Proactive resource allocation in Mobile Networks: A Predictive Guard Channel Scheme using Long Short-Term Memory

The escalating demand for mobile data services necessitates efficient resource management in cellular networks. Traditional reactive resource allocation schemes often suffer from delays and inefficiencies due to the dynamic nature of user mobility and traffic patterns. This paper proposed a proactive resource allocation scheme utilizing a predictive guard channel mechanism powered by Long Short-Term Memory (LSTM) networks. The proposed scheme anticipated future resource demands based on historical traffic and mobility data, enabling the dynamic adjustment of guard channel sizes to mitigate handoff failures and improve Quality of Service (QoS). We presented the system architecture, mathematical model, and algorithmic flow, followed by a performance evaluation comparing our LSTM-based scheme with a conventional guard channel scheme. Simulation results showed that the new call blocking probabilities for Static Guard Channel Scheme ranged from 2 to 7.5 while that of Predictive Guard Channel Scheme ranged from 1.0 to 7.0. Also, the handoff call blocking probabilities for Static Guard Channel Scheme ranged from 1 to 11.9 while that of Predictive Guard Channel Scheme ranged from 0.5 to 5.0 which shows that the Predictive Guard Channel Scheme significantly reduces new call blocking probability and handoff drop probability, and improves resource utilization, making it a promising solution for future mobile networks.