6533b82dfe1ef96bd1291eb0

RESEARCH PRODUCT

Hidden Markov Model Based Machine Learning for mMTC Device Cell Association in 5G Networks

subject

description

Massive machine-type communication (mMTC) is expected to play a pivotal role in emerging 5G networks. Considering the dense deployment of small cells and the existence of heterogeneous cells, an MTC device can discover multiple cells for association. Under traditional cell association mechanisms, MTC devices are typically associated with an eNodeB with highest signal strength. However, the selected eNodeB may not be able to handle mMTC requests due to network congestion and overload. Therefore, reliable cell association would provide a smarter solution to facilitate mMTC connections. To enable such a solution, a hidden Markov model (HMM) based machine learning (ML) technique is proposed in this paper to perform optimal cell association. As such, we consider MTC devices with network-assisted decision-making capabilities for selecting the most appropriate eNodeB for data transmission. The proposed HMM based ML technique focuses mainly on the reliability and availability of network resources. Correspondingly, two schemes are developed based on the classical reliability function and the next probable state of the HMM. Based on simulations under various configurations, we demonstrate the advantage of the proposed schemes over a random cell selection scheme.