US 7,218,665 B2Granted

Deferred decorrelating decision-feedback detector for supersaturated communications

AssigneeBAE Systems Information and Electronic Systems Integration Inc.

InventorThomas P McElwain

Life of this patent

Estimating the term…

Loading patent…

Abstract

The present invention provides an efficient means of estimating symbols transmitted in a multi-user environment in overloaded or super-saturated conditions by employing a deferred decorrelating decision feedback detector. In one embodiment, the present invention comprises a parameter estimation unit, filter bank, overloaded whitener and decision tree-based hypothesis testing. Parameter estimation defines the matched filter bank, whitening filters, and the terms of the hypothesis testing module. The whitening filter partially decouples co-channel interference and partially whitens the noise. The decision tree approach defers decisions until more evidence is accumulated and is a generalization that encompasses the jointly optimal maximum likelihood detector as well as the simpler decision feedback detectors.

Claims20

Dependent claims

A whitened front end for a tree-pruned multiuser detection (MUD) system, comprising:

a parameter estimator coupled to a data stream providing estimation values of said data stream, wherein said data stream is an aggregate of a plurality of signals, and each signal is assigned a channel, and wherein said plurality of signals exceeds a number of signal space dimensions of said MUD, resulting in an overloaded condition;
a filter bank having a plurality of filters coupled to said data stream and coupled to said parameter estimator, wherein each of said signals is individually coupled to said plurality of filters and produces a filtered output, and wherein said filters exceeds said number of signal space dimensions;
a whitener designer coupled to said parameter estimator, wherein said whitener designer computes a whitener factorization for said signals, wherein the signals exceeds said number of signal space dimensions;
an overloaded whitener unit coupled to said filter bank and said whitener designer, wherein said whitener unit applies said whitener factorization to said filtered output to produce a partially de-correlated and whitened output;
and, a symbol-hypothesis testing section coupled to said whitener designer and said whitened output, wherein said symbol-hypothesis testing section uses a hypothesis pruning to perform a decision tree search.

2 more elements

Dependencies are inferred from the claim text.