08s Cpe633 Test1 Solution

The University of Alabama in Huntsville
Electrical and Computer Engineering

CPE 633 01
Test 1 Solution
Spring 2008
1. (30 points) For the system diagram shown:
(a) Derive the upper bound for system reliability.
(b) Derive the lower bound for system reliability.
(c) Derive the exact reliability formula.
(d) If RA = RB = RC = RD = RE = RF = R = e- t, find the MTTF for the system.
E
C
A
F
D
(a) The paths in this system are AE, ACF, ADF, and BF.
Rsystem 1 (1 R pathi )
Rsystem 1 (1 RARE)(1 RARCRF)(1 RARDRF)(1 RBRF)

(b) The minimum cut sets in this system are AB, EF, BCDE, and AF.
Rsystem 1
(1 Qcuti ) , where Qcut i is the probability that minimal cut set i is faulty
Rsystem 1 [(1- (1 RA)(1 RB))(1 (1 RE)(1 RF)) (1 (1 RA)(1 RF))

(1 (1 RB)(1 RC)(1 RD)(1 RE))]
(c) Expanding around F, Rsystem = RF(Rsystem | F working) + (1 RF)(Rsystem | F not working)
Rsystem | F working = (1 (1- RA[1 (1 RE)(1 RC)(1 RD)])(1 RB))
= (1 (1- RA[1 (1 RE RC + RCRE)(1 RD)])(1 RB))
= (1 (1- RA[1 (1 RE RC + RCRE RD + RDRE + RCRD - RCRDRE)])(1 RB))
= (1 (1- RA[ RE + RC - RCRE + RD RDRE - RCRD + RCRDRE)])(1 RB))
= (1 (1- RARE - RARC + RARCRE - RARD + RARDRE + RARCRD - RARCRDRE)(1 RB))
= (1 (1- RARE - RARC + RARCRE - RARD + RARDRE + RARCRD - RARCRDRE - RB
+ RARBRE + RARBRC - RARBRCRE + RARBRD RARBRDRE RARBRCRD + RARBRCRDRE))
= RARE + RARC - RARCRE + RARD - RARDRE - RARCRD + RARCRDRE + RB
- RARBRE - RARBRC + RARBRCRE - RARBRD + RARBRDRE + RARBRCRD - RARBRCRDRE
Rsystem | F not working = RARE
Rsystem = RF(RARE + RARC - RARCRE + RARD - RARDRE - RARCRD + RARCRDRE + RB - RARBRE
- RARBRC + RARBRCRE - RARBRD + RARBRDRE + RARBRCRD - RARBRCRDRE) + (1
RF)RARE
(d) Rsystem = -R6 + 4R5 6R4 + 2R3 + 2R2
MTTFsystem = Rsystem(t )dt = e 6t + 4e 5t 6e 4 t + 2e 3t + 2e 2t
e 6t 4e 5t 6e 4t 2e 3t 2e 2t
=
+
+
+
+
4
3
2
6
5
4
3
2 1
1
= [0 0 + 0 0 0]
+

6 5 2 3
1 4
3
2 1 5 + 24 45 + 20 + 30
24
4
=
+
+
+ =
=
=
6 5 2 3
30
30 5
2. (20 points) Consider the processor/memory configuration shown below. List the conditions under
which it will fail,and compare them to a straightforward TMR configuration in which each unit
consists of a processor and a memory. Denote by Rp, Rm , and Rv, the reliability of a processor, a
memory, and a voter, respectively, and write expressions for the reliability of the two TMR
configurations.
The conditions under which this configuration would fail are

2 or 3 processors fail, 2 or 3 memories fail, 2 or 3 voters fail
Each of these units follows the TMR
3
R2 _ of _ 3 =
i=2
( )R (t )(1 R(t ))
i
3
i
3i
( )R
3
2
(t ) (1 R (t )) 3 2 +
( )R
3
3
(t )(1 R (t )) 33
3! 3
3! 2
R (t )(1 R (t )) +
R (t ) = 3 R 2 (t )(1 R (t )) + R 3 (t ) = 2 R 3 (t ) + 3 R 2 (t )
3!0!
2!1!
][
Rsystem = 2 R p (t ) + 3R p (t ) 2 Rm (t ) + 3Rm (t ) 2 Rv (t ) + 3Rv (t )

3
With each unit having a processor and memory and having one voter.
Rsystem = Rv 2(R p (t ) Rm (t ) ) + 3(R p (t ) Rm (t ) ) (t )

3
3. (15 points) Derive all codewords for the separable 6-bit cyclic code based on the generating
polynomial X3 + X2 + 1.
1
1101 001000
1101
101
11
1101 010000
1101
1010
1101
111
10
1101 011000
1101
010
110
1101 101000
1101
1110
1101
110
100
1101 110000
1101
100
11
1101 111000
1101
1100
1101
001
1110
1101 1000000
1101
1010
1101
1110
1101
010
Codewords: 000000, 001101, 010111, 011010, 100010, 101110, 110100, 111001
4. (20 points) A communication channel has a probability of 10-4 that a bit transmitted is erroneous. The
data rate is 6000 bits per second (bps). Data packets contain 148 information bits, a 16-bit CRC for
error detection, and 0, 8, or 16 bits for error correction coding (ECC). Assume that if 8 ECC bits are
added all single bit errors can be corrected, and if 16 ECC bits are added all double bit errors can be
corrected.
(a) Find the throughput in information bits per second of a scheme consisting of error detection with
retransmission of bad packets (i.e., no error correction).
(b) Find the throughput if 8 ECC check bits are used, so that single bit errors can be corrected.
Uncorrectable packets must be retransmitted.
(c) Finally find the throughput if 16 ECC check bits are appended, so that two bit errors can be
corrected. As in (b), uncorrectable packets must be retransmitted. Would you recommend
increasing the number of ECC check bits from 8 to 16?
(a) Each packet contains 164 bits. If any error occurs, it is detected (assuming that the CRC always
works) and the packet is discarded. The probability that a packet has no errors is
(164!/(164!0!))(1 - 10-4)164 = 0.9837. The data rate of the code is 148/164 = 0.9024. Thus, the
throughput in bits per second is 0.9837 *0.9024 * 6000 = 5326
(b) With the addition of 8 ECC check bits, each packet contains 172 bits. The probability that a
packet has at most one error is (172!(172!0!))(1 - 10-4)172 + (172!/(171!1!)) * 10-4(1 - 10-4)171 =
0.9994. The rate of the code is now 148/172 = 0.8430. Thus, the throughput with single bit error
correction is 0.9994 * 0.843 * 6000 = 5055.
(c) The second ECC byte increases the packet size to 180. The probability that a packet of 180 bits
has no more than two errors is (180!(180!0!))(1 - 10-4)180 + (180!/(179!1!))(10-4(1 - 10-4)179 +
(180!(178!2!))(10-4)2(1 - 10-4)178 = 0.. The code rate is now 148/180 = 0.8222, so the throughput
is 0.9999 * 0.8222 * 6000 = 4933. Increasing the error correction capability in this case resulted
in a reduction in the throughput, so no.
5. (15 points) Consider two computers A and B. (a) Assuming an exponential distribution, what is the
probability that at least one will survive 10,000 hours if their failure rate is 1 failure per million
hours? (b) What is the probability that both will survive 10,000 hours? (c) What is the probability that
A will survive 15,000 hours if it survived 5,000 hours?
(a) P{T A > 10,000 TB > 10,000} = 1 P{T A < 10,000 TB < 10,000}
= 1 P{T A < 10,000 TB < 10,000} = 1 ( FA(10000) FB(10000))
= 1 (1 R A (10000))(1 RB (10000)) = 1 (1 2 R(10000) + R 2 (10000))

= 2 R(10000) R 2 (10000)) = 2e 10
(10000 )
e 2*10
(10000 )
= 0.9999
(b) P{T A > 10,000 TB > 10,000} = P{T A > 10,000}P{T A > 10,000}
6
= R A (10,000) R B (10,000) = e 10 (10000 ) e 10 (10000 ) = 0.9802

P{T A > 15,000 T A > 5000} P{T A > 15000}
(c) P{T A > 15,000 | T A > 5000} =
=
P{T A > 5000}
R A (5000)
R A (15000)
e 10
( 5000 )
e 10
e 10
(15000 )
( 5000 )
= 0.9048

08s Cpe633 Test1 Solution

Uploaded by

Copyright:

Available Formats

08s Cpe633 Test1 Solution

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

08s Cpe633 Test1 Solution

Uploaded by

Copyright:

Available Formats

The University of Alabama in Huntsville

Electrical and Computer Engineering

Rsystem 1 (1 RARE)(1 RARCRF)(1 RARDRF)(1 RBRF)

Rsystem 1 [(1- (1 RA)(1 RB))(1 (1 RE)(1 RF)) (1 (1 RA)(1 RF))

MTTFsystem = Rsystem(t )dt = e 6t + 4e 5t 6e 4 t + 2e 3t + 2e 2t

The conditions under which this configuration would fail are

Rsystem = 2 R p (t ) + 3R p (t ) 2 Rm (t ) + 3Rm (t ) 2 Rv (t ) + 3Rv (t )

Rsystem = Rv 2(R p (t ) Rm (t ) ) + 3(R p (t ) Rm (t ) ) (t )

Codewords: 000000, 001101, 010111, 011010, 100010, 101110, 110100, 111001

= 1 P{T A < 10,000 TB < 10,000} = 1 ( FA(10000) FB(10000))

= 1 (1 R A (10000))(1 RB (10000)) = 1 (1 2 R(10000) + R 2 (10000))

= R A (10,000) R B (10,000) = e 10 (10000 ) e 10 (10000 ) = 0.9802

You might also like