A Computational Time Analysis of Discrete Simulated Kalman Filter Optimizer

Abstract

Simulated Kalman filter (SKF) is a population-based optimization algorithm based on the Kalman filter framework. To find the global optimum, the SKF applies a Kalman filter process that involves prediction, measurement, and estimation. However, the SKF can only operates in numerical search space. In literature, many techniques and modifications have been made to the SKF algorithms to function in a discrete search space. An example of the modified SKF is the discrete simulated Kalman filter optimizer (DSKFO). However, little research has been conducted on the DSKFO. This paper studies the computational time complexity of the DSKFO to acquire a better understanding of the algorithm's complexity. The analysis is done by comparing the computational time of the DSKFO against four combinatorial SKFs. The findings show that the DSKFO is the fastest algorithm for solving all TSP instances. The DSKFO requires just 13 s to solve the smaller TSP instance eil51, whereas SEDESKF, BSKF, DESKF, and AMSKF need 14, 34, 36, and 42 s, respectively. DSKFO solves the larger TSP instance dsj1000 in 79 s, whereas SEDESKF, BSKF, DESKF, and AMSKF need 182, 1104, 1125, and 1167 s, respectively.