Online adaptive algorithms have been largely applied for recursive estimation and tracking of sparse signals. In this paper we propose a distributed recursive least squares (RLS) algorithm incorporating an l 1-norm regularization with time-varying regularization coefficient that enables a recursive distributed solution with no losses with respect to the centralized solution. The method is especially useful in cooperative sensing when the parameters to be estimated are structurally sparse and time-varying. As well known, the l1-norm is useful to recover sparsity, but it also introduces a non negligible bias. To tackle this issue, we further apply a garotte correction to our distributed mechanism that strongly reduces the bias. Numerical results are included to validate the estimation and tracking capabilities of the proposed algorithm. © 2013 IEEE.

DISTRIBUTED RLS ESTIMATION FOR COOPERATIVE SENSING IN SMALL CELL NETWORKS / SARDELLITTI, Stefania; BARBAROSSA, Sergio. - (2013), pp. 5283-5287. ((Intervento presentato al convegno IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) tenutosi a Vancouver; Canada nel MAY 26-31, 2013 [10.1109/icassp.2013.6638671].

DISTRIBUTED RLS ESTIMATION FOR COOPERATIVE SENSING IN SMALL CELL NETWORKS

SARDELLITTI, Stefania;BARBAROSSA, Sergio
2013

Abstract

Online adaptive algorithms have been largely applied for recursive estimation and tracking of sparse signals. In this paper we propose a distributed recursive least squares (RLS) algorithm incorporating an l 1-norm regularization with time-varying regularization coefficient that enables a recursive distributed solution with no losses with respect to the centralized solution. The method is especially useful in cooperative sensing when the parameters to be estimated are structurally sparse and time-varying. As well known, the l1-norm is useful to recover sparsity, but it also introduces a non negligible bias. To tackle this issue, we further apply a garotte correction to our distributed mechanism that strongly reduces the bias. Numerical results are included to validate the estimation and tracking capabilities of the proposed algorithm. © 2013 IEEE.
9781479903566
File allegati a questo prodotto
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11573/772176
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 8
  • ???jsp.display-item.citation.isi??? 2
social impact