In this paper, we present a framework that integrates AI-based derivation of Access and Usage Control policies for IoT devices, using Large Language Models (LLMs) to automate the generation of policies from unstructured natural language commands. The framework employs a hybrid approach, combining LLMs with dedicated libraries to ensure efficient on-device execution. Our approach is based on a two-step process: first, a fine-tuned LLM converts user commands into structured JSON policy representations; then, a transformation module translates the JSON policies into fully compliant U-XACML policies. To ensure generality across different domains, we introduce a taxonomy-driven dataset creation, which enables policy creation for different environments such as smart homes, smart offices, and healthcare settings. Our evaluation demonstrates that the system achieves 93 % accuracy in policy generation and 91 % accuracy when handling ambiguous or noisy inputs. It also reaches 98 % agreement with expert-defined policies in real-world scenarios. Finally, on-device performance evaluations confirm the feasibility of running the model in practical settings, demonstrating reliable inference under constrained hardware conditions.

The Internet of Things (IoT) is an information service paradigm based on the integration of smart objects, mobile devices, and computers via the Internet. IoT technologies are key enablers for a multitude of applications in diverse fields, such as digital health, smart city, industrial automation, and supply chain. This raises new security and privacy challenges that can be addressed by advanced cryptographic methods. One of the most prominent is attribute-based encryption (ABE), which allows one to encrypt data while enforcing fine-grained access control on it. ABE is advantageous in many IoT applications since it allows data to be safely stored on untrusted storage, such as third-party cloud servers, hackable publish-subscribe brokers, physically accessible sensors, etc. This article surveys the ABE literature proposing schemes and solutions that are best suited for IoT applications. To do so, it first identifies three performance indicators that are key in IoT, namely, the data producer CPU efficiency, the data producer bandwidth efficiency, and the key authority bandwidth efficiency. Then, it analyzes only those schemes that are promising from the point of view of one or more indicators and, therefore, more applicable in typical IoT applications. As a further contribution, this article selects a subset of representative schemes and assesses their efficiency by thorough simulations. Such simulations show that no scheme excels in all three performance indicators at once, but some simultaneously perform well in two indicators.

Lately, many cloud-based applications proposed attribute-based encryption (ABE) as an all-in-one solution for achieving confidentiality and access control. Within this paradigm, data producers store the encrypted data on a semi-trusted cloud server, and users, holding decryption keys issued by a key authority, can decrypt data according to some access control policy. To be used in practical cases, any ABE scheme should implement a key revocation mechanism which assures that a compromised decryption key cannot be used anymore to decrypt data. Yuet al.(2010) introduced an ABE scheme with revocation capabilities that enjoys several unique advantages, such as reactivity and efficiency. In the scheme, the cloud server is entitled to update keys and ciphertexts in order to achieve revocation. Unfortunately, the cloud server retains the power to undo the revocation of a key (revocation undoing attack) so endangering confidentiality. In this article, we propose a revocable ABE scheme that still ensures the advantages of Yuet al.’s scheme, but it also resists to the revocation undoing attack. We formally prove the security of our scheme and show through simulations that the user experiences a slightly higher computational cost with respect to Yuet al.’s scheme.

The Internet of Things (IoT) is a technological vision in which constrained or embedded devices connect together through the Internet. This enables common objects to be empowered with communication and cooperation capabilities. Industry can take an enormous advantage of IoT, leading to the so-called Industrial IoT. In these systems, integrity, confidentiality, and access control over data are key requirements. An emerging approach to reach confidentiality and access control is Attribute-Based Encryption (ABE), which is a technique able to enforce cryptographically an access control over data. In this paper, we propose fABElous, an ABE scheme suitable for Industrial IoT applications which aims at minimizing the overhead of encryption on communication. fABElous ensures data integrity, confidentiality, and access control, while reducing the communication overhead of 35% with respect to using ABE techniques naively.

In the near future, a technological revolution will involve our cities, where a variety of smart services based on the Internet of Things will be developed to facilitate the needs of the citizens. Sensing devices are already being deployed in urban environments, and they will generate huge amounts of data. Such data are typically outsourced to some cloud storage because this lowers capital and operating expenses and guarantees high availability. However, cloud storage may have incentives to release stored data to unauthorized entities. In this work we present ABE-Cities, an encryption scheme for urban sensing which solves the above problems while ensuring fine-grained access control on data by means of Attribute-Based Encryption (ABE). Basically, ABE-Cities encrypts data before storing it in the cloud and provides users with keys able to decrypt only those portions of data the user is authorized to access. In ABE-Cities, the sensing devices perform only lightweight symmetric cryptography operations, thus they can also be resource-constrained. ABE-Cities provides planned expiration of keys, as well as their unplanned revocation. We propose methods to make the key revocation efficient, and we show by simulations the overall efficiency of ABE-Cities.

The capability to verify positions reported by devices is called secure location verification. The majority of the proposed solutions entail the use of many fixed anchors often along with special hardware, e.g., ultra-wideband and ultrasonic transceivers. However, the deployment and maintenance costs of such solutions make them scarcely attractive. A cheaper alternative is to use mobile entities as trusted infrastructure. In particular, Unmanned Aerial Vehicles (UAVs) represent a promising approach. Indeed, recent studies used them to face the secure location verification problem. In this paper, we introduce a low-cost approach based on a swarm of UAVs and a common radio frequency protocol, e.g., WiFi. By experimental simulations, we show that by using only three UAVs our system detects more than 99% of the attacks against an adversary that falsifies its position of at least 20 m. We also consider an adversary capable of tracking UAVs positions. The success probability of such an advanced adversary is smaller than 1% starting from a falsification distance larger than 35m.

The increasing diffusion of smart devices opens to a new era for collecting large quantities of data from urban areas. Sensing information can be collected by using existing network infrastructures, but also by adopting small, cheap and configurable aerial vehicles, namely drones. Our work focusses on studying how to optimize their adoption for smart city applications designed to gather sensing data from user's devices roaming on the ground. To this purpose, we used HUMsim, a tool which generates realistic human traces, to mimic pedestrian mobility. From this dataset, we extract some sociality features that we exploit to plan a social-aware drone trajectory with the goal of maximizing the opportunities of interaction between drone and devices. Our experiments compare social-aware and social-oblivious trajectories showing that knowing the way people move and interact boosts the amount of retrievable data.