Non-linear Precoding via Vector Perturbation for Massive MIMO

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current baseband non-linear precoding techniques for massive MIMO systems face scalability issues, leading to high computational complexity and suboptimal performance, especially with a large number of transmit antennas, as existing methods like maximum-likelihood search and tree/trellis searching schemes are impractical due to exponential complexity growth.

Innovation Solution

A signal precoder that performs non-linear precoding by determining optimized signal perturbations using a balancer and selector, which scale signal components based on channel properties, reducing complexity by selecting a subset of singular values from the channel matrix through singular value decomposition, and applying these perturbations to minimize energy transmission, thereby achieving near-optimum detection performance with linear computational scalability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If maximum-likelihood search is used to solve non-linear MIMO precoding, then detection performance is improved, but computational complexity increases exponentially

Engineering Contradiction:
Improvedetection performanceVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts and optimizes only the most significant singular values from the channel matrix, rather than processing all singular values. This selective extraction reduces the dimensionality of the search space from exponential to polynomial complexity while maintaining near-optimal detection performance.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transforms the original non-linear precoding problem into a simplified form by changing parameters: selecting a subset of singular values, applying linear precoding first, and then adding a perturbation vector. This parameter transformation converts the intractable exponential complexity into manageable polynomial complexity.

Inventive Principle:
Principle #35Parameter changes

2Device complexity

If tree search or trellis search is used to reduce ML complexity, then computational complexity is reduced, but scalability remains insufficient

Engineering Contradiction:
Improvecomputational complexityVSAvoidscalability
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent segments the precoding process into two independent stages: linear precoding and perturbation addition. This segmentation allows each stage to be optimized separately, with linear precoding handling the bulk of signal processing and perturbation handling only the residual optimization, enabling better scalability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Instead of performing complete non-linear optimization, the patent applies partial action by adding a perturbation vector to the linearly precoded signal. This partial non-linear adjustment achieves most of the performance benefit without the full computational burden, improving scalability.

Inventive Principle:
Principle #16Partial or excessive action

3Adaptability or versatility

If linear precoding is used for large MIMO systems, then scalability is improved, but performance degrades significantly

Engineering Contradiction:
ImprovescalabilityVSAvoidperformance
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent merges linear precoding and non-linear perturbation techniques into a hybrid approach. The linear precoder provides scalability and basic performance, while the added perturbation vector recovers most of the performance loss, achieving both scalability and near-optimal performance.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The perturbation vector acts as an intermediary that bridges the gap between linear and non-linear precoding. It is added to the linearly precoded signal to correct residual interference and improve performance without requiring a complete non-linear solution.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3224964B1Maximizing energy efficiency in non-linear precoding using vector perturbation
Publication Date: 2019.11.27 UNIVERSITY OF SURREY
  • EP3224964B1 patent drawingFigure 1
  • EP3224964B1 patent drawingFigure 2
  • EP3224964B1 patent drawingFigure 3

AI summary

A signal precoder is provided for use with a signal transmitter comprising a plurality of antennas for transmitting a plurality of signal components to a receiver over a communications channel in which one or more properties of the communications channel are described by a plurality of values which scale t he components of the signal in a plurality of dimensions during transmission, in which the precoder is arranged to perform non-linear precoding on a source signal prior to signal transmission, the precoder comprising a balancer for determining a plurality of optimised amounts of signal perturbation required to be applied to one or more source signal components to minimise the energy of the signal transmission, when respectively scaled by each of the values describing one or more properties of the communications channel, and a selector for selecting which of the optimised amounts of signal perturbations results in signal transmission having the least energy when scaled by the values describing the one or more properties of the communications channel, wherein the precoder is arranged to perform non-linear precoding on the source signal by applying the selected optimised signal perturbation to the source signal and to output the precoded signal. The embodiments of the present invention offer near-optimum detection performances and linear computational scalability simultaneously. The present invention also provides a signal transmitter, a communications system and a signal receiving method.