REPRODUCING KERNEL HILBERT SPACE METHODSIN MEASUREMENT SCIENCE

摘要

Basic functional model for random signal is presented for use in measurement theory. That is H? - reproducing kernel Hilbert space (RK-space), produced by the correlation function R(s, t) of random process x (t). RK-space H? presents an isomorphic representation of the process x(t). So it provides an adequate mathematical tool for solving several problems, such as linear filtering, extrapolation of random signal, and deterministic signal extraction from noise. Besides, the corresponding RK-norms are useful in metrology for employing as the measurement accuracy characteristics. As an illustration of the RK-approach, the pseudo-best B-estimates for the deterministic signal extraction from noise are considered.