Last-mile delivery operations describe the movement of goods from a distribution or fulfilment centre to the end customer’s location or drop-off locations/collection points. With growing customer expectations, the need to craft new last mile delivery models is becoming eminent. Mobile access hubs are flexible consolidation and transshipment points aiming at creating more sustainable city logistics systems by dynamically using urban space as logistics facilities.

The last mile delivery of goods via EV-based CAV hubs relies on a complex ecosystem using multiple heterogeneous devices, sensors, chargers, and AI intelligence, operating on different layers of the edge-cloud continuum; thus, the attack surface for the delivery operations is constantly increasing. Last-mile delivery services rely heavily on customer trust as they handle sensitive information, including customer addresses and contact details. If customer data is compromised during a cyber-attack, it can lead to privacy concerns and erode customer trust. Incidents like the cyber-attacks in the Courier, Express and Parcel (CEP) sector and on last-mile logistics apps can disrupt the last mile delivery ecosystem and lead to delays, impacting service levels and customer satisfaction.

This case scenario focuses on vehicle operations for last-mile delivery based on a mobile hub built on an autonomous electric vehicle with a dedicated area to accommodate app-openable mobile lockers. Predefined stops along the delivery routes are communicated to users who will have specific time slots to be able to pick up packages and envelopes from the lockers. The opening of these will be linked to the users registered via app on their smartphones. MEDIATE will leverage the potential of multimodal fusion and distributed learning as AIbased automation tools to boost cyber threat intelligence and insulate data poisoning effect through the complementarity of sensors and risk-aware synchronization and orchestration schemes. This use-case will demonstrate the MEDIATE secure platform solution in the automotive domain and, in particular, focusing on three main cases of the attack surface of last-mile deliveries using EV hubs: a) Attack on the sensing layer: Here we focus on potential intrusions on the guidance system, particularly the sensor system (e.g. LiDAR jamming or GPS spoofing), which can cause the vehicle’s control unit to behave erroneously, with grave implications for passenger and road user’s safety.

MEDIATE will utilize the redundancy and complementarity of sensors (e.g. visual and LiDAR), in conjunction with its multi-modal cooperative perception engine to mitigate attacks on all three bands of the operational spectrum: physical, sensing, and perception; b) Cloud-level attacks: MEDIATE also targets at mitigating vulnerabilities related to cloud infrastructures through robustifying the cyber-resilience across the higher levels of the spectrum as well including hacking the cloud platform. In this context, the consortium will launch a decentralized Federated Learning approach for scene understanding, to dramatically minimize communication costs and enhance privacy preservation. This way MEDIATE will address continual Secure integration of IoTs of questionable trustability in trusted ecosystem. c) Attack on the lockers: This scenario will demonstrate the MEDIATE solution in providing a security layer for the physical lockers installed on board the mobile hubs, for preventing violation aiming to gain access without permission and take parcels in an unauthorized manner. Here, the focus is on the locking system and the interface with the application (including access codes and relevant timings) and related user profiles.