GB1006641A - Improvements in and relating to devices for determining statistic parameters of signals - Google Patents

Abstract

1,006,641. Electric digital calculators. PHILIPS ELECTRONIC & ASSOCIATED INDUSTRIES Ltd. Feb. 5, 1964 [Feb. 8, 1963], No. 4852/64. Heading G4A. [Also in Division G1] In an arrangement for computing a statistical parameter such as auto- or cross-correlation of one or more input variables, each input variable is added to an auxiliary function, to produce a sum-function, the auxiliary functions being mutually statistically independent and being such that all levels occurring therein occur with the same frequency; scanning circuits are provided to produce series of equi-valued pulses representing the sign of the sum-functions; and these pulses are applied to a logical circuit which computes the required statistical parameter. The arrangement is characterized in that the auxiliary functions are composed of linear combinations of Rademacher functions, a Rademacher function r l (t), i = 1, 2, 3 ... being defined as a periodic function of t with period T/2<SP>i</SP> which has in the interval 0 < t < T/2<SP>i+1</SP> the value + 1 and in the interval T/2<SP>i+1</SP> < t < T/2<SP>i</SP> the value - 1. Fig. 2 shows the form of such a linear combination of Rademacher functions. A circuit for computing the auto-correlation function of an input function x(t) is shown in Fig. 4, the input function being applied to addition circuits A 1 , A 2 where it is added to two different statistically independent functions y 1 (t), Y 2 (t) each composed of a linear combination of Rademacher functions. The outputs z 1 (t), z 2 (t) of the adders are fed to limiters B 1 , B 2 which produce outputs representing the sign of the functions z 1 , z 2 , the sign functions being applied to scanners C 1 , C 2 which produce pulse trains representing the sign functions. The respective pulse trains are supplied to a logical circuit R, the first train after delay in a shift register DL. The circuit R (not described in detail) comprises counters and other digital circuits and produces the required output auto-correlation function. The arrangement may be applied to computing the mean value of a variable or for computing the cross-correlation of a plurality of variables.