Autonomous QoS-Based Mechanism for Resource Allocation in LTE-Advanced Pro Networks

In Clustering-Based Resource Allocation (CBRA) strategy, choosing an arbitrary number of clusters may trigger problems such as traffic similarity loss, which results in inade- quate resource allocation. This paper proposes a novel unsuper- vised machine learning mechanism that consists of combining the X-Means and Fuzzy C-Means (FCM) clustering algorithms. It establishes features related to the QoS parameters for dataset composition as a way of mapping the flow information. The X- Means algorithm estimates the ideal number of clusters corre- sponding to the provided dataset. The FCM algorithm classifies all network traffic flow from their common features, allowing the system to allocate resources by following the defined order of clusters to which each traffic belongs. The proposed mecha- nism exhibits good performance for real-time video application, compared to some scheduling algorithms employed in the system.