Optimized Numerical Integration for Embedded Architectures
Marina Bulat, Boris Antić · 2025
This paper explores the mathematical formalities and practical feasibility of implementing Modified Simpson's Rules for numerical integration in embedded system architectures. The study examines various Newton-Cotes-based integration methods, emphasizing the need for reduced numerical error and energy-efficient computation—critical factors in embedded applications. A formal proof of the derived formulas for the previously proposed Modified Simpson's Rules is provided, covering both the modification of Simpson's 1/3 Rule and Simpson's 3/8 Rule. The findings contribute to the optimization of numerical integration techniques tailored for embedded processing environments.