Event-Driven Source Traffic Prediction in Machine-Type Communications\n Using LSTM Networks

2021/01/12 by Thulitha Senevirathna, Bathiya Thennakoon, Senevirathna, Thulitha +9
Engineering · Computer Science · #Advanced Data and IoT Technologies #Wireless Signal Modulation Classification #Software-Defined Networks and 5G

paper · pdf · doi:10.48550/arxiv.2101.04365

Abstract

Source traffic prediction is one of the main challenges of enabling\npredictive resource allocation in machine type communications (MTC). In this\npaper, a Long Short-Term Memory (LSTM) based deep learning approach is proposed\nfor event-driven source traffic prediction. The source traffic prediction\nproblem can be formulated as a sequence generation task where the main focus is\npredicting the transmission states of machine-type devices (MTDs) based on\ntheir past transmission data. This is done by restructuring the transmission\ndata in a way that the LSTM network can identify the causal relationship\nbetween the devices. Knowledge of such a causal relationship can enable\nevent-driven traffic prediction. The performance of the proposed approach is\nstudied using data regarding events from MTDs with different ranges of entropy.\nOur model outperforms existing baseline solutions in saving resources and\naccuracy with a margin of around 9%. Reduction in Random Access (RA) requests\nby our model is also analyzed to demonstrate the low amount of signaling\nrequired as a result of our proposed LSTM based source traffic prediction\napproach.\n

Related

• A Directed Information Learning Framework for Event-Driven M2M Traffic Prediction
• Traffic Prediction Based on Random Connectivity in Deep Learning with Long Short-Term Memory
• Deep Learning with Long Short-Term Memory for Time Series Prediction
• NeuTM: A Neural Network-based Framework for Traffic Matrix Prediction in\n SDN
• A Long Short-Term Memory Recurrent Neural Network Framework for Network Traffic Matrix Prediction