Том 60, № 6 (2017)

The multistage digital automatic compensators of active noise interferences have been synthesized using the block Gram-Schmidt orthogonalization and LS-algorithm based on the least squares criterion. The specified automatic compensators ensure parallel-serial processing of signals. The application of RLS-algorithm made it possible to obtain a recurrent procedure for calculation of weighting coefficients of automatic compensator modules represented in the form of multiinput weighting adders. Statistical computer simulation was used to analyze the multistage digital automatic compensators based on the RLS-algorithm with simultaneous adaptation of weighting adders of all stages. As a result of parallelization of computational process, the synthesized multistage automatic compensators made it possible to increase the speed of processing (signal sampling frequency) by one order of magnitude and more as compared to that of automatic compensators built on single-stage scheme.

This paper presents comparative analysis of various algorithms of distributed power control used in Code Division Multiple Access (CDMA) systems. These algorithms include Distributed Balancing power control algorithm (DB), Modified Distributed Balancing power control algorithm (MDB), Fully Distributed Power Control algorithm (FDPC), Distributed Power Control algorithm (DPC), Distributed Constrained Power Control algorithm (DCPC), Unconstrained Second-Order Power Control algorithm (USOPC), Constrained Second-Order Power Control algorithm (CSOPC), and Fixed Step Distributed Power Control algorithm (FSDPC). These algorithms are compared based on the rate of convergence to the target signal-to-interference ratio, the rate of convergence of the power, and the rate of convergence of utilities. Simulation results show that the FDPC is the best choice according to these parameters.

The solution of the problem of computational time reduce during optimization of electronic circuits allows to enhance the development quality. Generalized methodology of the circuits optimization developed before on a basis of optimal control allows to define many different optimization strategies. Definition of Lyapunov’s function of optimization process and its analysis for different strategies allows to compare these strategies from viewpoint of computational burden and select the best of them. At the same time the most grounded approach for the search of optimal development strategy in this statement is Pontryagin maximum principle. But application of this principle for solution of non-linear problems is related to essential complications. In this paper it is obtained the solution of electronic circuit optimization problem during minimal amount possible processor time on a basis of Pontryagin maximum principle in general case of N variables. It is shown that effect studied before for acceleration of the process of optimization coincides the solution on a basis of the maximum principle. This fact is the theoretical explanation of the acceleration effect. From the other hand the principle of maximum can be the basis for development of the algorithm for electronic circuits optimization with minimal processor time cost.