Edge computing is transforming smart cities by enabling rapid, local data processing and reducing core network congestion, but decentralization also broadens the attack surface, exposing systems to novel security and privacy threats. Traditional cloud-based defenses often fall short in these dynamic and distributed environments. This paper delivers four main contributions: (1) a unified taxonomy of edge computing threats in smart cities, including data privacy/integrity risks, identity and access vulnerabilities, and emerging AI-driven and physical attacks; (2) a critical comparison of five leading defense strategies, AI-based anomaly detection, blockchain-enabled trust, hardware roots of trust, secure virtualization and privacy-preserving analytics, evaluated for security, overhead and deployment maturity, with practical recommendations; (3) an analysis of the sustainability implications of security solutions, focusing on energy and carbon impacts and their alignment with Sustainable Development Goals (SDGs); and (4) a forward-looking roadmap highlighting challenges such as quantum-resistant cryptography, federated learning, digital twin-based risk modeling, and real-time policy enforcement. Together, these contributions provide a structured framework for designing secure, reliable, and sustainable edge infrastructures in future smart cities.

Unmanned Aerial Vehicles (UAVs) have several applications in civil and military domains. The high data rate connectivity required by UAVs makes the use of Free Space Optical (FSO) communications very suitable. In this paper, we consider a three-layer optical UAV network architecture allowing the real-time delivery of collected data. The UAV network is exposed to several types of attacks such as replay attacks, black hole attacks, jamming attacks, etc. In this work, we are interested to a specific class of attacks which vises to perform an intentional interference with signals transmitted between two UAVs on an FSO link, which is the jamming attack. This attack can be seen as a form of denial of service attacks since it prevents the transmitter and the receiver to properly communicate through the FSO link and to use the network resources. In this paper, we present some jamming attack scenarios that can occur in an optical UAV network, then we present a countermeasure approach allowing to recover the transmitted signals in the presence of a jamming attack.

Recently, Visible Light Communication (VLC) has been proposed as a low-cost and low-complexity technology for vehicular communications. In this paper, we propose a vehicular network architecture that enables wireless Infrastructure-to-Vehicle (I2V) and Vehicle-to-Vehicle (V2V) communication to provide value added services. The wireless communication technique chosen to enable the data transmission is VLC due to its numerous advantages. A tree-based network access scheme is developed, providing a rapid connection and allowing to maximize at a large extent the network coverage beyond the area uncovered by lighting gateways. In particular, we propose a QoS oriented VANET architecture, which is based on a distributed and real-time computation of QoS metrics for multimedia services. QoS parameters are computed by the RSU announcing the route, and then updated by each intermediate node part of the tree under construction. Furthermore, we present a codeword-based node architecture that allows the optical switching of traffic from a vehicle node to another. Moreover, each vehicle in the network is identified by a structured codeword which allow to obtain a virtual tree that is not physically established. Based on the obtained virtual tree, we develop a multicast routing protocol for VANET.