An Improved Minor Component Analysis Algorithm Based on Convergence Analysis of 5G Multi-Dimensional Signals

With the availability of 5G, multi-dimensional signals’ characteristic analysis becomes more challenging. Minor component analysis (MCA) is a useful data analysis approach to estimate the important features of multi-dimensional signals. MCA neural network approaches can extract minor components from real-time data streams not only adaptively but also online. The rapid convergence of MCA neural network approaches is valuable in practical applications. We propose a deterministic discrete time-based method to analyze the convergence performance of MCA neural network algorithms and reveal the effects of initial weight vectors’ norm in MCA algorithms on convergence speed. Then, based on the convergence analysis, a new MCA algorithm is proposed to achieve faster convergence than existing MCA algorithms, which are proven by both the theoretical analysis and simulation results.