The Big-Data-Driven Intelligent Wireless Network: Architecture, Use Cases, Solutions, and Future Trends

Abstract

The concept of using big data (BD) for wireless communication network optimization is no longer new. However, previous work has primarily focused on long-term policies in the network, such as network planning and management. Apart from this, the source of the data collected for analysis/model training is mostly limited to the core network (CN). In this article, we introduce a novel data-driven intelligent radio access network (RAN) architecture that is hierarchical and distributed and operates in real time. We also identify the required data and respective workflows that facilitate intelligent network optimizations. It is our strong belief that the wireless BD (WBD) and machine-learning/artificial-intelligence (AI)-based methodology applies to all layers of the communication system. To demonstrate the superior performance gains of our proposed methodology, two use cases are analyzed with system-level simulations; one is the neural-network-aided optimization for Transmission Control Protocol (TCP), and the other is prediction-based proactive mobility management.