Enabling Ultra-Dense, Open-RAN, Vehicular Networks with Non-Linear MIMO Processing

Abstract

Future autonomous transportation systems necessitate network infrastructure capable of accommodating massive vehicular connectivity, despite the scarce availability of frequency resources. Current approaches for achieving such required high spectral efficiency, rely on the utilization of Multiple-Input, Multiple-Output (MIMO) technology. However, conventional MIMO processing approaches, based on linear processing principles, leave much of the system's capacity heavily unexploited. They typically require a large number of power-consuming antennas and RF-chains to support a substantially smaller number of concurrently connected devices, even when the devices are transmitting at low rates. This translates to inflated operational costs that become substantial, particularly in ultra-dense, metropolitan-scale deployments. Therefore, the question is how to efficiently harness this unexploited MIMO capacity and fully leverage the available RF infrastructure to maximize device connectivity. Addressing this challenge, this work proposes an Open Radio Access Network (Open-RAN) deployment, with Massively Parallelizable Non-linear (MPNL) MIMO processing for densely deployed, and power-efficient vehicular networks. For the first time, we quantify the substantial gains of MPNL in achieving massive vehicular connectivity with significantly reduced utilized antennas, compared to conventional linear approaches, and without any throughput loss. We find that an Open-RAN-based realization exploiting the MPNL advancements can yield an increase of over 300% in terms of concurrently transmitting single-antenna vehicles in urban mobility settings and for various Vehicle-to-Infrastructure (V2I) and Network (V2N) use cases. In this context, we discuss how implementing MPNL allows for simpler and more densely deployed radio units, paving the way for fully autonomous and sustainable transportation systems.