Particle Swarm optimization algorithm-based optimization and allocation method for improving ARPANET network reliability

Introduction

This study investigated the most efficient methods for ensuring the reliability of the Advanced Research Projects Agency (ARPA) network (Abuali et al., 1994) [1] . The system's reliability was measured by calculating the shortest paths within the connectivity matrices. According to H. Sulaiman et al. (2023b) [4] , "all paths are determined using Boolean algebra, followed by eliminating nodes to create minimal paths" (Hasslinger et al., 2005) [11] . The reliability function was analyzed to understand the security level of the implemented advanced system. The paper also addressed the mathematical challenge of optimizing reliability while considering the network's operational history. Reliability requirements for any component of a complex system are determined by its importance and the optimization of its location. The primary objectives are to reduce overall costs while enhancing the system's lifespan and reliability (Abuali et al., 1994; Elbaum & Sidi, 1996) [1,7] . Some components may require extensive customization, based on where they are in the framework, to increase overall dependability. Engineers face various challenges when optimizing mechanical and electrical systems (H Sulaiman et al., 2023a; Kirkpatrick, 1984) [3,14] . The aim is to enhance the distribution and reliability of complex systems while minimizing costs, which can be represented by factors such as weight, volume, or other parameters. One of the key factors in determining a component's reliability is ensuring that the model is cost-based, which validates the input component. The recommended cost factor settings can be modified (Abed et al., 2019) [2] , allowing technologists to evaluate how the scheme changes over time and determine whether each component meets basic reliability standards. Additionally, the model must consider the accuracy of the input system analysis. In larger systems, certain scenarios can pose significant challenges, leading to an increased workload, particularly when basic systems are involved. The results were obtained using the atom swarm optimization (PSO) algorithm, which assistances solve improvement issues with intricate systems. A logarithmic model was used to calculate costs.

Network optimization and (ARPA) reliability allocation

Deliberate a (ARPA) system anywhere machineries relevant to reliability are provided [1,5] . The succeeding proceedings are employed by us: 𝐶 𝑖 (𝑅 𝑖 )=(cost of section); 0 ≤ 𝑅 𝑖 ≤ 1=(dependability as an section); 𝑅 𝑠 =(reliability of the scheme);

(is the overall cost of the system, where 𝑎 𝑖 > 0); RG attitudes for "systems reliability" goal.

The end goal is its capacity to assign resources to all or specific components in a proper way. Nonlinear developing necessitates problems [2,3,16] , and the limit is still essential even if it is not linear, and it may be examined in terms of costs. Its modular architecture and the different roles that each component plays lead to a number of possible outcomes. Our power is the sum of the numerous distinct systems parts, each of that's has a distinct amount of dependability.

Assume that [𝐶 𝑖 (𝑅 𝑖 )] satisfies specific criteria and that the part cost purpose is equitable. Differentiated, optimistic purposes that rise from [⟹ 𝑑𝐶 𝑖 𝑑𝑅 𝑖 ≥ 0]. Are known by the name [8] . "The previous approach reduces the constraint on system reliability by obtaining an all-out framework cost foundation, subject to RG" [12] .

Optimization of Particle Swarms (PSO)

The presence of the distinctive ARPA gene highlights the use of Particle Swarm Optimization (PSO) as an unconventional method for network optimization [3,4] . PSO operates through well-defined steps, which include encoding, initialization, evaluation, cloning, crossover, mutation, and termination [6,7] .  Encoding: In this stage, the mutable quantity in the optimization delinquent are encoded as genes. A set of leading chromosomes is randomly selected from a pool of chromosomes. The population consists of these randomly chosen chromosomes, with the size of the population determined by the arbitrary number of chromosomes.  Evaluation: Reproduction aims to minimize the number of harmful chromosomes in the offspring while maximizing beneficial ones. Every chromosome in the population is given a fitness score determined by the gene order; the fitness value indicates the chance of a chromosome's survival when the gene order changes. Replicated mitochondria are assessed based on fitness indicators derived from various genetic combos.  Crossover: Maternal and paternal chromosomes are switched during this stage. To serve as paternal chromosomes, two chromosomes are chosen at random from the population. At crossover points which are selected so that there are fewer crossover spots than there are genes on the chromosome genes are switched. Two new chromosomes are then created by combining the genes from both parent chromosomes. This step focuses on where the crossing points intersect.  Mutation: During mutation, a new chromosome is created that differs from the existing population. A random selection process is used to select a modified chromosome, and among its genes, one distorted genetic factor is randomly selected and its worth is changed. This process is repeated until the population's average fitness level stabilizes.

Decoding the best chromosome in the neighborhood is the last step in resolving the optimization problem. In a previous study, PSO was partially applied to construct an ARPA network to improve bandwidth [13,14] . However, the overall ARPA network planning problem dismissed meta-inference.

While particle swarm optimization relies on genetic evolution, it disregards the knowledge generation process inherent to cultural progress. As stated by renowned evolutionary biologist Richard Dawkins, "Cultural transmission is analogous to genetic transmission in that, although basically conservative, it can give rise to a form of evolution".

One drawback of PSO-based systems is their rapid convergence on local preferences. The enhanced version of PSO, however, provides a solution to this issue by employing local search strategies to refine outcomes. In this experiment, the particle swarm optimization algorithm was used as the local search method, aiming to consistently improve the current fitness level. The algorithm continues searching for improved states, halting only when a more optimal state is found.

Use in a complex scheme

To facilitate estimations, we wish to set up the "ARPA" network so that each of its parts are connected in series or parallel. The following describes the dependability of a series network with n components and a parallel network:

In this case, [𝑅 𝑖 ] means the module i's reliability, while [𝑅 𝑁 ] signifies the ARPA net's dependability [6,8] . Equations ( 1) and ( 2) will be used to comparation each composite net's dependability with the providing p smallest paths.

In this case, "α" represents the index of the first module of a minimal pa z, while "ω" represents the index of the last component.

In order to assess the consistency of the complicated system shown in Figure 2, one might utilize Equation (3).

Conclusion

This study aims to allocate and enhance the reliability of the previously discussed complex network. The system optimization problem, which utilizes computer science methods, partitions the reliability of each component within the network. This challenge can also be approached as a nonlinear software design problem, incorporating restrictions related to human and substantial capitals (ARPA network reliability). The particle swarm optimization (PSO) technique, which considers the positions of various components within the complex network, was employed to address the optimal reliability allocation problem. The maximum and lowermost reliability allocations for the first and second mechanisms, as well as the fourth and sixth machineries, are detailed in the table above. The total system reliability allocation is (Rs=0.94). A key advantage of this model is that all calculations can be performed using mathematical software, as the allocation results were derived using the "Matlab" program, version R2020a.

References

1. Abuali FN, Schoenefeld DA, Wainwright RL. Terminal assignment in a communications network using genetic algorithms. In: Proc 22 nd Annual ACM Computer Science Conference (CSC'94); 1994; Phoenix, AZ, USA. p. 74-81.
2. Abed SA, Sulaiman HK, Hassan ZAH. Reliability allocation and optimization for ROSS of a spacecraft using genetic algorithm. J Phys Conf Ser. 2019;1234:032034.
3. Sulaiman HK, Hassan ZAH, Ali FH, et al. A model for allocation and optimization of reliability complex network using particle swarm optimization and genetic algorithms. Comput Sci. 2023;18(2):225-231.
4. Sulaiman HK, Ali FH, Hassan ZAH. Computational models for allocation and optimization of reliability for ROSS network. In: AIP Conf Proc. 2023;2591(1):020014.
5. Davis L, Cox A, Qiu Y. A genetic algorithm for survivable network design. In: Proc 5th Int Conf Genetic Algorithms; 1993. San Mateo (CA): Morgan Kaufmann; p. 408-415.
6. Elbaum R, Sidi M. Topological design of local area networks using genetic algorithms. In: Proc IEEE INFOCOM'95; 1995; Boston, MA, USA. Vol. 1, p. 6471.
7. Gerla M, Monteiro JAS, Pazos R. Topology design and bandwidth allocation in ATM networks. IEEE J Sel Areas Commun. 1989;7(8):1253-1262.
8. Gerla M, Kleinrock L. On the topological design of distributed computer networks. IEEE Trans Commun. 1977;25(1):55-67.
9. Goldberg DE. Genetic algorithms in search, optimization and machine learning. New York: Addison-Wesley; 1991.
10. Hasslinger G, Schnitter S, Franzke M. The efficiency of traffic engineering with regard to link failure resilience. Telecommun Syst. 2005;29:109-130.
11. Kim SB, Kim MJ, Lee SI. Mathematical models for dimensioning of ATM networks. In: Proc IEEE GLOBECOM'95; 1995; Singapore.
12. Feng X, Zhu X, Zhao W, Li X. Reliability of electric vehicle with wind turbine based on particle swarm optimization. Chem Eng Trans. 2018;66:1291-1296.
13. Kirkpatrick S, Gelatt CD, Vecchi MP. Optimization by simulated annealing. Science. 1983;220:671-680.
14. Liu X. Network capacity allocation for traffic with time priorities. Int J Netw Manag. 2003;13:411-417.
15. Rios M, Marianov V, Abaroa C. Design of heterogeneous traffic networks using simulated annealing algorithms. In: Proc ICOIN 2005. Lect Notes Comput Sci. 2005;3391:520-530.