Target Vector Identification With Sorted Distance Pruning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing speech and audio codecs face challenges in reducing computational complexity while maintaining coding efficiency, particularly in identifying the nearest code vector in structured codebooks, which is crucial for low-complexity mobile terminal communications.

Innovation Solution

The method involves identifying target vectors from candidate vectors by transforming the input vector into a sorted absolute-valued representation and using a reference vector to efficiently compute distances, reducing the need for explicit distance calculations by leveraging the triangle inequality, thereby minimizing computational complexity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If explicit distance calculations are performed for all candidate vectors to identify the nearest code vector, then measurement precision is improved, but device complexity increases

Engineering Contradiction:
Improvedistance calculation accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent transforms the input vector into a sorted absolute-valued representation before distance calculations. This preliminary transformation organizes the data in a way that enables efficient distance computation using the triangle inequality, reducing the need for full explicit distance calculations while maintaining accuracy in identifying the nearest code vector

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent extracts and utilizes only the necessary components for distance calculation by working with sorted absolute values rather than full vector components. This extraction approach allows the algorithm to compute distances using only relevant information, significantly reducing computational complexity while preserving measurement precision

Inventive Principle:
Principle #2Taking out (Extraction)

2Device complexity

If structured codebooks are used to reduce storage and complexity, then device complexity is reduced, but coding efficiency may be compromised

Engineering Contradiction:
Improvestorage and computation complexityVSAvoidcoding efficiency
Core Design Contradiction:
Device complexityVSProductivity

Solution Approach 1:

The patent changes the parameter representation by transforming input vectors into sorted absolute-valued forms. This parameter transformation enables the use of structured codebooks with lattice structures while maintaining coding efficiency, as the transformed parameters allow for more efficient search and distance calculation algorithms that work well with structured codebook organizations

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP2643833B1Low complexity target vector identification
Publication Date: 2020.01.01 NOKIA TECHNOLOGIES OY
  • EP2643833B1 patent drawingFigure 1~3
  • EP2643833B1 patent drawingFigure 4
  • EP2643833B1 patent drawingFigure 5

AI summary

It is inter alia disclosed to identify one or more target vectors from a plurality of candidate vectors, each candidate vector having sorted elements and being associated with a respective class of one or more code vectors of a codebook and at least one of the candidate vectors being associated with a respective class of two or more code vectors that comprise the respective candidate vector and at least one code vector obtainable from the respective candidate vector by one of permutation and signed permutation, the target vectors having, among all candidate vectors of the plurality of candidate vectors, smallest distances towards a at least sorted representation of an input vector. The identifying comprises checking, for a candidate vector of the plurality of candidate vectors, at least based on a distance between the candidate vector and a reference vector and on a distance between the reference vector and the at least sorted representation of the input vector, if a distance between the at least sorted representation of the input vector and the candidate vector is larger than a distance between the at least sorted representation of the input vector and the reference vector. The identifying further comprises computing, for the candidate vector, the distance between the at least sorted representation of the input vector and the candidate vector only if the checking yields a negative result.