Advanced Computing: An International Journal (ACIJ), Vol.15, No.

6, November 2024

METHODS AND ALGORITHMS FOR ASSESSING

COMPUTER NETWORK PERFORMANCE
Aziz Ishmukhamedov

Department of Software of Information Technologies, Tashkent University of

Information Technologies named after Muhammad al-Khwarizmi, Tashkent, Uzbekistan

ABSTRACT
The article proposes an approach that makes it possible to obtain an assessment of the reliability of a data
transmission system, containing a description of computational circuits, algorithms, and models for
assessing various aspects of reliability: the method for calculating operability is presented for the first
time, although it has already become widespread in the design of special-purpose systems.

KEYWORDS
Reliability assessment, operability, system reliability, computational schemes, algorithms, evaluation
model.

1. INTRODUCTION
Quite often, in the works [1,3,4,5,6], to determine the reliability of computer systems, as
properties to provide communication, while maintaining the values of established quality
indicators in time under specified operating conditions, not only and not so much the
characteristics of the condition of technical means, their processing for failure, but also the
number of user requirements for message reliability indicators, the probability of timely delivery
began to be included messages, i.e. due to a wide range of user requirements for reliability
indicators. taking into account all the stated requirements at the design stage of computer systems
can be implemented by modeling the functioning of the system taking into account failures –
restoration and evaluation of the technical condition of the system identified with network
connectivity (structural reliability). The connectivity property does not provide the necessary
degree of detail about all possible network states at the current time, allows you to operate only in
two categories (whether the connectivity requirements are met or not), without determining how
“badly” connectivity is broken in case of non-fulfillment of the specified requirements. This is
necessary for a more detailed assessment of the state of the computer network. The step-by-step
significance of the task of determining the technical condition of the network is increasing.
Almost until recently, the definition of “bottlenecks” of the network, in which there is a high
probability of failure of elements that will worsen its functioning or lead to a complete failure of
the system, was based on the well-known provisions of the theory of reliability, including the
concept of a complete failure of the system in case of failure. At the same time, even in the
presence of equipment failure or disconnection of part of the system elements, it is possible to
transmit and receive information. Known methods of assessing the reliability of computer
systems are not taken into account. The step-by-step introduction [2] of conceptual definitions of
the technical condition of objects and systems, as well as corresponding quantitative indicators,
made it possible to correctly solve a large number of network operation tasks.

DOI:10.5121/acij.2024.15601 1
 Advanced Computing: An International Journal (ACIJ), Vol.15, No.6, November 2024

2. INDICATORS TO EVALUATE NETWORK STATE.

The connectivity property does not provide the necessary degree of detail about all possible
network states at the current time, since it allows you to operate only two categories (whether the
connectivity requirement is met or not), does not determine how “severely” connectivity is
disrupted if the specified requirements are not met. Given this, it is necessary to have a more
detailed assessment of the state of the computer network. Therefore, the importance of the task of
determining the technical condition of the network is increasing. Almost until recently, the
definition of “bottlenecks” in the network, in which there is a high probability of failure of
elements that will worsen its functioning or lead to complete failure, was based on well-known
provisions of reliability theory, including the concept of complete system failure when its
elements fail. However, even in the presence of equipment failures or disconnection of part of the
system elements, information can be transmitted and received. Well-known methods for
evaluating the reliability of computer systems do not take these features into account. Therefore,
the introduction in [2] of conceptual definitions of the technical conditions of the object and
system, as well as the corresponding quantitative indicators, made it possible to correctly solve a
large number of network operation tasks. When designing and operating a computer network, it is
necessary to determine the state of the network and its elements, so that, on the one hand, to
assess the state of the network from the position of an external observer (maintenance personnel),
and on the other hand, to have more detailed information for operational management of network
resources. The methodological concept of the state of complex systems is predominant today. The
concept of state plays an important role in describing the changes that occur in the network and
its elements over time. To assess the reliability of the network and its elements, the following
alternative states are mainly used: serviceability, malfunction or operability, inoperable. In certain
stages of the network life cycle, mere observation is insufficient. Therefore, a comprehensive
evaluation of serviceability, or specifically network operability, becomes essential. A
fundamental concept introduced to gauge the technical condition of the system is the notion of
personnel competence and overall network performance as perceived by maintenance personnel.
This concept facilitates the development of visual, quantitative representations over time,
illustrating the operability level throughout network operation. Consequently, it enables
meticulous decision-making regarding maintenance structure. Recently, much attention has been
paid to the development and implementation of rational strategies for technical operation as an
effective means of ensuring the required values of reliability indicators of complex systems. As
such an indicator for assessing the technical condition of the network, the work [2] introduced the
indicator 𝛼𝑘𝑖𝑗 – the level of network operability, which allows determining the degree of network
operability from the position of the i–th user when interacting with the j–th user. Developing this
approach, this section proposes a method for evaluating the network performance as a whole, i.e.
from the position of the service personnel. Let's consider a more detailed method of assessing the
operability of the communication network from the user's position, described in [2]. Each user is
interested in the fact that the computer network meets his needs for the delivery of messages in a
certain mode, with specified reliability indicators, the main of which are: the probability of timely
delivery of messages, message delivery time and reliability. As a rule, the user is interested in the
operability or inactivity of the network. In some cases, it is important to know the health of the
network or the quality of its functioning at a given time. The definition of the specific state of the
network is delineated from the perspective of the i-th user linked with the j-th user (where i,j
=1,2,...,N, N being the number of users). Network serviceability, denoted as 𝑍𝑢𝑖𝑗 , pertains to the
network state characterized by the existence of all m potential paths 𝑃𝑖𝑗 between the i-th and j-th
users, signifying compatibility. This is mathematically represented by the equation:

𝑞
𝑊𝑖𝑗 = ⋃𝑚
𝑞−1 𝑃𝑖𝑗 , (1)

2
 Advanced Computing: An International Journal (ACIJ), Vol.15, No.6, November 2024

Where m represents the number of paths. Network nonlinearity, 𝑍𝑢𝑖𝑗

̅ , refers to a state wherein at
𝑟
least one connection 𝑆𝑖𝑗 (𝑟 = 1,2, … , 𝑟 − denoting the number of regulated message delivery
modes) is present. Network operability, 𝑍𝑢𝑖𝑗
̅ , denotes a state characterized by the presence of at
least one data transmission path between 𝑃𝑖𝑗 the i-th and j-th users. Conversely, network
𝑖𝑗
inoperability, 𝑍𝑢̅ , describes a state where the last f-th (where f=1,2,…q) of all q (where q=1,2
,…m) possible paths between the i-th and j-th users fail. The event leading to the transition from
ij ij
state Zp state Zp is termed a network failure 𝐻𝑂𝑖𝑗 . The primary functional objective of the
𝑖𝑗
network, denoted as 𝑍𝑃𝐹 , is to ensure timely and error-free message delivery between the i-th and
𝑖𝑗
j-th users, under specific time and mode conditions. Network malfunction, 𝑍𝑃𝐹 , denotes a
condition where the network achieves timely and error-free message delivery in the designated
𝑖𝑗 𝑖𝑗
mode. The event leading to the transition from the state 𝑍𝑃𝐹 to the state 𝑍𝑃𝐹 will be called a
𝑖𝑗
violation of the network 𝐻𝐻̅ . For any user, as noted above, the greatest interest is finding the
network in the 𝑍𝑝С state, which is a necessary and sufficient condition for ensuring normal
functioning in all modes, that is, for analyzing the visibility of the network state from now on.
positions, i.e. classification of network states into two classes of states: operability and network
inactivity. 𝑍 𝐶 ∊ 𝑍𝑝С 𝑜𝑟 𝑍 𝐶 ∊ 𝑍𝑝̅С and

𝑍𝑝С ⋂𝑍𝑝̅С = 0 (2).

As highlighted earlier, the paramount concern for any user is to ascertain the network's state as
𝑍𝑝С which serves as both a necessary and sufficient condition for ensuring seamless operation
across all modes. This entails analyzing the network's visibility from the perspective of
categorizing its states into two distinct classes: operability and network inactivity. The listed
states can be described using Boolean variables. Each state of the 𝑃𝑖𝑗 path corresponds to a
𝑃
Boolean variable 𝑍𝑝 𝑖𝑗 , whose valid implementations are 1 or 0, i.e.

𝑃𝑖𝑗 1, if the path is free,

𝑍𝑝 = { (3).
0, in the opposite case).

In turn, the state of each path is determined by the state of the network elements entering it. Each
i-th element of the network corresponds to a Boolean variable 𝑃𝑖𝑗 , valid implementations of
which are 1 or 0, i.e.

1, if the n − th element is operable,

𝑍𝑝𝑛 = { (4).
0, in the opposite case).

That state of the network is uniquely determined through the known state of its elements forming
𝑖𝑗
this path. To provide a more nuanced assessment of the network state, the indicator 𝛼𝑘 (𝑡) is
introduced. This indicator represents the level of network operability at a given time, enabling the
determination of operability from the perspective of the i-th user interacting with the j-th users. It
is defined as follows:
П𝑙𝑖𝑗
𝛼𝑘𝑖𝑗 (𝑡) = 𝑚
. (5).

Where: k is the index of the level of working capacity, m is the number of possible paths between
the i-th and j-th users.𝑃𝑖𝑗𝑙 , 𝑙(𝑙 = 1,2, … , 𝑚) – the number of operable paths existing at a given
time between the i–th and j-th corresponding users. For any moment of time t ≤ 0, the state of the

3
 Advanced Computing: An International Journal (ACIJ), Vol.15, No.6, November 2024
network Z(t) is interpreted as a random variable. The progression of the network state over time
can be construed as a stochastic process {Z(t), t=0} characterized by a finite set of states.
Consequently, the quantitative assessment of network health from the perspective of two users at
any given time can be computed using equation (5). Serviceability is the condition in which the
object meets all the established requirements. Malfunction is a condition in which an object does
not meet at least one of the established requirements. Operability denotes a state where an object
can execute predetermined functions while adhering to specified parameter values and
established limits. Conversely, inoperability occurs when at least one specified parameter fails to
meet requirements despite the object's ability to perform designated functions. Proper functioning
describes a state wherein an object fulfills all regulated functions required at the current time,
while maintaining parameter values within predefined limits. On the other hand, improper
functioning arises when an object fails to perform some regulated functions necessary at the
present time and/or does not maintain specified parameter values within established limits. Based
on these technical condition definitions, it's evident that: - In a state of serviceability, the object
remains operational. - In a state of operability, the object functions correctly across all modes. -
In a state of improper functioning, the object is inoperable and defective. Moreover, a properly
functioning object may still be inoperable and defective. A functional object may also be
defective. Let's consider the events leading to the transition from one technical condition to
another. The event (H) that leads to the transition of an object: - from Zu to Zu̅ is called damage
(Нп); - from 𝑍р to 𝑍р̅ is called damage (Но); - from ZPF to ZPF ̅̅̅̅ is called damage (Нн). The
transition process: - from Zu to Zu̅ is called the “Restoration of serviceability” of Bu; - from
Zр to Zр̅ we will call the “Restoration of operability” of Вр; - from Zр to Zр̅ we call “Restoration
of proper functioning” Вpf. To determine the types of technical condition of the communication
network, based on the general technical concepts of the technical conditions of the object,
secondary concepts are introduced, such as: - network resource, a set of tools necessary to
perform one of the network functions; - single resource, the amount of resource determined by
the minimum amount of function performed for a given system (for example, a single amount of
memory, a single bandwidth); a single connection, denoted as 𝑆𝑖𝑗 (𝑖, 𝑗 = 1,2, … , 𝑁; 𝑖 ≠ 𝑗), exists
within the network framework, where N represents the total number of users. This connection
comprises a sequential arrangement of individual network resources proficient in executing all
tasks associated with the data delivery process between the i th and j th users, characterized by
p
unified quality indicators. - connection Sij , the minimum set of single connections capable of
delivering data from i and j to the user with the specified destination indicators for the p–th mode
((p = ̅̅̅̅̅
1, p, where p – is the number of delivery modes). The connection is characterized by the
following parameters: a) length:

𝑝
𝑙(𝑆𝑖𝑗 ) = ∑𝑘𝑞=1 𝑙 (𝑈
⏟ 𝑞 ). (6).
𝑝
𝑈𝜖𝑆𝑖𝑗

where l(U) – is the length of the communication line between the plate nodes; K - the number of
communication lines encompassed within this connection. b) Time of existence:

p p p p
t (Sij ) = t yer (Sij ) + t coxp (Sij ) + t зав (Sij ). (7).

p
c) Bandwidth: μ (Sij ); d) priority of service; e) broadcasting capability (multi-targeting); f) the
discreteness of the input; g) reliability – the probability of implementation and the probability of
connection restoration. Path Пij (or a set of connections), the minimum set of network resources
that allows you to organize several connections, in any necessary combination of their types
between the i-th and j-th users, i.e.

4
 Advanced Computing: An International Journal (ACIJ), Vol.15, No.6, November 2024

𝑃𝑖𝑗 = ⋃𝑝𝑝=1 𝑆𝑖𝑗

𝑝
. (8).

Note that the path is characterized by the same parameters as the connection. Due to the ultimate
reliability of network resources, as well as due to the need to ensure a given probability of timely
delivery by i - th and j – th users, in addition to the main path, it provides backup paths that
together make up a set of paths. The set of paths Wij is the set of all existing or possible paths
between the i-th and j-th users.
𝑞
𝑊𝑖𝑗 = ⋃𝑚
𝑞=1 П𝑖𝑗 . (9).

where m-is the number of paths. Wij is characterized by the number of paths and the probability
of the existence of at least one workable path between the i-th and j-th users. Then the
communication network can be represented as L sets of paths, where L-is the number of user
pairs. The minimum number of independent paths between any two users is referred to as
network connectivity, denoted as

ℎ = 𝑚𝑖𝑛𝑖𝑗∈𝑙 𝑊𝑖𝑗 . (10).

The serviceability of the set of paths Wij is characterized by the presence of all operable Pij .
Failure of any 𝑃𝑖𝑗 leads to a transition to the fault state. The operability of 𝑊𝑖𝑗 is characterized by
the presence of at least one operable 𝑃𝑖𝑗 , with a partial failure of which 𝑊𝑖𝑗 goes into a state of
inactivity, and with a full one – into a limiting state. The operability of 𝑃𝑖𝑗 is characterized by the
presence of all operable connections. Failure of any connection leads to the transition of 𝑃𝑖𝑗 to the
state of inactivity. The failure of the last of the existing path connections (complete path failure)
p
leads to the transition of 𝑃𝑖𝑗 to the limit state. The operability of Sij is characterized by the
presence of all single connections, the failure of any of them leads to a transition to a state of
inactivity. The definitions of the types of technical conditions of the object discussed above and
the introduction of secondary concepts allow us to define the types of technical conditions of a
computer network. Let's define the technical conditions of the network from the perspective of
the i-th user corresponding with the j-th user (where i, j=1,2,…, N; i≠j), where N-is the number of
users. Network health, denoted Zuc , represents the state of the network characterized by the
presence of all m possible paths 𝑃𝑖𝑗 between the i-th and j-th users, depicted as a set of paths

𝑞
𝑊𝑖𝑗 = ⋃𝑚
𝑞=1 𝑃𝑖𝑗 . (11).

Network malfunction Zuc – is a network condition in which at least one connection is inoperable.
The occurrence that triggers the shift from the state of Zuc to the state of Zuc̅ is termed network
damage, denoted as Нcп. Network operability, represented as 𝑍𝑢𝑐 , indicates the state of the network
characterized by the existence of at least one path 𝑃𝑖𝑗 between the i-th and j-th users. Network
inactivity Zрc is a network condition in which the last 𝑃𝑖𝑗 path of q (q = 1,2, … , m) possible paths
𝑐
between the i-th and j-th users fails. The event causing the transition from the state of 𝑍𝑃𝐹 to the
𝑐 c
state of 𝑍̅̅̅̅
𝑃𝐹 is termed a network failure, denoted as Н н . The state of proper functioning over the
established connection, referred to as Zрc 0, signifies the state of the network where data delivery
c
is accurately and promptly guaranteed. The condition of improper functioning Z̅̅̅̅ PF is a network
condition in which error-free and/or timely data delivery is not provided. The event leading to the
𝑐 𝑐 c
transition from the state of 𝑍𝑃𝐹 to the state 𝑍𝑃𝐹
̅̅̅̅ will be called a violation of Нн . Since for any
user, as already noted above, the greatest interest is finding the network in the state Zрc 0, which is
a necessary and sufficient condition to ensure proper functioning in all modes, we will analyze

5
 Advanced Computing: An International Journal (ACIJ), Vol.15, No.6, November 2024
the type of technical condition of the network from these positions, i.e. classify the network states
into two subsets of states network operability or inactivity

ij
𝑍 𝑖𝑗 𝜖 𝑍𝑝𝑖𝑗 or 𝑍 𝑖𝑗 𝜖 𝑍𝑝̅𝑖𝑗 and Zij ⋂ Zp̅ = ∅. (12).

The network's state is uniquely determined by the states of the set of paths (Wij ), and the state of
Wij , in turn, is determined by the states of the paths between the corresponding i-th and j-th users.
The considered basic concepts of the computer system operation process allow us to draw the
following conclusions: - to formulate the task of general and technical operation of a computer
system in a specific formulation; - identify user classes depending on the set of required delivery
modes; - identify the necessary computer system resources to meet the needs of users; -to develop
an algorithm for managing computer system resources.

3. ALGORITHM FOR EVALUATING THE OPERABILITY OF A COMPUTER

NETWORK.
There are various models and methods for calculating the performance characteristics of a
computer system. In most cases, simulation models are used. However, the use of such models to
calculate various networks with acceptable accuracy requires significant machine time.
Analytical methods are more economical than simulation models in terms of machine time costs,
but they do not always fully describe the real process, since they have to introduce strong
simplifying assumptions, and thus make it convenient for an analytical solution. Different
mathematical methodologies are employed to construct models aimed at computing the
performance characteristics of a computer system. A special role is assigned to statistical models
of the queuing system, which have a number of advantages compared to other models. The flow
of messages transmitted over a computer network usually has a statistical character, i.e. the time
intervals between the receipt of messages are random variables with the specified distribution
laws. The order of service for incoming messages is assumed to be preset. So, if the transmitted
messages belong to different categories of urgency, priority service disciplines are usually used in
such systems. he operational model of such a system can be depicted as a multi-phase queuing
system, where each phase comprises a multi-channel priority queuing system with expected
values. The process of passing a message through such a network is modeled in phases with the
preservation and transfer from phase to phase of all the characteristics of each priority message.
Let's define a computer network in the form of a directed graph G (v, u), where the vertices V
𝑉 = {𝑣𝑖 }, 𝑖 = 1,2, … , 𝑁; correspond to the nodes of the computer network, and the arcs
U={u_j},j=1,2,…,S; represent the communication lines. The laws of propagation of incoming
message streams entering the system and circulating in it are assumed to be Poisson, the duration
of message service is determined using an exponential distribution law with the same parameter
for all streams. For the described computer network, which has a hierarchical structure, it is
necessary to develop a model for evaluating the functional characteristics of reliability. When
constructing a mathematical model, the functional features of the computer network under
consideration should be taken into account: - priority flows; - different communication channel
performance; - failure rate and channel recovery. The evaluation of delivery characteristics
involves modeling a specific direction of information flow using a multi-phase queuing system
model. Each phase of this model is depicted as a multi-channel queuing system of the 𝑀 ⃗⃗ 𝑘 /𝑀/𝑛/
𝑟 < ∞, type, which is elaborated upon in this work. Using this model, it is also possible to
calculate the characteristics of a network queuing system, such as the probability of timely
maintenance, performance, stay time, waiting time, the number of messages (requests) in the
system, etc. Let's describe the implementation of some computer network functioning processes
in the model. Each message entering the computer network is characterized by the time of receipt

6
 Advanced Computing: An International Journal (ACIJ), Vol.15, No.6, November 2024
and priority. At a random moment in time, the system receives messages (requests) for
maintenance. Since the input stream is Poisson, the time of receipt of the k - th message is
determined recursively by the following equation: The equation

𝑡𝑘 = 𝑡𝑘−1 + (−1)(1⁄𝜆вх ) ln 𝜉𝑘 . (13).

describes the iterative process where: 𝜆вх represents the intensity of message reception; 𝜉𝑘 - are
random numbers uniformly distributed in the interval (0,1). At the input of the system, 𝑞∗ (𝑞 =
1,2, … , 𝑞∗ ) incoming message streams are received. Messages are unequal in importance:
messages from the q-th stream are more important than messages from the 𝑞 + 1-st stream, and
the latter, in turn, are more important than messages from the 𝑞 + 2-nd stream, etc. (note that all
messages in the stream are equivalent). Messages are “lined up” according to priorities and in the
order of receipt. The affiliation of each message to a particular priority is determined randomly
from the proportion of messages of each priority in the total stream. The messages are serviced
by 𝑆 ∗ uniform channels, respectively numbered (1,2,3, … , 𝑆, … , 𝑆 ∗ ).. The system serves incoming
messages in order of importance, so that messages with the lowest priority number from among
those in the queue always arrive in the vacant channel. If a low priority request has been received
for maintenance in any channel, then its maintenance continues to the end even if messages of
higher priorities are received by the system during its maintenance, i.e. there is a service with
relative priority (without interruption). The transmission rate of the message obeys the
exponential distribution law with a value of

𝜇обс , i.e. 𝜇𝑖∗ = −𝜇обс · ln 𝜉𝑖 . (14).

The considered model takes into account the failure (failure) and restoration of service channels.
Each channel has an inherent uptime, which is a random variable with a corresponding
distribution law. The duration of channel recovery after failure is also a random variable
governed by a specified distribution law. If the channel fails at the moment when another
message is being serviced in it, then this message is returned to the beginning of the queue of the
appropriate priority. Simultaneously, it's possible for the same message to enter both the service
area and the waiting area repeatedly. The model takes into account the following message service
failures: - due to queue overflow; - due to exceeding the waiting time for maintenance. Given the
significance of the system time advancement mechanism in constructing such models, let's
explore potential methods for establishing system time. The model's operation should occur in
artificial time, guaranteeing event occurrence in the correct sequence and with appropriate time
intervals between them. As events may occur simultaneously in various parts of a real system, it's
essential to develop a time-setting mechanism to synchronize the actions of system components
within a specified time interval. There are two main methods for setting time: - the fixed time
step method, in which the system time is counted at predetermined time intervals of constant
length; - the method of the step to the next event. When using which, the state of the simulated
system is updated with the occurrence of each significant event, regardless of the time intervals
between them. Each of these methods has its advantages and disadvantages. For instance, the
method of advancing to the next event eliminates the need to designate an arbitrary artificial time
increment. This avoids the danger that the time increment value selected without the user's
knowledge will change the simulation results. In addition, in this case, events are considered and
served as simultaneous only if they are marked with the same time of occurrence. On the other
hand, the fixed-step method works better if many events occur during the simulation cycle, and
the mathematical expectation of the duration of events is low, as well as when the exact nature of
significant events is not clear, as, for example, it happens at the initial stage of the study. When
constructing a model of a computer network's operation, the approach of stepping to the next
event is chosen. This is mainly due to the fact that more accurate results are obtained, and there is
also no need to determine the magnitude of the time increment. The following significant events
7
 Advanced Computing: An International Journal (ACIJ), Vol.15, No.6, November 2024
are selected in the model: the receipt of a message into the system and the release of
communication channels. The promotion of the system time T in the model is performed as
follows. At the initiation of the simulation, the variable T is initialized to zero, denoted as T=0.
Subsequently, the time of arrival for the first message is generated, and T is updated accordingly
1
at this juncture. 𝑇 = 𝑡пост . All subsequent promotions of T depend on checking the condition:
which of the two nearest events – the receipt of a message or the release of a data channel, will
happen earlier, i.e.

𝑇 = min(𝑡пост , 𝑡осв ). (15).

Let's consider an algorithm for evaluating network performance in terms of correlating

parameters of users and service personnel (the network as a whole). The computational basis of
the method for assessing network performance is the following algorithm: 1. The initial data for
the simulation are formulated as follows: - The communication network structure is represented
by an oriented graph G(v, u), where vertices correspond to network nodes, and arcs (edges)
represent communication lines. - the characteristics of communication lines are described by the
connectivity matrix 𝐵 = ǁ𝛽𝑖𝑗 ǁ, (𝛽𝑖𝑗 = 1, if there's a connection between sets i and j, and 𝛽𝑖𝑗 = 0;
otherwise. - the set of switching nodes is denoted as 𝑉 = {𝑣𝑖 }, 𝑖 = 1,2, … , 𝑁;. - The set of linear
connections (elements) is represented by U={u_j }, where ,j=1,2,…,S. - the simulation time
interval is denoted as 𝑇𝑀 ;. - Failure rates of elements are given by 𝜆отк, where j=1,2,…,S. - the
𝑗
recovery intensities of elements are represented by 𝜆восст , where j=1,2,…,S. Determines the
number of possible paths between the i–th and j–th users 𝑚𝑖𝑗 (𝑖 ≠ 𝑗; 𝑖, 𝑗 = 1,2, … , 𝑁), at the initial
(t_0) moment of time (when all network elements are functioning). 2. Defines the number of
possible communication paths between all pairs of users, at the initial (𝑡0 ) moment of time (when
all network elements are functioning). 3. Defines the number of possible communication paths
between all pairs of users, at the initial (t0) time point.

𝑀𝑐 = ∑𝑖𝑗 𝑚𝑖𝑗 . (16).

4. A random number n is formed in relation to the law of distribution of the incoming flow
(failures and recoveries). 5. Formation of the timestamp for the next state change (either failure
or recovery) of the element is expressed as:

𝑡𝑘 = 𝑡𝑘−1 + 𝜂 . (17).

where 𝑡𝑘 represents the time of occurrence of the subsequent event. 𝑡𝑘 the moment of receipt of
the previous event. 6. Defines the number of existing paths 𝑃𝑖𝑗 (𝑡𝑘 ) between all users, at 𝑡𝑘 point
in time. 7. Defines the number of existing 𝑃𝑖𝑗 (𝑡𝑘 ) paths between all users, at 𝑡𝑘 point in time. 8.
Determines the degree of network operability, from the position of the i–th user corresponding
with the 𝑡𝑘 user:
𝑃𝑖𝑗 (𝑡𝑘 )
𝛼𝑖𝑗 (𝑡𝑘 ) = . (18).
𝑚𝑖𝑗

9. Defines 𝛼с -the level of network operability as a whole (from the point of view of maintenance
personnel) at a given time

∑𝑖𝑗 𝑃𝑖𝑗 (𝑡𝑘) 𝑃с (𝑡𝑘 )

𝛼𝑖𝑗 (𝑡𝑘 ) = ∑𝑖𝑗 𝑚
or α 𝛼𝑐 (𝑡𝑘 ) = . (19).
𝑀𝑐
𝑖𝑗

10. A check is performed to see if the moment 𝑡𝑘 – the arrival of a random event for a given
simulation interval has been exceeded 𝑡𝑘 < 𝑇𝑀 . If the condition is met, then control is transferred

8
 Advanced Computing: An International Journal (ACIJ), Vol.15, No.6, November 2024
to step 4. If the condition is met, i.e. the simulation interval is exhausted, you should proceed to
processing the simulation results, step 11. 11. Analysis results.

4. CONCLUSIONS
The reliability of modern computer networks is a complex property characterized by taking into
account a wide variety of parameters for its definition. These parameters include structural and
operational characteristics of the service. Existing methods of performance assessment: a)
analytical – are not suitable for multi-pole systems; b) statistics take into account only failures
and recoveries of technical elements of the communication network. The statistics take into
account only failures and recoveries of technical elements of the communication network. Many
available calculation methods, as a rule, reliability assessment is carried out according to one of
the parameters (for example, network connectivity, time to failure, cost of element restoration,
etc.). This is not entirely true, because in real networks, if some of its elements fail, the network
can perform its functions. The operability of a computer system - is a new reliability assessment
indicator that takes into account the performance of its functions by the system. The purpose of
this study is to create an automation system for evaluating and improving reliability, taking into
account the structure of the network properties and functional characteristics. The computation of
the principal functional attributes of a hierarchical computer network, assuming arbitrary
message transmission flow and predetermined service order, is conducted using a multi-phase
queuing system (QS) model. Each phase of this model is depicted as a QS of the 𝑀 ⃗⃗ 𝑘 /𝑀/𝑛/𝑟 <
∞. type. The structural reliability of the network can be estimated as the average proportion of
connections between the elements of the network graph, which is preserved while its arbitrary
elements are damaged. The proposed method of assessing the operability has no restrictions on
the number of corresponding pairs in the network, it allows you to track the dynamics of network
operability. As a criterion of “operational” reliability, i.e. an assessment of the necessary costs
for the reliable functioning of the communication network, the indicator “reduced costs” is
justified as reflecting the structural and technical characteristics of the communication network.
The developed automation system for evaluating the reliability of a computer network for the
first time allows us to obtain the dependence of functional characteristics on the level of
operability, structural characteristics. This makes it possible, almost for the first time, to have the
dynamics of changes in the system's operability, taking into account failures-restoration of
elements on the one hand and their influence (failures) on the characteristics of the system's
functioning on the other. This conclusion is clearly confirmed by a computational experiment.
Three schemes for improving the reliability of a computer network are presented. These are the
restoration of network elements, the introduction of a reserve and a combined scheme. The
selection of each particular scheme is made considering the costs necessary for its
implementation. Consequently, the aforementioned costs are taken into account. A computational
experiment conducted on a computer network with a specified configuration validates and
demonstrates the viability of the proposed approach for evaluating and enhancing reliability. The
analysis of the current state of the reliability problem on computer networks carried out in this
study shows that in determining reliability as a property to ensure communication, while
maintaining the values of established quality indicators in time under specified operating
conditions, not only the characteristics of the state of technical means, their operating time, but
also many user requirements for indicators of reliability of message delivery, reliability and the
error-free transmission of messages, the probability of timely delivery of messages. Of course,
taking into account these requirements does not fit into the schemes of traditional reliability
calculation methods and requires a new, systematic approach to the formation of a
comprehensive reliability assessment, which is the subject of this work, the scientific and
practical results of which are as follows: 1. A systematic approach to assessing the operability of
a computer network is proposed, including its functional, structural, and operational aspects for

9
 Advanced Computing: An International Journal (ACIJ), Vol.15, No.6, November 2024
networks of arbitrary configuration, which makes it possible to provide controls with objective
and reliable information about the state of a computer network. The need to calculate the
structural parameters and performance indicators of a computer network is caused by the
following reasons: the need to assess the state of the network in order to make a decision on
network management in conditions of damage to elements and the need to evaluate intermediate
network options at the stage of its synthesis. 2. A new method for calculating the structural
reliability of complex multifunctional structures, such as computer networks, is proposed. The
structural reliability of the network can be estimated as the average proportion of connections
between the elements of the network graph, which is preserved while its arbitrary elements are
damaged. 3. A method is proposed for assessing the technical condition of a computer network
based on assessments of the operability of both individual corresponding nodes and the entire
network. The method imposes no limitations on the number of matching pairs in the network and
enables monitoring the dynamics of network performance. 4. The criterion for estimating the cost
of network maintenance was reasonably chosen, based on the assumptions that measures to
restore elements should be the simplest and the cost of diagnostic tools should be minimal. 5. The
structure and management principles of the automation system for obtaining a comprehensive
health assessment have been developed, including blocks for calculating the functional
characteristics of health, assessing the level of system performance, mechanisms for improving
system performance, in case of failure of its elements. 6. The dependence between the functional
characteristics of reliability and the level of its operability is obtained. This allows you to identify
the risk zones of the system, putting it into an inoperable state. Therefore, a method is introduced
for evaluating the technical condition of a computer network by assessing the operability of
individual correlating nodes (pairs) as well as the overall compatibility of the network (users).
This method imposes no restrictions on the number of correlating pairs in the network, enables
monitoring the network's performance dynamics, and consequently serves as a convenient tool in
synthesizing network maintenance solutions. Moreover, it emerges as a key component in the
design of future digital systems.

REFERENCES

[1] Половко А.М. & Гуров С.В., (2006) “Основы теории надежности”, БХВ-Петербург Publishers,
704 p.
[2] Захаров Г.П. & Захаренко Г.П. (1989) “Детерминированная модель оценки живучести и
уязвимости сетей”, АН СССР Publishers, Техническая кибернетика, No. 2.
[3] Громов Ю.Ю., (2010) “ Надежность информационных систем ”, ГОУ ВПО ТГТУ Publishers, 160
p.
[4] Гузик В.Ф. & Самойленко А.П., (2008) “ Принципы проектирования интегральной модели
оценки надежности информационно-вычислительных систем”, ЮФУ. Технические науки
Publishers, pp 36-39.
[5] Василенко Н.В. & Макаров В.А., (2004) “ Модели оценки надежности программного
обеспечения ”, Вестник Новгородского государственного университета Publishers, No. 2 pp 126–
132.
[6] Чекал Е.Г. & Чичев А.А., (12) “Надежность информационных систем”, УлГУ Publishers, 118 p.
[7] Н.Рахимов, &, О. Примкулов (2023) “ Ахборот тизимларида мантиқий хулосалаш
самарадорлигини ошириш ёндашуви”, International Scientific and Practical Conference on
Algorithms and Current Problems of Programming. Pp 56-59

AUTHOR
Aziz Ishmukhamedov

10