GREEN: Precise Geolocation in Metaverse using Reinforcement Learning-Enabled Sensor Placement

The advancement of mobile networks is swiftly progressing to support an increasingly diverse array of applications. The next generation of cellular networks has been providing services through distinct, logically separate segments known as network slices. A service type gaining momentum within this framework involves augmented and virtual reality (AR/VR) devices, which are instrumental in the emergence of the Metaverse. The successful realization of the Metaverse not only demands strict control over user equipment (UE) latency but also relies heavily on the precise positioning of user devices in a three-dimensional (3D) environment. As the next evolution, 6G networks will function within higher frequency spectrums and more compact setups. In such settings, achieving precise positioning indoors poses a considerable challenge due to the nature of signals at these higher frequencies. To address this, we introduce a novel framework, leveraging established 3GPP standards to craft a hybrid radio access network (RAN) that integrates Wireless Fidelity (Wi-Fi) routers to assist with positioning, offloading this task from base stations (BSs) in indoor settings. We highlight how the strategic placement of the Transmission and Reception Points (TRPs) in our system critically influences the positioning accuracy of AR/VR devices within the Metaverse. We tackle and resolve a complex NPhard problem concerning TRP deployment through a novel deep reinforcement learning (DRL) framework. Our findings reveal that strategic TRP placement significantly enhances positioning accuracy over arbitrary ad hoc placement strategies.