.png){kind=icon}
.png){kind=icon}
Return
FPGA-Based Compact Differential Evolution for General-Purpose Optimization in Resource-Constrained Devices
.png){kind=icon}
PRE
AI
DOI：10.1109/TII.2024.3353851.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Resource-constrained devices in open environments face diverse optimization problems, so general-purpose optimization capabilities become important but are currently lacking. Our algorithm framework aims to fill this gap and better understand the issues when implementing general optimization in hardware, especially using field-programmable gate array. Therefore, the challenge is to design an algorithm framework that can handle different optimization problems with fewer hardware resources while maximizing solution performance. Based on the Zynq XC7Z020-2CLG400I device and the compact differential evolution (cDE) algorithm, this article describes the unified software and hardware architecture, the cDE algorithm that incorporates ensemble mutation and crossover strategy as well as uniform mutation operation (cDE-emc-um), providing an efficient and low-resource algorithm framework while maintaining generality. This article also theoretically analyzes the global convergence and low computational complexity of the cDE-emc-um and tests the general optimization capabilities of our algorithm framework through two optimization problems.
Keywords:
Optimization
Hardware
Task analysis
Field programmable gate arrays
Software algorithms
Uncertainty
Robot sensing systems
Compact differential evolution (cDE)
field-programmable gate array (FPGA) implementation
general-purpose optimization
global convergence
resource-constrained devices
Journal
IF:
9.9
Papers:
8.3K
Citations:
6.0W