Title of the Invention

“ASSESSMENT OF REDUNDANCY ALLOCATION FOR

COMPONENT SOFTWARE DEPENDABILITY AT MULTIPLE
STAGES IN A FUNCTIONAL SYSTEM”
1
Chetan Kumar Sharma, 2Rajeev Kaushik, 3Ruchi Parashar, 4Ambika
Chauhan
1
Associate Professor, Department of Mathematics, Noida International
University, G B Nagar, U.P., INDIA
2
Assistant Professor, Department of Physics, Shri K K Jain (P/G) College,
Khatauli, Muzaffar Nagar, U.P., INDIA
3,4
Research Scholar, Department of Mathematics, Noida International
University, Greater Noida, U.P., INDIA
E-mails:
1
[email protected]
2
[email protected]
3
[email protected]
4
[email protected]

ABSTRACT
In this manuscript, we analyses that only one element be added to a specific
stage that has the lowest reliability; yet, these limitations also allow for the
addition of multiple elements to a different stage that has the highest reliability.
It is possible to handle a large number of linear or nonlinear constraints with
little increase in computational effort. Due to their varied nature, components in
every complicated practical system would inevitably have nearly identical
reliability but drastically varying costs. With a multi-state constraint, the best
redundancy allocation will maximize the optimum and stage while also
minimizing unreliability parameters. If these conditions are not met, the stages
will be removed and the more pertinent parameters will be obtained again to
find the stages with the highest reliability.
KEYWORDS: Software Reliability, Redundancy Allocation, Multi-state
Constraint, Parallel series System, Mean Time to Failure
 OBJECTIVES

Dependable systems are preferred because they are “reliable” so, the main
objectives of the dependable or reliable system have the following
characteristics:
Reliability: There are very few mistakes and the system operates as it should.
Availability: There is extremely little to no downtime for the system and
services.
Safety: The systems don't present unacceptably high threats to users' health or
the environment.
Confidentiality: It is not appropriate to disclose data or other information
without purpose and consent.
Survivability: The system services ought to be resilient enough to endure
mishaps and intrusions.
Integrity: It is not acceptable to alter system data without permission and intent.
Maintainability: Hardware and service maintenance for the system shouldn't be
too expensive or complicated.

BACK GROUND OF THE INVENTION

Rapid growth of science and technology has thrown a challenge before
designers, engineers and manufactures for the reliable performance of their
system, particular in defense, biological system, industries and space research
programmers. A high degree of reliability is an absolute necessity. No one can
afford to take risk with device which does not operate particularly at an instant
when human life and national security is at stake. As if a missile misfires, the
dud might affect the entire outcome of the battle, the failure of technological
devices used in dialysis and any other medical treatment may cause of death of
a patient, a leaky brake cylinder could result in personal injury and undue
expense, a space satellite may be rendered completely useful if switch fails to
operate.

One of the major functions of reliability engineering is to achieve the desired

level of system reliability or improve the system reliability as far as possible.
The most effective and most commonly used method is structured redundancy.
Being empirically verifiable in terms of dependability is one of the keys of
trustworthy systems. This implies that fads in methodology, even those that are
highly popular, should be evaluated impartially according to their actual
efficacy. The quantity of errors is one indicator of dependability. Errors in
design or implementation that lead to malfunctions are called faults. Any non-
attainment of any of the system's functional specifications is considered a
failure. Failures can vary in severity, from trivial to catastrophic, contingent on
how they affect the system and its surrounding environment. Errors are
considered minor failures.
In certain system applications, certain reliability criteria must take precedence
over others. Correctness, for instance, is essential in banking transactions; if
correctness cannot be ensured, the transaction must be cancelled and rolled
back. The base station uses all of the signals, even those from faulty sensors,
and creates a composite image from all of the data available. In contrast, there
may be several sensors in charge of a deep-sea oil rig. If acceptable levels of
survivability, maintainability, and availability are attained within a budgetary
constraint, then 100% accuracy may be forfeited. In a similar vein, intelligence
communications require secrecy and security but may not care all that much
about delays of a few seconds or minutes.

Thus, the potential impact of the underlying problems can be used to prioritize
them. When a system exhibits lack of dependability, it indicates that one or
more of its dependability traits are deficient, which can be brought on by
malfunctions or other potential reasons for the system to fail. It is possible for
faults to appear while a system is operating. These errors are said to as active. In
the absence of this, the flaws might exist and even show themselves later. The
term "dormant" refers to these kinds of defects, and the goal of the testing
process in systems engineering is to find as many active and dormant faults as
possible prior to the tested system being deployed and used generally.

BASIC MATHEMATICAL STRUCTURE

Devices nowadays are made up of thousands of parts, subassemblies, and
components that are frequently interconnected to the point that the failure of one
could result in the system as a whole. Certain elements, sections, and
subassemblies inside the system might also be strategically important.
Technology systems and their products are becoming more and more
complicated every day. Therefore, it is now crucial to increase the effectiveness
of these complex systems in every way. One way to lower the degree of system
unreliability is to try better understanding of failures, better manufacturing
procedures, careful planning and designing of new systems, and appropriate
component selection. From high-reliability space users to low-cost ordinary
consumers, from political leaders to industrial managers, from academicians to
practical engineers, the issue of reliability looms large. Reliability requirements
for a given system or mission will depend on its intended use. It is unrealistic to
anticipate that the parts used in a TV set will be as reliable as those used in a
spacecraft or missile.

We have considered 𝑛 number of stages in a system connected in series where

stage 𝑖 is a parallel configuration of 𝑋𝑖 components each with reliability𝑃𝑖 . It is
assumed that all elements are working simultaneously and for a stage to fail all
these elements must fail. For the whole system to be operating all the stages
must be operating the system reliability is expressed by
𝑛

𝑅𝑠 = ∏ 1 − (1 − 𝑃𝑖 )𝑋𝑖
𝑖=1
𝑛

= ∏ 1 − 𝑄𝑖 𝑋𝑖
𝑖=1

The problem is to maximize 𝑅 subject to

𝑚

∑ 𝐶𝑖𝑗 (𝑋𝑗 ) ≤ 𝐾𝑗 ; 𝑗 = 1, 2, 3, … , 𝑚
𝑖=1

Notation
𝑷𝒊 : Reliability of 𝑖𝑡ℎ component
𝑄𝑖 : Unreliability of 𝑖𝑡ℎ component
𝑅𝑠 : System Reliability
𝑄𝑠 : System Unreliability
𝑋𝑖 : Number of total components
𝑛: Number of Stages
𝐾𝑗 : Available resource for constraint 𝑗
𝑚: Total number of different types of constraint
𝐶𝑖𝑗 (𝑋𝑖 ): Resourse 𝑗 consumed in stage 𝑖 with 𝑋𝑖 components connected

Probability:
Probability is defined by
 𝑁
𝑃(𝐶 ) = lim [ ]
𝑛→∞ 𝑛

Where,
𝑃(𝐶 ) is the probability of occurrence of event 𝐶
𝑁 is the number of times event 𝐶 occurs in the repeated experiments.
Some of the probability properties are as follows:
i. The probability of occurrence of an event, say A, is 𝑂 ≤ 𝑃(𝐴) ≤ 1
ii. Probability of the sample space 𝑆 is 𝑃(𝑆) = 1
iii. The probability of the negation of the sample space 𝑆 is 𝑃(𝑆̅) = 0

Where 𝑆̅ is the negation of the sample space S.

iv. The probability of occurrence and nonoccurrence of an event, say 𝐴, is
always
𝑃(𝐴) + 𝑃(𝐴̅) = 1
Where,
𝑃(𝐴) is the probability of occurrences of event 𝐴
𝑃 (𝐴̅) is the probability of nonoccurrence of event 𝐴
v. The probability of the union of 𝑛 independent events is
𝑛

𝑃 (𝐴1 + 𝐴2 + 𝐴3 + ⋯ + 𝐴𝑛 ) = 1 − ∏(1 − 𝑃(𝐴𝑖 ))

𝑖=1
vi. The probability of an intersection of n independent events is
𝑃(𝐴1𝐴2 𝐴3 … 𝐴𝑛 ) = 𝑃 (𝐴1) 𝑃 (𝐴2 ) 𝑃(𝐴3 ) … 𝑃(𝐴𝑛 )
Mean Time To Failure:
This is expressed by
∞

MTTF = E(t) = ∫ 𝑡 𝑓 (𝑡)𝑑𝑡

0

or,
∞

MTTF = ∫ 𝑅 (𝑡)𝑑𝑡
0

or,
 MTTF = lim 𝑅(𝑠)
𝑠→0

Where,
MTTF is the item's mean time to failure.
E(t) is the expected value.
𝑠 is the Laplace transform variable.
𝑅(𝑠) is the Laplace transform for the reliability function 𝑖. 𝑒., 𝑅 (𝑡)

Reliability Measurement:

The choice of reliability measures for maintained systems requires the

consideration of whether the main penalty or cost of system failures depends on

 The total duration of failures, or

 The frequency of failures

The right actions will depend on the system's availability if the overall length of
failures is significant. The system's mean up-time or mean down-time will
determine the suitable measure if the frequency of failures is significant. Since
availability only considers the overall amount of time the system is in an
acceptable state and does not reveal how this time is distributed, the
significance of the mean up-time (or mean downtime) is evident.
The factors associated with the above measures:

 The component failure process.

 The system structure,
 The maintainability and its policies, and
 The states in which the system is defined as success.

A general procedure for the reliability analysis is as follows:

 Definition of the reliability goals and criteria,

 Determination of physical model, basic component events, and repair
rate.
 Collection of component and subsystem characteristics with respect to
various stresses such as thermal, electrical, etc.
 Selection of methods and mathematical models.
 Numerical evaluation of mathematical models.
  Comparison of results and
 Development of alternative design if the reliability goals are not reached.

Some of the important methods to enhance reliability are:

 Paints improvement method,

 Effective and creative design,
 System simplification,
 Use of over-rated components,
 Structural redundancy, and
 Maintenance and repair.

The cost of various methods of achieving, reliability will vary according to the
following:

 Types of components,
 Cost of maintenance,
 Accessibility of the products for the maintenance
 Time and manpower available for the design,
 Constraints such as weight, volume, etc.

Reliability metrics include availability, mean downtime, failure frequency,

mean time to failure, and survival probability are some of the markers of system
efficacy. These metrics provide the necessary standards for comparing and
assessing different designs, allowing the system planner to select the alternative
design policy that best achieves the objectives within certain techno-economic
constraints. Apart from the dependability attributes of individual subsystem
components, the overall reliability characteristics of the system are greatly
influenced by the topological configuration and subsystem architecture. These
attributes do share certain similarities. It is comparable to the weakest link
phenomena in security, for example, in that the strength of the entire system is
determined by the weakest link in the chain. Thus, each of the aforementioned
qualities has to be present for a system or product to be considered dependable.
Conversely, the degree to which these characteristics are absent from a system
is strongly related to its unreliability. Additionally, dependability is usually not
a binary phenomenon (present or absent), but rather depending on grades and
accepted standards. These thresholds are specific to infrastructures like
electrical, electromechanical, and quantum as well as applications like data
processing, communications, and process control.

WORKING METHODOLOGY FOR REDUNDANT ALLOCATION

Most algorithms were always independent of the values of the constraints
associated with each step. Due to these limitations, only one element may be
added to a stage with the lowest reliability; on the other hand, numerous
elements may be added to a stage with the highest dependability. This situation
is particularly common when the stages are made up of nearly identical parts
that have different prices.
A component is added up to the point at which, given a factor defined as the
product of increments in constraints times the decrement in unreliability, its
addition has the highest value.
𝑷𝒊 𝑄𝑖 𝑋𝑖
𝐹𝑖 (𝑋𝑖 ) =
∏𝑚
𝑗=1 ∆𝐶𝑖𝑗 (𝑋𝑖 )

It is observed that 𝐹𝑖 (𝑋𝑖 ) is a function of 𝑖, the particular stage and 𝑋𝑖 , the

number of elements in that stage, hence in the process of computation, the value
of this factor keeps changing even for a fixed 𝑖. Therefore
𝑷𝒊 𝑄𝑖 𝑋𝑖+1
𝐹𝑖 (𝑋𝑖 + 1) = 𝑚
∏𝑗=1 ∆𝐶𝑖𝑗 (𝑋𝑖 )

In a linear constraint, it is therefore very convenient to evaluate all 𝐹𝑖 (𝑋𝑖 ) by

using recursive relation which simply requires successive multiplication by 𝑄𝑖
𝐹𝑖 (𝑋𝑖 + 1) = 𝑄𝑖 ∗ 𝐹𝑖 (𝑋𝑖 )
The procedure outline above can be summarized in the form of various steps:
1. Let 𝑋𝑖 = 1 ∀ 𝑖
2. Calculate 𝐹𝑖 (𝑋𝑖 )∀𝑖
3. Mark the stage * having highest value of stage selection factor 𝐹𝑖 (𝑋𝑖 )

DATA ANALYSIS FOR THE REDUNDANT ALLOCATION

Due to its many benefits over alternative strategies for enhancing system
reliability, redundancy is applied in system design in nearly all systems.
 Any desired level of reliability can be achieved (if the available sources
permit).
 Increasing in reliability per unit resource spent is highest when optimal
redundancy techniques are employed.
  Design through redundancy needs comparatively less skill on the part of
the designer.

Consider a four stage system for optimum redundancy allocation with two
linear constraints.
The data are:
𝒏=𝟒 𝑲𝟏 ≤ 𝟓𝟒 𝑲𝟐 ≤ 𝟏𝟏𝟗
Stage 1 𝑃1 = 0.65 𝐶11 = 1.2 𝐶12 = 7
Stage 2 𝑃2 = 0.70 𝐶21 = 2.2 𝐶22 = 6
Stage 3 𝑃3 = 0.75 𝐶31 = 3.3 𝐶32 = 5
Stage 4 𝑃4 = 0.80 𝐶41 = 4.4 𝐶42 = 4

Four stage system for optimum redundancy allocation

6-7
7
5-6
6
5 4-5
4
3-4
3
2 2-3

1 K≤ 120
1-2
0 K≤ 56
Stage 1 0-1
Stage 2 Probability
Stage 3
Stage 4

Diagram 1: Four Stage System for Optimum Allocation

There are two linear constraints in this case.
Therefore,
𝑷𝒊 𝑄𝑖 𝑋𝑖
𝐹𝑖 (𝑋𝑖 ) = ∀𝑖 = 1,2,3,4 𝑤ℎ𝑒𝑟𝑒 𝑋𝑖 = 1
𝐶𝑖 1 𝐶𝑖 2
𝑷𝒊 𝑄𝑖 𝑋𝑖
𝐹𝑖 (1) = ∀𝑖 = 1,2,3,4
𝐶𝑖 1 𝐶𝑖 2
Hence,
 𝐹1 (1) = 0.02708, 𝐹2 (1) = 0.015909, 𝐹3(1) = 0.01136, 𝐹4(1) = 0.009090

F(1)-Values

0.03

0.02

0.01

0
1 2 3 4
F-Values 0.02708 0.01591 0.01136 0.00909

Diagram 2: Two Linear Constraints Allocation

Construct a Table for Four Stages Unreliability Parameters
𝐗𝟏 𝐗𝟐 𝐗𝟑 𝐗𝟒 𝐅𝟏 (𝐗 𝐢 ) 𝐅𝟐 (𝐗 𝐢 ) 𝐅𝟑 (𝐗 𝐢 ) 𝐅𝟒 (𝐗 𝐢 ) ∑ 𝐗 𝐂 𝑭𝒊 ( 𝑿 𝒊
𝐢 𝐢 𝟏 ∑ 𝐗 𝐢 𝐂𝐢 𝟐
+ 𝟏)

1 1 1 1 2.708* 1.590 1.136 0.909 11.1 22 0.948

2 1 1 1 0.948 1.590* 1.136 0.909 12.3 29 0.477
2 2 2 1 0.948 0.477 1.136* 0.909 14.5 35 0.284
2 2 2 1 0.948* 0.477 0.284 0.909 17.8 40 0.332
3 2 2 1 0.332 0.477 0.284 0.909* 19 47 0.182
3 2 2 2 0.332 0.477* 0.284 0.182 23.4 51 0.143
3 3 2 2 0.332* 0.143 0.284 0.182 25.6 57 0.116
4 3 2 2 0.116 0.143 0.284* 0.182 26.8 64 0.071
4 3 3 2 0.116 0.143 0.071 0.182* 30.1 69 0.036
4 3 3 3 0.116 0.143* 0.071 0.036 34.5 73 0.043
4 4 3 3 0.116* 0.043 0.071 0.036 36.7 79 0.040
5 4 3 3 0.040 0.043 0.071* 0.036 37.9 86 0.018
5 4 4 3 0.040 0.043* 0.018 0.036 41.2 91 0.013
5 5 4 3 0.040* 0.013 0.018 0.036 43.4 97 0.014
6 5 4 3 0.014 0.013 0.018 0.036* 44.6 104 0.007
6 5 4 4 0.014 0.013 0.018* 0.007 49 108 0.004
6 5 5 4 0.014* 0.013 0.004 0.007 52.3 113 0.005
 7 5 5 4 0.005 0.013* 0.004 0.007 53.5 120 0.003
7 6 5 4 0.005 0.003 0.004 0.007* 55.7 126 0.001
7 6 5 5 0.005 0.003 0.004 0.001 60.1 130

Mark the Stage (*) having highest value of stage selection factor. A redundant
component proposed to be added to that stage.
For a system having 𝑛 stages in series with 𝑋𝑖 redundant components at stage-𝑖,
the system unreliability is given by
𝑛

𝑄 = 1 − ∏(1 − 𝑄𝑖 )𝑋𝑖
𝑖=1

This can be approximated as,

𝑛

𝑄 ≅ ∑ 𝑄𝑖 𝑋𝑖
𝑖=1

The problem is to maximize 𝑅 subject to

𝑛

∑ 𝐶𝑖𝑗 (𝑋𝑗 ) ≤ 𝐾𝑗 ; 𝑗 = 1, 2, 3, … , 𝑚
𝑖=1

and,
𝑛

𝑅 = ∏ (1 − 𝑄𝑖 𝑋𝑖 )
𝑖=1

𝑅 = (1 − 𝑄1 𝑋1 )(1 − 𝑄2 𝑋2 )(1 − 𝑄3 𝑋3 )(1 − 𝑄4 𝑋4 )

= (1 − 0.035)(1 − 0.018)(1 − 0.020)(1 − 0.005) ≅ 0.9240340
 Reliable Area

1
0.995
0.99
0.985
0.98
0.975
0.97
0.965
0.96
0.955
0.95
1st reliabile Area 2nd reliabile Area 3rd reliabile Area 4th reliabile Area

Diagram 3: Optimum Solution for Four Stages Reliable Area

Therefore, optimum solution is 𝑋 = [7 6 5 5] and 𝑅 ≅ 0.9240340

CLAIM
1. A maximum optimum and a minimized unreliability parameter are
provided by the stage in the optimal redundancy allocation with multi-
state constraints.

Conflicts of Interest
On the behalf of all inventors, the First inventor states that there are no conflicts
of interest.