received sequence (at the receiver side) will contain (within a section of the total received sequence) a pattern that matches the 'circular convolution' between the raw data set (without the CP added) and the channel impulse response. And the length of the circular convolution (eg Mod N circular convolution) is assumed to be the length of the raw data set. And, importantly - if any old arbitrary sequence were attached to the front end of the original raw data set (other than a CP of suitable length), then the result of the linear convolution will not contain (within a portion of the total received sequence) a pattern that matches the circular convolution result. If only sources attempt to explain things properly - rather than leave people scratching their heads - then it would be great. But that's the life story of so many documents ----- it's useless if things aren't explained properly.
500:'a b c d e' with channel impulse response h = , and leads people to believe that the received signal will just be linear convolution between x and h. But there's a catch. The values 'a', 'b', 'c' etc each represent a complex vector symbol, and the duration of each of those vector symbols is typically much longer than the impulse response 'h'. So things might make more sense in the discrete-time domain (when generating simulated received sequences) by choosing a sampling period that is much smaller than the rate that which 'a', 'b', 'c' etc get clocked out. That is, the channel impulse response 'h' is generally meant to be much shorter than the duration of 'a' or 'b' or 'c' etc. So everything becomes clear if we break down the sequence 'x' into finer resolution samples prior to carrying out linear convolution.
479:
information is cryptic. Sources that properly show a step-by-step example with truly useful guiding side-information appear to be rare, or don't exist. The first issue is, material about cyclic prefix is introduced, but nobody teaches anything like "ok....here is a complex-valued sequence with a cyclic pre-fix, and we now use the real part and imaginary part of each element of this set of sequential elements to modulate cos and sin carriers, for mixing up to a frequency transmittable by an antenna. Then, at the receiver....we need do this...followed by this....and this...". It's the same story with many systems. It's not that students or exponents aren't capable of understanding it all. The problem is that there are no sources that tell it like it really is.
324:
S was to some reasonable precision? If the receiver does know this delay, how does it know it? Does the transmitter occasionally send a pilot at the same sub-carrier frequency as S and the receiver use this pilot to estimate the delay? If so, what if that delay changes between when the last pilot is received and the next symbol is received. To me it all comes down to knowing what that delay is, otherwise how could the receiver know (if S were a QPSK symbol for example) if the symbol val is 00, 01, 10, or 11 since knowing that is all based on phase relative to the sub-carrier for S - phase which is offset by said time delay?
71:
53:
368:
window to match. This alignment will typically serve well enough for the next few data symbols, until the next pilot is transmitted. In a highly mobile channel (where the delay and multipath vary quickly), more complicated approaches are required, such as interpolating between successive delay estimates, predictive delay tracking, or data-aided estimation (whereby the receiver effectively compares what it received against a resynthesised waveform based on what it's managed to demodulate).
407:
reply would suggest that the pilots would be used to do the time synchronization inside that black box whereas my information only says, "After the initial A-to-D conversion, the receiver is synchronized to the signal" - making no mention of pilots. perhaps my information is simply incomplete and pilots are used for time as well as frequency synchronization. If that is so, for me that opens up some more questions:
377:, there are no pilots. The receiver will typically synchronise by using the fact that the cyclic prefix is a repetition of another part of the symbol, and hence can run a sliding auto-correlator and look for peaks. This approach is less accurate, but typically doesn't matter because the simpler modulation scheme of DAB (DQPSK) doesn't require a high-performance receiver in order to demodulate adequately.
22:
400:" (emphasis mine) and then "re-align its sampling window to match". My information gives no indication that pilots are used for time synchronization - my information only shows a "time synchronization" black box in the receiver following the A-to-D conversion. After time synchronization, the cyclic prefix is removed (makes sense - you wouldn't want to remove the cyclic prefix
81:
568:
lengths? If they're going to say that the channel impulse response has "L" elements, then at least explain it, and comment on how to treat it - such as later assume that "H" is extended to N elements(where element 0 to L-1 are whatever they are supposed to be, and then the remaining elements up to index N-1 are assumed to be all zero.
280:
This symbol extension is specified in the standard as ¼, 1/8, 1/16 or 1/32 of the symbol time to provide flexibility for different environments, but the WiMAX Forum specifies as a mandatory implementation the 1/8 fraction. The maximum multipath spread distances supported for symbol duration of 102.86
567:
Also, why is the channel impulse response represented by "L" points? The article has "X" in the frequency domain with N points (ie. 0 to N-1 index), and "H" with L points. So what's going on here regarding the X(k)H(k) product shown in the article? How can you multiply the two when they're different
406:
time synchronization, right?), then the signal is converted into the frequency domain via an FFT, and then the "pilots are enlisted to maintain frequency synchronization". So I think it comes down to my desire to know with more certainty what constitutes that "time synchronization" black box. Your
323:
I am not sure how the delay spread is determined or if it necessary to know it to any precision. But, if all of my above suppositions are correct, how can the receiver know what to discard as cyclic prefix samples unless said receiver somehow knows what the delay (from transmission to reception) of
268:
The cyclic prefix may be viewed in more general terms. Transmitting a periodic waveform makes linear and circular convolution the same. This leads to frequency-domain processing at the receiver possible with the DFT. This isn't unique to OFDM. Indeed, an active area of research is single carrier
499:
I will also add that the question asked by Billy is an excellent one. Text books and sources don't appear to show what the incoming received sequence (in some form) is meant to look like. OFDM discussions involving cyclic prefix often involve confusing examples, such as x = (with example prefix of
393:
Oli_Filth, thank you for the reply. I think here is the nub of my confusion.... My information on OFDM (text from a professional education course on WiMAX) says the receiver uses pilots for frequency synchronization whereas you have indicated that the receiver will identify "the location of the
478:
That's the main issue. Text book and online sources don't appear to truly 'properly' teach or show clearly what OFDM synchronisation actually involves. There are certainly articles and papers that discuss 'synchronisation'. But one will often find that the diagrams are not clearly explained, and
303:
I am pondering this article from an OFDM perspective. The article states "the receiver is typically configured to discard the cyclic prefix samples". To me this statement begs the question - how does a receiver recognize cyclic prefix samples? Please correct any of the following suppositions:
225:
I personally would avoid writing something like 'makes the linear convolution look like circular convolution'. That is somewhat vague. It would be much more meaningful by saying something like adding a cyclic prefix allows the modified data sequence (that now contains the CP on the front end) to
367:
is the sub-carrier frequency spacing). It is able to do this because the transmitted waveform is typically interspersed with pilot symbols, which the receiver can correlate for and determine symbol timing. Once it has identified the location of the pilots in time, it can re-align its sampling
548:
Could someone add information on Cyclic Prefix/Guard
Interval length to the article? I.e. why should you choose for what value of L. To minimize overhead, L should be as small as possible of course. And larger then N also seems illogical. But what then, and why?
180:
Makes the linear convolution of a frequency-selective channel look like circular convolution, so that sub-carrier orthogonality is maintained (so no inter-carrier interference), allowing frequency-domain channel estimation and equalisation to be
316:
4) Some of what the receiver samples as S will be cyclic prefix (unless the delay from the transmitter to the receiver for S equals the guard interval in which case the receiver misses the cyclic prefix altogether).
320:
5) The length of the delay of S from the transmitter to the receiver we do not know, we only know that it is somewhere within the delay spread (which is hopefully less than the guard interval).
245:
I have tried to add some rudimentary stuff. Please note that it may not be clear, or consistent with the pattern of math symbol use. If so, please correct it or bring it to my notice. Thanks.
597:
127:
133:
269:
with frequency-domain equalization (SC-FDE). I don't have time right now to contribute SC-FDE to
Knowledge, but someone will, I'm sure, in the future. -- sct 2007-05-17
592:
310:
2) A symbol transmitted on an arbitrary sub-channel (call it S) will be delayed by some amount of time within the delay spread before the sampling receiver grabs it.
255:
Added a few more things in a hurry, forgot to even label the changes! Sorry. I think I'll stop now, and let the experts take over. Comments would be appreciated!
103:
359:
delay at the receiver. But that doesn't matter; all the receiver needs to do is ensure that it correctly "opens" its sampling window such that it picks up 1/
284:
In that case, according to my understanding, then the cyclic prefix can be used to determine the distances a signal 'should' be expected to optimally cover.
200:
the guard interval. If the guard interval were longer than the CP, the circular convolution principle would break down, and the CP would be useless.
428:-If pilots are used for time synchronization, how might that look as far as FFT's, filtering, IFFT's, etc. inside the time synchronization black box?
94:
58:
456:
343:
307:
1) The receiver samples for a period equal to the inverse of the symbol rate (which is constant regardless of sub-carrier frequency).
529:
313:
3) The delay spread is arbitrary depending on channel conditions, multipath, distance between transmitter and receiver, etc.
33:
155:
This article is, at best, misleading. All three comments below are right and have been integrated to the main page.
374:
86:
419:-Are pilots included in the transmitter's cyclic prefixing operation? Does it matter whether they are or not?
21:
452:
339:
70:
52:
533:
39:
448:
335:
554:
550:
525:
444:
331:
289:
569:
501:
480:
227:
211:
156:
102:
on
Knowledge. If you would like to participate, please visit the project page, where you can join
256:
246:
277:
Can someone clarify if the cyclic-prefix is equivalent or is the same as the symbol extension.
573:
505:
484:
231:
215:
160:
99:
177:
As a guard interval, it eliminates inter-symbol interference from the previous OFDM symbol.
285:
378:
201:
327:
Anyone who might be able to better explain this, please respond. Thanks in advance.
586:
76:
171:"to allow multipath to settle before the main data arrives at the receiver"
210:
All three comments above are right and being integrated to the main page.
519:
After convolution the correct range for m is one period of length N and
577:
558:
537:
509:
488:
460:
381:
347:
293:
259:
249:
235:
219:
204:
164:
15:
440:
Any further enlightenment would be greatly appreciated. :o)
299:
How does a receiver recognize cyclic prefix samples?
98:, a collaborative effort to improve the coverage of
132:This article has not yet received a rating on the
355:You are correct, in general, we don't know the
598:Unknown-importance Telecommunications articles
8:
19:
523:
47:
112:Knowledge:WikiProject Telecommunications
593:Start-Class Telecommunications articles
115:Template:WikiProject Telecommunications
49:
265:Please let me add to your comments...
7:
92:This article is within the scope of
522:L-1 <= m <= N - 1 + (L-1)
194:"often equal to the guard interval"
38:It is of interest to the following
14:
281:μs are shown in the table below.
79:
69:
51:
20:
241:Some math to clarify the CP use
169:The cyclic prefix (CP) is not
95:WikiProject Telecommunications
1:
544:Missing some on CP/GI length
510:00:41, 14 December 2018 (UTC)
461:23:26, 13 December 2010 (UTC)
382:21:01, 10 December 2010 (UTC)
348:02:19, 10 December 2010 (UTC)
236:00:29, 14 December 2018 (UTC)
205:00:41, 1 September 2006 (UTC)
106:and see a list of open tasks.
192:The length of the CP is not
189:orthogonal to one another.
173:. It serves two purposes:
118:Telecommunications articles
614:
538:17:49, 7 August 2015 (UTC)
294:17:10, 10 March 2008 (UTC)
260:15:10, 11 March 2007 (UTC)
250:14:45, 11 March 2007 (UTC)
220:16:14, 22 March 2010 (UTC)
165:16:41, 22 March 2010 (UTC)
134:project's importance scale
578:19:53, 2 March 2018 (UTC)
489:20:16, 2 March 2018 (UTC)
373:In some systems, such as
363:worth of waveform (where
131:
64:
46:
559:10:42, 8 June 2016 (UTC)
87:Telecommunication portal
28:This article is rated
151:Technical inaccuracy
185:Frequencies do not
109:Telecommunications
100:Telecommunications
59:Telecommunications
34:content assessment
540:
528:comment added by
464:
447:comment added by
351:
334:comment added by
148:
147:
144:
143:
140:
139:
605:
463:
441:
350:
328:
120:
119:
116:
113:
110:
89:
84:
83:
82:
73:
66:
65:
55:
48:
31:
25:
24:
16:
613:
612:
608:
607:
606:
604:
603:
602:
583:
582:
546:
541:
517:
442:
329:
301:
275:
243:
153:
117:
114:
111:
108:
107:
85:
80:
78:
32:on Knowledge's
29:
12:
11:
5:
611:
609:
601:
600:
595:
585:
584:
581:
580:
563:
545:
542:
521:
516:
513:
498:
496:
495:
494:
493:
492:
491:
470:
468:
467:
466:
465:
434:
432:
431:
430:
429:
423:
422:
421:
420:
413:
411:
410:
409:
408:
387:
385:
384:
370:
369:
300:
297:
274:
273:clarification!
271:
263:
262:
242:
239:
224:
209:
183:
182:
178:
152:
149:
146:
145:
142:
141:
138:
137:
130:
124:
123:
121:
104:the discussion
91:
90:
74:
62:
61:
56:
44:
43:
37:
26:
13:
10:
9:
6:
4:
3:
2:
610:
599:
596:
594:
591:
590:
588:
579:
575:
571:
566:
565:
564:
561:
560:
556:
552:
543:
539:
535:
531:
527:
520:
514:
512:
511:
507:
503:
490:
486:
482:
477:
476:
475:
474:
473:
472:
471:
462:
458:
454:
450:
446:
439:
438:
437:
436:
435:
427:
426:
425:
424:
418:
417:
416:
415:
414:
405:
404:
399:
398:
392:
391:
390:
389:
388:
383:
380:
376:
372:
371:
366:
362:
358:
354:
353:
352:
349:
345:
341:
337:
333:
325:
321:
318:
314:
311:
308:
305:
298:
296:
295:
291:
287:
282:
278:
272:
270:
266:
261:
258:
257:Kumar Appaiah
254:
253:
252:
251:
248:
247:Kumar Appaiah
240:
238:
237:
233:
229:
222:
221:
217:
213:
207:
206:
203:
199:
195:
190:
188:
179:
176:
175:
174:
172:
167:
166:
162:
158:
150:
135:
129:
126:
125:
122:
105:
101:
97:
96:
88:
77:
75:
72:
68:
67:
63:
60:
57:
54:
50:
45:
41:
35:
27:
23:
18:
17:
562:
547:
524:— Preceding
518:
497:
469:
449:BillyBarty68
443:— Preceding
433:
412:
402:
401:
396:
395:
386:
364:
360:
356:
336:BillyBarty68
330:— Preceding
326:
322:
319:
315:
312:
309:
306:
302:
283:
279:
276:
267:
264:
244:
223:
208:
197:
193:
191:
186:
184:
170:
168:
154:
93:
40:WikiProjects
530:12.5.37.115
30:Start-class
587:Categories
551:JHBonarius
394:pilots in
286:Kendirangu
181:performed.
379:Oli Filth
202:Oli Filth
187:"become"'
526:unsigned
457:contribs
445:unsigned
357:absolute
344:contribs
332:unsigned
570:KorgBoy
502:KorgBoy
481:KorgBoy
228:KorgBoy
212:Mel aad
157:Mel aad
515:Error?
403:before
36:scale.
196:, it
574:talk
555:talk
534:talk
506:talk
485:talk
453:talk
397:time
340:talk
290:talk
232:talk
216:talk
161:talk
375:DAB
128:???
589::
576:)
557:)
536:)
508:)
487:)
459:)
455:•
346:)
342:•
292:)
234:)
218:)
198:is
163:)
572:(
553:(
532:(
504:(
483:(
451:(
365:f
361:f
338:(
288:(
230:(
214:(
159:(
136:.
42::
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.
