Performance analysis of a deterministic channel estimator for block transmission systems with null guard intervals

Sergio Barbarossa, Anna Scaglione, Georgios B. Giannakis

Research output: Contribution to journalArticle

24 Scopus citations

Abstract

A deterministic algorithm was recently proposed for channel identification in block communication systems. The method assumed that the channel is finite impulse response (FIR) and that null guard intervals of length greater than the channel order are inserted between successive blocks to prevent interblock interference and allow block synchronization. In the absence of noise, the algorithm provides error-free channel estimates, using a finite number of received data, without requiring training sequences and without imposing a restriction neither on the channel, except for finite order and time invariance, nor on the symbol constellation. Using small perturbation analysis, in this paper, we derive approximate expressions of the estimated channel covariance matrix, which are used to quantify the resilience of the estimation algorithm to additive noise and channel fluctuations. Specifically, we consider channel fluctuations induced by transmitter/receiver relative motion, asynchronism, and oscillators' phase noise. We also compare the channel estimation accuracy with the Cramér-Rao bound (CRB) and prove that our estimation method is statistically efficient at practical SNR values for any data block length. Finally, we validate our theoretical analysis with simulations and compare our transmission scheme with an alternative system using training sequences for channel estimation.

Original languageEnglish (US)
Pages (from-to)684-695
Number of pages12
JournalIEEE Transactions on Signal Processing
Volume50
Issue number3
DOIs
StatePublished - Mar 2002
Externally publishedYes

    Fingerprint

Keywords

  • Channel estimation
  • Equalization
  • Theoretical bounds

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Signal Processing

Cite this