Pilot Pouring in Superimposed Training for Channel Estimation in CB-FMT

Cyclic block filtered multi-tone (CB-FMT) is a waveform that can be efficiently synthesized through a filter-bank in the frequency domain. Although the main principles have been already established, channel estimation has not been addressed yet. This is because of assuming that the existing techniqu...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on wireless communications 2021-06, Vol.20 (6), p.3366-3380
Main Authors: Chen-Hu, Kun, Garcia, M. Julia Fernandez-Getino, Tonello, Andrea M., Armada, Ana Garcia
Format: Article
Language:English
Subjects:
CB-FMT
Channel estimation
Data transmission
Estimation
Filter banks
Frequency domain analysis
OFDM
pilot pouring
Pilots
Pouring
Prototypes
Receivers
SC-FDMA
superimposed training
Symbols
Time-frequency analysis
Training
Waveforms
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Cyclic block filtered multi-tone (CB-FMT) is a waveform that can be efficiently synthesized through a filter-bank in the frequency domain. Although the main principles have been already established, channel estimation has not been addressed yet. This is because of assuming that the existing techniques based on pilot symbol assisted modulation (PSAM), implemented in OFDM-like schemes, can be reused. However, PSAM leads to an undesirable loss of data-rate. In this paper, an alternative method inspired by the superimposed training (ST) concept, namely pilot pouring ST (PPST), is proposed. In PPST, pilots are superimposed over data taking advantage of the particular spectral characteristics of CB-FMT. Exploiting the sub-channel spectrum, the pilot symbols are poured in those resources unused for data transmission. This spectral shaping of pilots is also exploited at the receiver to carry out channel estimation, by enhancing those channel estimates that exhibit a low data interference contribution. Furthermore, a frequency domain resource mapping strategy for the data and poured pilot symbols is proposed to enable an accurate estimation in strongly frequency-selective channels. The parameters of the proposed scheme are optimized to minimize the channel estimation mean squared error (MSE). Finally, several numerical results illustrate the performance advantages of the proposed technique as compared to other alternatives.
ISSN:1536-1276
1558-2248
DOI:10.1109/TWC.2021.3049530