Applied & Interdisciplinary
Hamid Rahimi; Reza Sheibani; Gelareh Veisi
Abstract
In this paper, a hybrid framework called E-SNM-CNN-LSTM was proposed for cellular traffic prediction and base station sleep-wake decision support in heterogeneous networks. First, traffic data were processed using an entropy-based refinement step to reduce noise and short-term spikes. Then, a combination ...
Read More
In this paper, a hybrid framework called E-SNM-CNN-LSTM was proposed for cellular traffic prediction and base station sleep-wake decision support in heterogeneous networks. First, traffic data were processed using an entropy-based refinement step to reduce noise and short-term spikes. Then, a combination of CNN and LSTM with the adaptive memory gate (AMG) was employed to extract spatial-temporal patterns. Meta-parameter tuning was subsequently performed using a newly proposed Star-nosed mole (SNM) algorithm. The performance of the proposed method was evaluated on four datasets, Milan, Madrid LTE, Telecom Shanghai, and Trentino, and compared with baseline models and the Transformer--ConvLSTM reference framework. Results showed that the proposed model provided small and stable improvements in MAE, RMSE, and R\textsuperscript{2} for traffic prediction, while achieving higher accuracy and a lower false-off rate in sleep decision-making. In high-density scenarios such as Telecom Shanghai and sparse-data conditions such as Trentino, the framework improved both prediction accuracy and energy-based decision-making quality. Furthermore, simulations of a two-layer macro-small cell network showed that the proposed model increased energy savings and reduced the number of on/off switches. Independent evaluation of the SNM algorithm on the CEC BC 2017 benchmark functions also demonstrated competitive and stable performance in search capability and overall ranking. Overall, the findings indicate that the proposed framework is a practical and reliable solution for efficient energy management in 5G and beyond networks, supporting sustainable operation and intelligent resource utilization across heterogeneous wireless communication environments while maintaining robust performance under diverse network traffic conditions.
Machine Learning & Data Science in Optimization
Saeed Mirzajani; Majid Roohi
Abstract
The prediction of chaotic time series is essential for understanding highly nonlinear and sensitive systems, with the Lorenz system serving as a standard benchmark due to its intricate and non-periodic dynamics. Classical forecasting approaches often struggle to capture such irregularities, ...
Read More
The prediction of chaotic time series is essential for understanding highly nonlinear and sensitive systems, with the Lorenz system serving as a standard benchmark due to its intricate and non-periodic dynamics. Classical forecasting approaches often struggle to capture such irregularities, motivating a shift toward deep learning–based strategies. In this study, we develop two hybrid models—Feedback Long Short-Term Memory (FB-LSTM) and Feedback Variational Stacked LSTM (FBVS-LSTM), specifically designed for multivariate prediction of the Lorenz system. By embedding feedback structures into LSTM networks, the proposed methods deliver enhanced short-term prediction performance without substantial computational costs. Comparative simulations indicate that our frameworks surpass traditional RNNs and baseline LSTM models, achieving prediction accuracies up to 94%. These findings indicate that feedback-enhanced architectures offer effective and practical tools for forecasting chaotic systems, with potential applications in both scientific research and engineering practice.
