I have a question on CP-OFDM. In recent years people are talking about filter bank approach where they are trying to filter out the out of band side lobes using filters. Besides they are presuming that this approach may provide some robustness and spectral efficiency since they can get rid of CP. Is this true and what aspects we can remove CP and make spectral efficiency equal one.

Ref. F. Schaich, T. Wild, Y. Chen , „Waveform contenders for 5G – suitability for short packet and low latency transmissions“, accepted for IEEE VTCs’14, Seoul, Korea, April 2014


In CP-OFDM, the transmitted signal is the output of the IDFT appended with a cyclic prefix. No pulse-shaping is used.

In so-called "new waveforms" proposed as candidate for 5G, pulse shaping is heavily used.

In FBMC, each subcarrier (i.e each complex exponential) is convoluted with a pulse with a certain shape (usually a bell shape with beginning and end at zero amplitude). As a consequence of convolution, the amplitude of each filtered subcarrier at the beginning and at the end of the multicarrier symbol goes to zero. Consequently, in case of multi-path, this zero-valued beginnings and ends avoid ISI, since the part of the previous symbol overlapping most with the current one is its end, which as we said is zero-valued (or almost) and the current symbol has little energy at its beginning (zero or almost).

Hope this helps.

| improve this answer | |
  • $\begingroup$ If I understand correctly, FBMC makes CP-OFDM to zero-padded OFDM (if we remove CP and do zero padding that might help us to reduce ICI and IBI noises). In such case how can we achieve full spectral efficiency as we are not transmitting data for full block. May you share some thoughts in this regard. Thank You $\endgroup$ – ibrahimkbd Jun 12 '15 at 14:41
  • $\begingroup$ To the best of my knowledge CP-OFDM does use pulse shaping at the boundaries of the OFDM symbols, although the length of the pulse-shaping window is pretty short (4-8 samples) $\endgroup$ – Naveen Jun 12 '15 at 15:27
  • $\begingroup$ FBMC is not zero-padded OFDM. When you convolute a sequence of length N and a sequence of length L, the output is of length N+L-1. This natural increase in length due to the subcarrier-per-subcarrier filtering convolution is the only thing taking place in FBMC. The main interest of FBMC and other filter-based waveforms is the reduction of out-of-band emissions. Up to my knowledge, there is not any waveform achieving full spectral efficiency in doubly dispersive channels. $\endgroup$ – Th B Research Jun 16 '15 at 2:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.