Enhancing Network Performance Using Machine Learning-Driven Genetic Algorithms

View at publisher Summarise with VIRAT
Authors: N. Akhila, Suneel Pappala, S. Rama Sree
Cited by: 0
References: 19

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The increasing competency of the heterogeneous wireless network, such as the IoT network, cell ad hoc network, and the upcoming 5G/6G networks, requires effective and adaptive optimization protocols to allow maintaining the performance in changing environments. Traditional genetic algorithms (GA) have been used extensively in optimization of routing, resource allocation and fault toleration systems, although they are generally slow to converge, and computationally expensive. On the other hand, machine learning (ML) methods offer predictive efficiency at the lack of flexibility in rapidly changing conditions. The challenges are introduced by way of an ML-assisted GA framework to aim at optimizing network behaviour. It aims at improving throughput, minimisation of latency, higher packet delivery ratio (PDR) and minimisation of energy usage by combining ML surrogates with GA exploration. The data set is based on reallife data, CAIDA Internet backbone traces and CRAWDAD Wi-Fi mobility logs alongside laboratory data sets, such as NS-3 synthetic scenarios. ML models are used both as proxies in place of QoS prediction over a variety of GA knowledge (mutation, crossover) and as a feedback loop to achieve flexibility via surrogate retraining. It has been found that the hybrid structure outperforms GA-only and ML-only baselines, with up to${1 7. 6 \%}$higher throughput, which is lessened by a factor of 25.7, delay is minimized, and reliability. Convergence is also lowered at a faster rate and minimized number of generations is used to get near optimal solutions. Similar to the implementation of ML prediction and GA exploration, a resilient scalable optimization scheme is provided by the concept, which can eventually be extended to reinforcement learning, federation learning acceptance, and validation in the ultra-dense mart of 6G space.

Fields of Study & Subjects

Network Traffic and Congestion Control Metaheuristic Optimization Algorithms Research Software-Defined Networks and 5G

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
2001–2008
Publication date
21 January 2026
Language
EN