Stochastic linear detection

US10819468B2 · kind B2 · utility

Inventors

• Ahmad Rezazadehreyhani · San Jose, US
• Jonathan Hedstrom · Salt Lake City, US
• Chung Him Yuen · Salt Lake City, US
• Kohl Riekhof · Cottonwood Heights, US
• Daryl Wasden · Draper, US
• Behrouz Farhang · Salt Lake City, US

Key dates

• Filing date: Dec 5, 2019
• Grant date: Oct 27, 2020
• Expiry date: Dec 5, 2039

Classification

• Technology area (CPC H): Electricity
• CPC primary: H04L2025/03426
• WIPO field: Digital communication
• WIPO sector: Electrical engineering

Abstract

Apparatuses, systems, and methods are disclosed for stochastic linear detection. A digital signal processor determines information about a plurality of transmitted signals based on a plurality of received signals. An initialization module determines an estimator matrix and a noise shaping matrix based on channel state information that relates the transmitted signals to the received signals. A sample generation module stochastically generates a plurality of signal estimates so that each signal estimate is a sum of a fixed component and a random component. The fixed component may be based on applying the estimator matrix to a vector of the received signals, and the random component may be based on applying the noise shaping matrix to generated noise. An output module sends soft information to an error-correcting code (ECC) decoder for decoding bits carried by the transmitted signals. The soft information may be based on the plurality of signal estimates.