Traffic prediction assisted wavelength allocation in vehicle-to-infrastructure communication: A fiber-wireless network based framework

The advent of the next generation of connected and autonomous cars offers immense opportunities for both users as well as service providers. In particular, fiber-wireless (FiWi) based vehicle-to-infrastructure (V2I) network can facilitate some of the stringent requirements of sixth-generation (6G) vehicular networks, including higher capacity, lower delay, and ubiquitous connectivity. FiWi based V2I network integrates the next generation passive optical network 2 (NG-PON2) with IEEE 802.11p based V2I network. In this work, we first review the various kinds of vehicular data traffic and their desired key performance indicators (KPIs), namely throughput, delay, and reliability. Depending on the KPI requirements, the V2I traffic is classified among four classes and assigned to different transmission containers (T-CONTs) of the optical network unit (ONU). Further, in order to minimize the delay of the network, we propose a machine learning (ML) based T-CONT priority assignment wavelength allocation algorithm that minimizes the number of wavelength switching instances in the PON. The performance of the proposed ML-based wavelength allocation algorithm is compared with the other approaches, namely random and equal T-CONT based wavelength allocation algorithms. Simulation results demonstrate the efficiency of the proposed algorithm vis-a-vis other approaches in terms of end-to-end (e2e) delay, throughput, and reliability.