Skip to content
Conference

A Survey on LLM-based Network Resource Optimization

Jul 2026 · International Conference on Ubiquitous and Future Networks · pp. 328-330 · 0 citations · 17 references

Abstract

With the rapid proliferation of high-bandwidth and low-latency services such as virtual reality (VR), holographic communications, and large-scale Internet of Things (IoT), the complexity of network resource management has increased significantly. Network resource optimization plays a crucial role in improving throughput, reducing latency, and enhancing energy efficiency by enabling efficient utilization of limited wireless and wired resources. Conventional approaches, including rule-based methods, mathematical optimization, and reinforcement learning-based techniques, can achieve satisfactory performance in specific environments. However, they suffer from limitations such as poor generalization to dynamic environments, high modeling complexity, and difficulties in real-time decision-making. To overcome these limitations, recent studies have begun to explore network resource optimization based on Large Language Models (LLMs). This paper presents a comprehensive survey of LLM-based network resource optimization techniques. Existing studies are classified according to the role of LLMs, and the characteristics and limitations of each approach are analyzed.

View source

Similar papers

Open access 2026

A Novel Power Optimization Technique for Sliced 5G Network

—To reduce power consumption and extend network lifespan, academic and industrial groups have focused on energy-efficiency approaches for Next Generation Networks (NGNs). Fifth-generation (5G) networks offer a large number of services at high data rates, low latency, and massive connectivity. Increasing volumes of heterogeneous traffic from billions of devices, ranging from smartphones to intelligent transport systems, significantly challenge network resource utilization, particularly power consumption. This study targets energy-efficient resource allocation in sliced 5G systems, ensuring service-level guarantees for heterogeneous applications through intelligent optimization. This work proposes a novel hybrid optimization framework for energy-aware resource provisioning in 5G sliced networks using Hybrid Grey Wolf–Tasmanian Devil Optimization (HGWTDO) with a Linear Pattern Search (LPS) refinement technique. While HGWTDO combines the global search ability of Grey Wolf Optimization (GWO) and the exploitation abilities of the Tasmanian Devil Optimizer, the addition of LPS provides accurate local convergence. LPS has been integrated into the proposed solution to enhance optimization results. The solution is augmented with a Classification Tree-based classification that assigns users to their corresponding slices for Mobile Broadband (eMBB), Ultra-Reliable Low Latency Communications (URLLC), and massive Machine-Type Communication (mMTC) based on quality of service (QoS) requirements. The suggested system provides improved power efficiency under QoS constraints and is an intelligent, scalable solution for energy-aware 5G network slicing compared with existing techniques.

P. Raddy, Sudhanva A M, Arathi R. Shankar · 0 citations
Open access Jul 2026

A unified machine learning framework for intelligent resource allocation toward 6G wireless communications.

A Dual-Stage Multi-Time-Scale Temporal Attention-Based LSTM network (D-MTSTA-LSTM) has been architected, which effectively learns short- and long-term relationships in network trends, thereby precisely predicting optimal communication routes and associated power and spectrum allocation.

Nishu Gupta, Rupali Bhartiya, S. Rathod et al. · 0 citations
Open access Jul 2026

A Multi-Stage Adaptive Approach for Resource Optimization and Mobility Management in 5G Networks

The 5G wireless networks are fast with low latency and support a very large number of connected devices, making it possible to support advanced applications and services. Such environment demands efficient management of network resources that can be achieved by evaluation of traffic load, distribution of devices, and state of signals to ensure quality-of-service (QoS) and network performance. Conventional resource allocation and mobility management techniques tend to have challenges of flexibility, computational effectiveness, as well as optimal choice of devices in disparate network conditions. In order to address these issues, this paper introduces a multi-stage resource management system that is adaptive in 5G networks to address QoS and mobility efficiency issues. The framework combines the set of network parameters with Adaptive Multivariate Kernel Resource Estimation (AMKRE) in case of assessing and normalizing available resources. Resource-Aware Device Selection (ORADS) is used to select the appropriate devices and then it is followed by Robust Filter-Based Rank Evaluation (RFBRE) where devices are ranked on the basis of reliability and link quality. The algorithm that optimizes the resources distribution taking into consideration the space constraints is known as DistanceAware Adaptive Particle Swarm Optimization (DAAPSO), and the algorithm that provides the continuity of mobility is known as Hybrid Predictive Soft Handover Control (HPSHC). The multi-dimensional paradigm improves the resource utilization, quality-of-service (QoS) provisioning, and mobility management in 5G networks.

G. Ramasamy, C. Chandrasekar · 0 citations
Conference Jul 2026

5G NetOptima: AI-Driven Real-Time Latency and Bandwidth Optimization for Intelligent Resource Allocation in Next-Generation Networks

With the advent of fifth-generation (5G) networks, the world has become wireless thanks to ultra-low latency, high bandwidth, and massive connectivity that have become an essential part of applications like autonomous vehicles, industrial IoT, smart cities, and real-time multimedia streaming. Conventional fixed or intuitive-based resource allocation schemes fail to adjust well to changing network states and user demands with varied needs leading to delay of service, congestion and the poor use of spectrum. In this paper, 5G NetOptima, which is an AI-based real-time resource allocation framework, is introduced and optimizes both bandwidth and latency at the same time. The suggested system uses machine learning models predicting the intelligent allocation decisions by analyzing network parameters such as user density, traffic type, channel quality, and the priority of services continuously. Dynamic priorities are given to latency sensitive and mission critical services, whereas bandwidth utilization is optimized over the entire bandwidth to improve the Quality of Service (QoS) and Quality of Experience (QoE). The vast simulations show that 5G NetOptima is more efficient than traditional methods of the allocation, as the 5G system facilitates the reduction of latency by the significant degree, enhanced the throughput, and enhanced the load balancing throughout the network. The most important novelty of the work is related to its combined AI-oriented structure which adjusts to all network parameters in real time, provides an efficient, scalable, and intelligent solution to 5G networks of the next generation.

Ashok Kumar.R, Dharshani.R, J. Amirtha et al. · 0 citations