Correlator Bank Circuit Reusing Product Values for IoT
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional correlation operations in IoT devices require large numbers of multipliers and adders, leading to high costs, size, and power consumption due to the need for extensive multiplication and addition operations with numerous correlation coefficient vectors.
Innovation Solution
The correlation computation circuit reuses product values by calculating them for a reference coefficient vector and then adjusting for other vectors using difference vectors, significantly reducing the number of multiplications and additions required.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a bank of parallel correlators is used to perform correlation operations with multiple coefficient vectors, then the correlation computation capability is improved, but the number of multipliers and adders increases significantly, leading to higher cost, larger size, and increased power consumption
Solution Approach 1:
The patent merges the computation of multiple correlation outputs by identifying and reusing common product terms. Instead of computing each correlation independently with separate multipliers, the invention combines the coefficient vectors and shares common multiplication operations, thereby reducing the total number of multipliers and adders while maintaining the ability to compute multiple correlations simultaneously
Solution Approach 2:
The patent creates a universal computation structure that can handle multiple coefficient vectors using the same set of multipliers and adders. By organizing the computation to process multiple correlations through a shared resource pool, the system achieves multi-functionality where the same hardware components serve multiple correlation computation tasks, reducing overall hardware requirements
2Adaptability or versatility
If the number of correlation coefficient vectors is increased to handle more IoT protocols, then the adaptability is improved, but the power consumption increases due to the large number of multipliers and adders required
Solution Approach 1:
The patent merges power-consuming multiplication operations by combining coefficient vectors and sharing common product computations. This reduction in the number of active multipliers directly lowers dynamic power consumption while maintaining support for multiple IoT protocols through the unified correlation computation architecture
Solution Approach 2:
The patent changes the computational parameters by reorganizing how correlations are computed - shifting from independent computation of each correlation to a combined computation approach. This parameter change in the computation methodology reduces the operational count of power-intensive multipliers and adders, thereby reducing overall power consumption while maintaining protocol versatility
Data Source
AI summary
In some embodiments, a circuit that performs a correlation between vectors of coefficients and a vector of input values (input vector) is disclosed. A multiplication block outputs a first vector of product values equal to a first one of a reference vector of coefficients multiplied by the input vector. A first calculation circuit sums the elements in a first vector of product values to generate a first output value. The correlation computation circuit also includes a second calculation circuit configured to: sum a difference vector of product values that correspond to each element in a difference vector of coefficients multiplied by each element in a difference vector of inputs to generate a summed output, multiply the summed output to generate a multiplied summed output, and output a second output value that is equal to the first output value minus or plus the multiplied difference value.