抄録
Combining the modern theory of stochastic dynamical systems with the classical one of signal detections, this paper presents studies on the signal detection problems via realtime data processing. Descriptions in this paper are divided into four parts.
First, introducing some successful results of research at the primary stage of developing the theory of signal detection under white Gaussian noise observations, the motivation of the present studies is clearly shown. The logarithmic likelihood-ratio function plays an important role to detect the message signal corrupted with a random noise.
Secondly, the scenario of this paper leads the readers to the version of the signal detection under colored noise observations. Practical applications with real-time data processing are also shown after digital simulation experiments.
The third aspect is to develop a method for an early-stage signal detection in which a feasible approach is proposed to determine adaptively the threshold with both the decision time and the noise level.
The final stage of this paper shows an extension of the theory to deal with the detection of a spatially distributed signal, the detection of a moving signal source and ultimalely of a noise source location and/or faulty parts in a crippled system.
