This monograph deals with the construction of optimal, asymptotically optimal and quasioptimal sequential rules for hypothesis testing and estimation of parameters in the presence of complete a priori information as well as under the condition of a priori indeterminacy. Special attention is paid to the analysis of the sequential rules and their comparison with the non-sequential ones based on a formerly fixed time of observation. The sequential rules are proposed for testing both simple and complex hypotheses, the testing of two or more hypotheses being included.
This monograph covers the cases of independent homogeneous observations as well as inhomogeneous correlated ones, at both continuous and discrete times of observations. The experimental results indicate that the sequential methods make possible to increase significantly the throughput (capacity) of information systems of different types (communication, coordinate determination, management, queueing etc.), which is important for efficient use of computers within contemporary information networks. This monograph is intended first for research workers in information transmission and processing as well as in optimal management.
Chapter headings: (1) Optimal sequential processes of decision making; (2) Sequential hypothesis testing at discrete times of observations; (3) Sequential estimation of constants and variables; (4) Simultaneous sequential hypothesis testing and estimation of parameters; (5) Sequential hypothesis testing at non-stationary correlated observations and continuous times of observations; (6) Synthesis and analysis of sequential procedures for hypothesis testing under the condition of a priori indeterminacy.