Cumulant-based deconvolution and identification - several new families of linear-equations

F C Zheng, Steve McLaughlin, Bernard Mulgrew

Research output: Contribution to journalArticlepeer-review

10 Citations (Scopus)


This paper presents several new families of cumulant-based linear equations with respect to the inverse filter coefficients for deconvolution (equalisation) and identification of nonminimum phase systems. Based on noncausal autoregressive (AR) modeling of the output signals and three theorems, these equations are derived for the cases of 2nd-, 3rd and 4th-order cumulants, respectively, and can be expressed as identical or similar forms. The algorithms constructed from these equations are simpler in form, but can offer more accurate results than the existing methods. Since the inverse filter coefficients are simply the solution of a set of linear equations, their uniqueness can normally be guaranteed. Simulations are presented for the cases of skewed series, unskewed continuous series and unskewed discrete series. The results of these simulations confirm the feasibility and efficiency of the algorithms.

Original languageEnglish
Pages (from-to)199-219
Number of pages21
JournalSignal Processing
Issue number2
Publication statusPublished - Jan 1993


Dive into the research topics of 'Cumulant-based deconvolution and identification - several new families of linear-equations'. Together they form a unique fingerprint.

Cite this