US Army Materiel Command AMCP 706-197 — AMCP 706-197 Reliability Prediction Handbook — US Army Engineering Design Guide PDF

Downloaded from https://www.everyspec.com AM7W47 J putation F x ,, ), and an encoding function e than t, errors, then P, the probability of a such that sr.tem error, can be calculated as H(x ) - e{Fd(xl) (1048) N where P-I- (1 - a,) (10-40) i-1 H(xt ) - fHj (0 t ), H. (x Q ) ., ff. (0 t ) }whr e{Fd(x') I - [eIl {F[d(x't)I }, where e2 {F[d(x'l) },... ,e. {F[d(xe)J l = (T)p(f)I1 - Ilei e, {Ffd(x')1 = e, [F, jd(x, )J, i" 5-t F2 fd(xl), .. ,1 , ld(xt)) ( "= +7 8+1 ' (tj) F [d(x)] -f F Id, (x'), d. (xe), ...

,, (XQ)] njp(t) 1 (10-41) d,(x') -=d,(xx .,4) (10-39) Let a measure of improvement I, the improv-ment factor, be dei'ied av the ratio of Here, dxl ) is the decoding function cowre- the system error probability before and after sponding to some error-correcting code that is coding, assumed to have been used on the output of the preceding network, and e is the encoding 1 = P8 IPA (10-42) function corresponding to some code which, where of course, also must be accommodated on the BA are subscripts refer ing to Before and input of the following network. The net result After coding, respectively.

is to replace one network by another where the two networks are related through two If, for simplicity, it is assumed that the sys- error-correcting codes, such that, in the tem is sufficiently homogeneous that all the absence of errors, a given input and output networks have the same number of inputs and state of the second is the encoded form of the ouitputs, and the same error-correction ca- corresponding input and output states of the pacity(n = n 1 ,Q = t, and t, = t,, for all i and j), first.

then The performance of devices using coding a, = j = a (10-43) redundancy can improve the correctness of for all i,j, and output signals and also the engineering con- fidence of the individuals using the equip. = 1 - (1 - )N - Ne, (1044) ment. If the decoding function becomes Na 1 complex, the usefulness of coding redundancy Thus is minimized, and this appears to be the major drawback of coding redundancy. I:,- & a / &A (10-45) To estimate reliability improvement, con- A detailed explanation of the practical sider first a system model that will be used to problems associated with this type of design is estimate a system error probability. In this presented in Ref. 12.

model, a system consists of N combinatorial networks arranged in an arbitrary order (any 10-4 DECISION-WITH-SWITCHING RE- combination of series and parallel). Network j DUNDANCY has n 1 outputs being generated by devices having a fan-in of Q 1 , each of which has an 104.1 STANDBY REDUNDANCY error probability ofp( 1 ). Let aj be defined as the probability that more than t, of the n, A system in which a component or unit is outputs are in error, where t, is the maximum stanesing by idly (cold standby) and operates number of errors that can be corrected by the only when the preceding unit fails is said to code used in the output of network j.

Assume be using standby or sequential redundancy that a system error is oblained if one or more (Refs. 1 and 13). A standby system usually networks generate an output having more requires failure-sensing and/or switching net- 10-15