Abstract

Abstract
Contracts and assertions are effective methods to enhance software quality by enforcing preconditions, postconditions, and invariants. Previous research has demonstrated the value of contracts in traditional software development. However, the adoption and impact of contracts in the context of mobile app development, particularly of Android apps, remain unexplored. To address this, we present the first large-scale empirical study on the use of contracts in Android apps, written in Java or Kotlin. We consider contract elements divided into five categories: conditional runtime exceptions, APIs, annotations, assertions, and other. We analyzed 2,390 Android apps from the F-Droid repository and processed 52,977 KLOC to determine 1) how and to what extent contracts are used, 2) which language features are used to denote contracts, 3) how contract usage evolves from the first to the last version, and 4) whether contracts are used safely in the context of program evolution and inheritance. Our findings include: 1) although most apps do not specify contracts, annotation-based approaches are the most popular; 2) apps that use contracts continue to use them in later versions, but the number of methods increases at a higher rate than the number of contracts; and 3) there are potentially unsafe specification changes when apps evolve and in subtyping relationships, which indicates a lack of specification stability. Finally, we present a qualitative study that gathers challenges faced by practitioners when using contracts and that validates our recommendations.

Abstract
The shift to AI-native 6G networks demands autonomous slicing strategies that can adapt to diverse and evolving edge and IoT service needs. Two paradigms have emerged: Learn to Slice (L2S), where AI optimizes network slicing for general services, and Slice to Learn (S2L), where slices support AI model training, often offloaded from Internet of Things (IoT) devices. Existing S2L approaches typically optimize communication or computation in isolation. This paper presents the first unified framework that jointly optimizes communication resources, computation capacity, and AI hyperparameters to maximize the average accuracy of multiple concurrent AI services. We address the complexity of this joint problem by applying L2S-inspired techniques to enhance S2L, introducing two autonomous agents: EXP3 from online convex optimization and DQN from deep reinforcement learning. Extensive experiments demonstrate and contrast the effectiveness of these agents in maximizing aggregated AI accuracy, supporting knowledge transfer, and sustaining robust performance under adversarial and long-term conditions, thereby enhancing the realization of zero-touch network management for AI services in 6G networks, supporting resource-constrained IoT. © 2025 Elsevier B.V., All rights reserved.

Abstract
Conventional power distribution networks (PDNs), in which individual voltage regulators power the entire integrated circuit (IC), are ineffective for high-power, large-area ICs. In highperformance systems-on-chip (SoCs) and microprocessors (in particular those designed for AI applications), shrinking technology nodes are leading to higher current densities, which impose thermal constraints and limit the portion of the chip that can be simultaneously powered (“dark silicon”). PDNs with point-of-load regulation offer a promising alternative. The distributed nature of their design inherently relaxes thermal constraints while minimizing high-current routing overhead (IR drops), thereby improving the PDN efficiency. In this work, the concept of on-chip distributed voltage regulation is introduced. Previously reported distributed voltage regulator designs are reviewed, emphasizing their major achievements and limitations. Then, the challenges that hinder a more ubiquitous adoption of such designs, namely stability (analysis) and unbalanced load sharing, are discussed. Existing solutions addressing these challenges are also presented. Finally, a comparative analysis of the performance of these regulators is presented, and insights into the future direction of distributed voltage regulation are offered. © (2025), (Universidade do Porto - Faculdade de Engenharia). All rights reserved.

Abstract

Abstract
In many organizations with a distributed operation, not only is data collection distributed, but models are also developed and deployed separately. Understanding the combined knowledge of all the local models may be important and challenging, especially in the case of a large number of models. The automated development of consensus models, which aggregate multiple models into a single one, involves several challenges, including fidelity (ensuring that aggregation does not penalize the predictive performance severely) and completeness (ensuring that the consensus model covers the same space as the local models). In this paper, we address the latter, proposing two measures for geometrical and distributional completeness. The first quantifies the proportion of the decision space that is covered by a model, while the second takes into account the concentration of the data that is covered by the model. The use of these measures is illustrated in a real-world example of academic management, as well as four publicly available datasets. The results indicate that distributional completeness in the deployed models is consistently higher than geometrical completeness. Although consensus models tend to be geometrically incomplete, distributional completeness reveals that they cover the regions of the decision space with a higher concentration of data.