ABSTRACT
Modular redundancy, the traditional approach to fault tolerance, is prohibitively expensive because of the overhead in replicating the hardware. In this paper we discuss alternative techniques for fault tolerance in sequence enumerators that are implemented as linear finite-state machines (LFSM’s). Our approach replaces a given LFSM with a larger, redundant LFSM that preserves the evolution and properties of the original one. The state of the larger LFSM is a linearly encoded version of the state in the original machine and allows an external mechanism to perform error detection and correction by identifying and analyzing violations of the code restrictions. In this paper, we characterize the class of appropriate redundant LFSM’s and demonstrate a variety of possibilities for fault tolerance, ranging from no redundancy to full replication.
Key Words: Fault tolerance, linear finite-state machines, concurrent error detection and correction.
ABSTRACT
An early result on the Smith-MacMillan form of a rational matrix is used for evaluating the degree of the determinant of a polynomial matrix using numerically reliable techniques. This allows for accurate determinant zeroing and determinant interpolation, thus improving existing numerical methods for polynomial matrix determinant computation.
Key Words. Polynomial Matrix, Determinant Computation, Numerical Methods.
ABSTRACT
An efficient implementation of the operation by ax + , on which the construction of second order digital filters and complex numbers multipliers is based, is presented. The quantitiesx and y are 2’s complement numbers in serial form. The numbers a and b are constant coefficients in Canonical Signed Digit form (CSD) while the result is obtained in se-rial 2’s complement form. The proposed scheme operates in pipeline mode with 100% hard-ware efficiency, namely, no sign extension words between successive data words are re-quired. The implementation is based on merging of two serial multipliers, which yields sig-nificant hardware reduction.
Key Words. Serial multipliers, Systolic multipliers, Digital filters
ABSTRACT
In this paper we propose two algorithms for the computation of the Drazin inverse, based on the Leverrier-Faddeev algorithm. These algorithms represent extensions of paper [3] and a continuation of the paper [4]. A few matrix equations which include rational matrices are solved by means of the Drazin inverse and the Moore-Penrose inverse of rational matrices.
Keywords. Drazin inverse, Leverrier-Faddeev algorithm, matrix equations.
ABSTRACT
In this paper, we demonstrate a performance of a PD type self-learning fuzzy logic controller (SLFLC), which has been implemented as a function block for the very popular Matlab+Simulink environment. The SLFLC described in detail in [1 ] utilizes a reference model and a sensitivity model for learning of SLFLC parameters. The effectiveness of the SLFLC function block has been demonstrated on the model of a closed-loop engine speed control system provided in Matlab for demo purposes. The results show very clearly how the SLFLC brings improved performance into the selected control system example.