An LSTM-Based Network Slicing Classification Future Predictive Framework for Optimized Resource Allocation in C-V2X

Abood, Mohammed Salah; Wang, Hua; He, Dongxuan; Fathy, Maha; Rashid, Sami Abduljabbar; Alibakhshikenari, Mohammad; Virdee, Bal S.; Khan, Salahuddin; Pau, Giovanni; Dayoub, Iyad; Livreri, Patrizia; Elwi, Taha A. · 2023 · OpenAlex-citations

DOI: 10.1109/access.2023.3332225

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of optimizing resource allocation in Cellular Vehicle-to-Everything (C-V2X) networks, specifically within the context of 5G network slicing. The authors identify that traditional network slicing methods often suffer from bias, leading to non-optimal resource distribution where certain slices become overloaded while others remain underutilized. Existing Machine Learning (ML) approaches typically allocate resources based on current network conditions and immediate service requirements, ignoring temporal trends and future demand patterns. To resolve this, the study proposes a Deep Learning framework that combines immediate classification with future predictive analysis to ensure Ultra-Reliable Low-Latency Communication (URLLC) and high Quality of Service (QoS) for autonomous vehicles and other critical applications. The proposed methodology utilizes a hybrid model comprising Artificial Neural Networks (ANN) and Long Short-Term Memory (LSTM) networks. The ANN component classifies incoming traffic requests into appropriate network slices—Enhanced Mobile Broadband (eMBB), Massive Machine Type Communication (mMTC), or URLLC—based on service requirements such as Packet Delay Budget (PDB) and Packet Loss Rate (PLR). Concurrently, the LSTM model performs time-series prediction to forecast network conditions and slice load for the next 30 seconds (approximately 10,000 requests). This predictive capability allows the system to anticipate congestion and proactively reroute traffic from saturated slices to underutilized ones. The study employs a dataset derived from previous research [27], which includes features such as use case type, UE category, supported technology, time of day, and QoS metrics. The resource allocation is mathematically modeled to distribute available resources based on predicted demand and QoS priorities. The findings demonstrate that the integrated LSTM-ANN framework effectively optimizes network performance by balancing load across slices. By predicting future demand, the model prevents network saturation and maintains low latency and high reliability, which are critical for C-V2X applications like collision avoidance and automated driving. The approach allows for dynamic resource adjustment, ensuring that critical services retain necessary resource margins while less critical traffic is diverted to available capacity. The simulation results indicate improvements in network efficiency measures, including reduced latency rates and enhanced reliability compared to static or purely reactive allocation methods. The significance of this work lies in its contribution to the development of intelligent, self-optimizing 5G networks. By incorporating future predictive analysis into network slicing, the framework addresses the limitations of current ML-based solutions that rely solely on historical or current data. This approach supports the scalability and spectrum efficiency required for dense urban environments and emerging IoT applications. The authors conclude that this method enhances the sustainability and performance of C-V2X infrastructure, facilitating the broader adoption of autonomous vehicles and smart city technologies by ensuring robust, low-latency communication under varying network conditions.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• v2x connected vehicle