Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
We use rectangular windows in OFDM ( a trivial case when no matched filters are used), this results undesired magnitude responses for the transmitted. However, Filtered OFDM systems and FBMC (Filter Bank Multicarrier) systems use transmit and receive matched filters for OFDM transmission.
We use rectangular windows in OFDM ( a trivial case when no matched filters are used), this results undesired magnitude responses for the transmitted. However, Filtered OFDM systems and FBMC (Filter Bank Multicarrier) systems use transmit and receive matched filters for OFDM transmission.
Thanks for replying. I am trying to draw the block diagram of a typical OFDM system, and I am wondering if these filters are there or not. I need to compare this with a typical single carrier system where these filters are appeared. So, you are saying that they are used in OFDM system the same way as in single carrier systems, right?
Thanks for replying. I am trying to draw the block diagram of a typical OFDM system, and I am wondering if these filters are there or not. I need to compare this with a typical single carrier system where these filters are appeared. So, you are saying that they are used in OFDM system the same way as in single carrier systems, right?
But in this case I need a bank of filters in both sides?! Can not I just put a transmit filter at the end of the transmitter side, and a receive filter at the beginning of the receiver part?
No because every subcarrier will be raised on a different carrier frequency, so the filtering (bandpass) must be raised on the same carrier for every subcarrier and a single filter won't work.
No because every subcarrier will be raised on a different carrier frequency, so the filtering (bandpass) must be raised on the same carrier for every subcarrier and a single filter won't work.
you can use the same Lpf filter for each modulation symbol before multiplying by the carrier freq of a subcarrier or multiply by carrier first and then filter by a BPF centered around the subcarrier central freq. I mean that you don't need to use a filter bank if you filter before serial to parallel conversion and u can only use one filter after the output of the QAM/QPSK/...etc mapper.
I can understand that the transmit filter may be placed before S/P converter, but I could not for the receive filter. I mean, in regular systems, we first filter and then sample, right?
Another question, regarding CP, do we add it after the IDFT at the transmitter, and we remove it before the DFT at the receiver? I mean, it is not entered in IDFT/DFT process, right? I see some papers on single-carrier frequency domain equalization, which is basically works in the same manner as OFDM, i.e.: block-base with DFT/IDFT units at the receiver to equalize the channel at the frequency domain and transform it back to the time domain, and they add at the end of the block a pseudo-noise (PN) sequence to force the linear convolution to be circular to apply the DFT, however, they do the DFT/IDFT for the whole block (including the PN) and then after the IDFT, the last L samples are dropped. Why not drop them before the DFT instead after the IDFT?
- You cannot filter after S/P if you are using IFFT block, but you can do that if you are multiplying by a subcarrier for each parallel stream. In this case, you have to use a different filter in each bank
- In conventional OFDM, CP is added after IFFT and removed before FFT. Adding PN sequence is not a CP, but I don't know how it could be handled.
Ok, now if the channel has memory of L symbols, the first data symbols will be affected by the CP, then why do we add CP and not putting simply zeros to keep the data interference-free from CP?
- CP and Zero Guard interval can both cancel ISI between OFDM symbols, so why do we not simply use a zero guard ?
-> Because if we used a zero guard, there will not be continuity between CP and useful data (one has carrier and other has zero amplitude), so ICI occurs because delayed versions of the OFDM symbol will not be orthogonal to the basic version.
-> Your argument that zero guard will give an interference-free from CP is moot because the delayed versions from OFDM symbols itself interfere with each other (this is fading and not ISI). SO using a zeros' GI is nonsense.
But we are concerned with the mutual orthogonality within an OFDM symbol, and not between OFDM symbols. Yes, the OFDM symbols' useful data will interfere with each other, and this is ISI definitely not fading, but this is unavoidable, while the CP is under designer's control.
- Yes we are concerned with mutual orthogonality but also the delayed version of subcarrier 1 will interfere with the original version of subcarrier 2 if it has no CP and has a zeros interval
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
