With the support of Vehicle-To-everything (V2X) communications, connected and automated road safety systems could offer a revolutionary, complementary solution to standalone on-board sensors, while drastically reducing the probability of accidents, as well as the likelihood of fatal injuries for Vulnerable Road Users (VRUs). In this paper, we explore different options for such safety systems, in terms of overall architecture, V2X technology and network deployment, or sensor perception. First, in a decentralized architecture using Pedestrian-To-Vehicle (P2V) and Infrastructure-To-Vehicle (I2V) connectivity relying on C-V2X sidelink (e.k.a., PC5 Mode 4), we examine two concrete cases, with V2X-equipped and passive VRUs respectively. In addition, we consider a centralized solution involving Vehicle-to-Network (V2N) links based on the 5G New Radio (5G-NR) technology, where pedestrian detection relies on the fusion of vehicles' sensors data in a Multi-access Edge Computing (MEC) system. An hybrid approach combining V2N and I2V links is also evaluated. All these use cases are simulated using a full-stack simulator, where we analyze the overall probability of driver awareness, which combines both probabilities of successful sensors operation and critical information reception via V2X at the vehicle (i.e., prior to the crash). Results emphasize the pros/cons of each system architecture independently, as well as the benefits from hybrid multi-technology approaches to raise the driver's awareness about the presence of non-equipped VRUs.