Time-Frequency Atom Decomposition with Quantum-Inspired Evolutionary Algorithms

Time-frequency atom decomposition (TFAD) provides a flexible representation for non-stationary signals, but the extremely high computational effort greatly blocks its practical applications. Quantum-inspired evolutionary algorithms (QEA) are efficient optimization methods with strong search capability and rapid convergence. This paper proposes the application of a modified variant of QEA to the TFAD problem. The problem on TFAD with evolutionary algorithms is formulated. By using gray coding, elite groups, and an appropriate termination criterion, the modified QEA is developed to search the suboptimal time-frequency atoms from a very large and redundant time-frequency dictionary. Also, this paper discusses the reduction of the computational time in terms of parameter setting, and presents an application example of radar emitter signals. Extensive experiments show the effectiveness and practicability of the presented algorithm.