Reliability Analysis of Communication Networks: UIUC ECE 413 Problem Set 7 Solutions, Assignments of Statistics

Solutions to problem set 7 of the university of illinois ece 413 course, focusing on the reliability of multiple connections, failure rate function for discrete random variables, reliability of a self-healing ring communication network, network reliability problem, a series parallel network, and packet length choice for noisy channels.

Typology: Assignments

Pre 2010

Uploaded on 03/16/2009

koofers-user-yp9
koofers-user-yp9 🇺🇸

10 documents

1 / 3

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
University of Illinois Spring 2007
ECE 413: Solutions to Problem Set 7
1. The reliability of multiple connections
The probability of total outage on a given day is pk, so the probability of total outage at least one
day during a year is 1 (1 pk)365. For p= 0.001, this gives values 0.3059 for k= 1, 0.000365 for
k= 2, and 3.65 e-07 for k= 3. We find that k= 2 suffices (even for a leap year).
2. Failure rate function for discrete random variables
(a) Since {T=k}⊂{Tk}, the definitions of conditional probability and pmf yield λT(k) =
P(T=k|Tk) = P(T=k)
P(Tk)=pT(k)
P
l=kpT(l). Or, equivalently, λT(k) = pT(k)
1Pk1
l=1 pT(l). (In particular,
λT(1) = pT(1). Note that λT(k) is only well defined if P(Tk)>0.)
(b) By definition, λT(kf) = 1 if and only P(T=kf) = P(Tkf)>0. So λT(kf) = 1 if and only
if P(T > kf) = P(Tkf)P(T=kf) = 0 and P(T=kf)>0. Thus, the condition λT(kf)=1
means that Thas a largest possible value, and that value is kf.
(c) For 1 kn,λT(k) = pT(k)
1Pk1
l=1 pT(l)=
1
n
1k1
n
=1
nk+1 . (That is, the first nvalues of λTare
1
n,1
n1, . . . , 1
1.)
(d) One solution is to first derive a general formula. Note that
P(Tk+ 1) = P({Tk+ 1}∩{Tk})
=P(Tk+ 1|Tk)P(Tk)
= (1 P(T=k|Tk))P(Tk)
= (1 λT(k))P(Tk) (1)
Starting with the initial condition P(T1) = 1, (1) can be used to solve for P(Tk) for all k.
The result is P(Tk) = (1 λT(1))(1 λ2(T)) · · · (1 λT(k1)) for k1. Therefore,
P(T=k) = P({Tk}∩{T=k})
=P(Tk)P(T=k|Tk) = P(Tk)λT(k)
= (1 λT(1)) · · · (1 λT(k1))λT(k).
In particular, if λT(k) = pfor all k1, then pT(k) = (1 p)k1pfor k1. In other words, Thas
the geometric distribution with parameter p.
(e) Fix k0. Consider conditional probabilities given the event {Tk}. Equivalently, consider
conditional probabilities given that both links are still working at time k. Then the conditional
probability that the network fails at time k, is the conditional probability that the first link fails at
time k, plus the conditional probability that the second link fails at time k, minus the conditional
probability that both links fail at time k. That is λT(k) = λ1(k) + λ2(k)λ1(k)λ2(k). (Note: If
λ1(k) and λ2(k) are both very small, then λT(k)λ1(k) + λ2(k).)
1
pf3

Partial preview of the text

Download Reliability Analysis of Communication Networks: UIUC ECE 413 Problem Set 7 Solutions and more Assignments Statistics in PDF only on Docsity!

University of Illinois Spring 2007 ECE 413: Solutions to Problem Set 7

  1. The reliability of multiple connections The probability of total outage on a given day is pk, so the probability of total outage at least one day during a year is 1 − (1 − pk)^365. For p = 0.001, this gives values 0.3059 for k = 1, 0.000365 for k = 2, and 3.65 e-07 for k = 3. We find that k = 2 suffices (even for a leap year).
  2. Failure rate function for discrete random variables (a) Since {T = k} ⊂ {T ≥ k}, the definitions of conditional probability and pmf yield λT (k) = P (T = k|T ≥ k) = P P^ ((TT^ =≥kk)) = P∞ lp=Tk^ ( pkT) (l). Or, equivalently, λT (k) = (^1) −PpkT−^ ( 1 k) l=1 pT^ (l)^ . (In particular,

λT (1) = pT (1). Note that λT (k) is only well defined if P (T ≥ k) > 0 .) (b) By definition, λT (kf ) = 1 if and only P (T = kf ) = P (T ≥ kf ) > 0. So λT (kf ) = 1 if and only if P (T > kf ) = P (T ≥ kf ) − P (T = kf ) = 0 and P (T = kf ) > 0. Thus, the condition λT (kf ) = 1 means that T has a largest possible value, and that value is kf.

(c) For 1 ≤ k ≤ n, λT (k) = (^1) −PpkT−^ ( 1 k) l=1 pT^ (l)^

1 n 1 − k− n^1 =^

1 n−k+1. (That is, the first^ n^ values of^ λT^ are 1 n ,^

1 n− 1 ,... ,^

1 1 .) (d) One solution is to first derive a general formula. Note that

P (T ≥ k + 1) = P ({T ≥ k + 1} ∩ {T ≥ k}) = P (T ≥ k + 1|T ≥ k)P (T ≥ k) = (1 − P (T = k|T ≥ k))P (T ≥ k) = (1 − λT (k))P (T ≥ k) (1)

Starting with the initial condition P (T ≥ 1) = 1, (1) can be used to solve for P (T ≥ k) for all k. The result is P (T ≥ k) = (1 − λT (1))(1 − λ 2 (T )) · · · (1 − λT (k − 1)) for k ≥ 1. Therefore,

P (T = k) = P ({T ≥ k} ∩ {T = k}) = P (T ≥ k)P (T = k|T ≥ k) = P (T ≥ k)λT (k) = (1 − λT (1)) · · · (1 − λT (k − 1))λT (k).

In particular, if λT (k) = p for all k ≥ 1, then pT (k) = (1 − p)k−^1 p for k ≥ 1. In other words, T has the geometric distribution with parameter p. (e) Fix k ≥ 0. Consider conditional probabilities given the event {T ≥ k}. Equivalently, consider conditional probabilities given that both links are still working at time k. Then the conditional probability that the network fails at time k, is the conditional probability that the first link fails at time k, plus the conditional probability that the second link fails at time k, minus the conditional probability that both links fail at time k. That is λT (k) = λ 1 (k) + λ 2 (k) − λ 1 (k)λ 2 (k). (Note: If λ 1 (k) and λ 2 (k) are both very small, then λT (k) ≈ λ 1 (k) + λ 2 (k). )

  1. Reliability of a self-healing ring communication network (a) We need to find P (X ≥ 2), where X is the number of failed links. The assumptions imply that X has the binomial distribution with parameters n = 5 and p = 0.001. Thus, using q = 1 − p, P (X ≥ 2) = 1−P (X = 0)−P (X = 1) = 1−q^5 − 5 pq^4 = 0. 00000998 ≈ 0. 000010. Another way to do this is P (X ≥ 2) = P (X = 2) + P (X = 3) + P (X = 4) + P (X = 5) = 10p^2 q^3 + 10p^3 q^2 + 5p^4 q + p^5 = 0 .00000997 + 0.00000001 + 0.000000000005 + 0. 000000000000001 (b) That is P (X = 2) = 10p^2 q^3 = 0.00000997, (which is very nearly equal to the answer to part (a)) (c) Link 1 must fail, and at least one of liks 2,3,4, or 5 must fail, which has probability p[1 − q^4 ] ≈

(d) At least one of links 1 or 2 must fail, and at least one of links 3,4, or 5 must fail, which has probabilty [1 − q^2 ][1 − q^3 ] ≈ 0. 000006. (Note, on a logarithm scale, the answers to parts (a)-(d) are all about the same.)

  1. Network reliability problem (a) P (F |F 1 F 2 ) = 1, because every directed s − t path uses either link 1 or link 2. (b) Given the event F 1 F 2 c , the original network is equivalent to the one pictured. Node a and links 1,3,4 are deleted, because they can’t be used, and nodes s and c are merged, because c can always be reached from s. Analysis of the equivalent network yields P (F |F 1 F 2 c ) = (2p − p^2 )^2 = 0. 0361. Equivalently, with q = 1 − p, P (F |F 1 F 2 c ) = (1 − q^2 )^2. Similarly, P (F |F 1 c F 2 ) = (2p − p^2 )^2 = 0. 0361.

6

(b) (^) (c)

6

7

8

b

c d

t

7

8

b

d

t

s

5 s^ a c

3

4

5

(c) Given the event F 1 c F 2 c , the original network is equivalent to the one pictured. Nodes s, a, and c are merged, because a and c can always be reached from s. Analysis of the equivalent network yields P (F |F 1 c F 2 c ) = (p^2 + p − p^3 )^2 = 0. 011881. (d) P (F ) = p^2 P (F |F 1 F 2 ) + pqP (F |F 1 F 2 c ) + pqP (F |F 1 c F 2 ) + q^2 P (F |F 1 c F 2 c ) = 0.0261. (e) The network is tolerant to single link failures, and there are only two possible double link failures that can make the network fail, namely F 1 F 2 and F 7 F 8. It is much more likely that either of these two events occurs than it is that three or more links will fail in this small network, for p = 0. 001. Thus, for p = 0.001, P (F ) ≈ P (F 1 F 2 ∪ F 7 F 8 ) ≈ P (F 1 F 2 ) + P (F 7 F 8 ) = 0. 000002.

  1. A series parallel network (a) P (F ) = P (F 1 ) + P (F 2 F 3 ) − P (F 1 F 2 F 3 ) = 0.1 + 0. 01 − 0. 001 ≈ 0. 109. Therefore,

P (F 1 |F ) =

P (F 1 F )

P (F )

P (F 1 )

P (F )

(b)

P (F 2 |F ) =

P (F 2 F )

P (F )

P (F 2 F 3 ∪ F 2 F 1 )

P (F )