Machine-learning based method for MIMO detection complexity reduction
US10826581B1 · kind B1 · utility
Assignee
Inventors
Key dates
| Filing date | Jul 19, 2019 |
| Grant date | Nov 3, 2020 |
| Priority date | — |
| Expiry date | Jul 19, 2039 |
Classification
- Technology area (CPC H)Electricity
- CPC primaryH04L2025/03426
- WIPO fieldDigital communication
- WIPO sectorElectrical engineering
Abstract
A machine-learning based multiple-input, multiple-output (MIMO) demapper for a wireless device may include a classifier that selects which MIMO demapper to use for a sample for a particular tone. For example, a wireless device may receive via a plurality of antennas a plurality of signals including a plurality of tones. The wireless device may determine selection features for each tone of the plurality of tones. The wireless device may select, for each tone, by the classifier based on the selection features, a selected demapper from at least a first MIMO demapper and a second MIMO demapper. The second MIMO demapper may have a different performance characteristic than the first MIMO demapper. The wireless device may detect, for each tone, one or more streams using the selected demapper for the tone. A stream may refer to a sequence of bits.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.