Suppose X 1, X 2, X 3 and X 4 denote the lifetimes of the four components of the minicomputer. Here it is given that the lifetimes of the components are mutually independent exponentially distributed random variables with j-th component has mean lifetime 1/j, j = 1,2,3,4. Thus the survival functions are P[ X 1 t ] et , P[ X 2 t ] e2t , P[ X 3 t ] e3t and P[ X 4 t ] e4t a) The probability that no component of a minicomputer fails during the first 0.2 years of operation is given by P[ X 1 0.2, X 2 0.2, X 3 0.2, X 4 0.2] = P[ X 1 0.2].P[ X 2 0.2].P[ X 3 0.2].P[ X 4 0.2] , since the lifetimes of the components are mutually independent. = e0.2 e2*0.2 e3*0.2 e4*0.2 e2 = 0.135335 b) At any time ‘t’ the probability that component 1 of a minicomputer is the first component of that minicomputer to fail is given by P[ X 1 t , X 2 t , X 3 t , X 4 t ] = P[ X 1 t ].P[ X 2 t ].P[ X 3 t ].P[ X 4 t ] = (1 et )e2t e3t e4t = (1 et )e9t Let S denote the lifetime of the minicomputer and FS (t ) P[ S t ] denote the survival function of the minicomputer system. Here it is given that the system ( minicomputer) remains functioning as long as at least 2 of the 4 components are functioning. Thus the given system is a 2 out of 4 system. Thus the survival function of the minicomputer system is given by FS (t ) P[ S t ] = P[ X 1 t , X 2 t , X 3 t , X 4 t ] + P[ X 1 t , X 2 t , X 3 t , X 4 t ] + P[ X 1 t , X 2 t , X 3 t , X 4 t ] + P[ X 1 t , X 2 t , X 3 t , X 4 t ] + P[ X 1 t , X 2 t , X 3 t , X 4 t ] + P[ X 1 t , X 2 t , X 3 t , X 4 t ] + P[ X 1 t , X 2 t , X 3 t , X 4 t ] + P[ X 1 t , X 2 t , X 3 t , X 4 t ] + P[ X 1 t , X 2 t , X 3 t , X 4 t ] + P[ X 1 t , X 2 t , X 3 t , X 4 t ] + P[ X 1 t , X 2 t , X 3 t , X 4 t ] = et e2t e3t e4t + et e2t e3t (1 e4t ) + et e2t (1 e3t )e4t + et (1 e2t )e3t e4t + (1 et )e2t e3t e4t + et e2t (1 e3t )(1 e4t ) + et (1 e2t )e3t (1 e4t ) + et (1 e2t )(1 e3t )e4t + (1 et )e2t e3t (1 e4t ) + (1 et )e2t (1 e3t )e4t + (1 et )(1 e2t )e3t e4t 10 t 7t 6t 4t 3t 2 9t 2 8t 2 5t = 3 c) The expected lifetime of the minicomputer is given by m E ( S ) FS (t )dt 0 10 t 3 = 2 9t 2 8t 7t 6t 2 5t 4t 3t t 0 3 2 2 1 1 2 1 1 158 = 10 9 8 7 6 5 4 3 315 = 0.501587 = (1/2) years approximately d) The variance of the lifetime of the minicomputer is given by = Var ( S ) 2 tFS (t )dt m 2 ( This is the standard formula for variance for 2 0 positive random variables) But 2 tFS (t )dt = 2 t 10 t 3 2 9t 2 8t 7t 6t 2 5t 4t 3t t 0 0 = 569897 = 0.358968 1587600 2 569897 158 170473 1587600 315 1587600 = 0.107378 e) Let N(t) denote the number of renewal in the interval [0, t]. Then { N(t), t ≥ 0} will be a Poisson process if the interarrival times ( the time between successive replacement of the minicomputer ), the random variable S must be exponentially distributed. That is the survival function of S must be of the form FS (t ) P[ S t ] e t . But here the distribution of S is not exponentially distributed as its survival function is of the form 10 t 7t 6t 4t 3t 2 9t 2 8t 2 5t F (t ) 3 , which is Thus 2 = S not the survival function of an Exponential random variable. Thus { N(t), t ≥ 0} is not a Poisson process. f) Even though { N(t), t ≥ 0} is not a Poisson process, it is a renewal process since the interarrival times are identically distributed with the survival function FS (t ) . By the theory of renewal process for sufficiently large ‘t’ n (t / m) P[ N (t ) n] , where (.) is the cumulative distribution function 3 t/m of a Standard Normal random variable ( For the Proof refer Ross( 1996, page 109) Ross, S.M Stochastic Processes, 2nd Ed. Wiley, New York ) Now the probability that at least 50 computers have to be replaced during the first 10 years of operation of the alarm system is given by 50 (10 / m) P[ N (10) 50] 1 P[ N (10) 50] 1 3 10 / m where m = 0.501587 and = 2 0.327686 Substituting these values we get P[ N (10) 50] 1 10.3062 =1-1 = 0