In wireless cellular or ad hoc networks where Quality of Service (QoS) is interference-limited, a variety of power control problems can be formulated as nonlinear optimization with a system-wide objective, e.g., maximizing the total system throughput or the worst user throughput, subject to QoS constraints from individual users, e.g., on data rate, delay, and outage probability. We show that in the high Signal-to- interference Ratios (SIR) regime, these nonlinear and apparently difficult, nonconvex optimization problems can be transformed into convex optimization problems in the form of geometric programming; hence they can be very efficiently solved for global optimality even with a large number of users. In the medium to low SIR regime, some of these constrained nonlinear optimization of power control cannot be turned into tractable convex formulations, but a heuristic can be used to compute in most cases the optimal solution by solving a series of geometric programs through the approach of successive convex approximation. While efficient and robust algorithms have been extensively studied for centralized solutions of geometric programs, distributed algorithms have not been explored before. We present a systematic method of distributed algorithms for power control that is geometric-programming-based. These techniques for power control, together with their implications to admission control and pricing in wireless networks, are illustrated through several numerical examples.

In recent years, AI technologies have been developed to promote students' self‐regulated learning (SRL) and proactive learning in digital learning environments. This paper discusses a comparative study between generative AI‐based (SRLbot) and rule‐based AI chatbots (Nemobot) in a 3‐week science learning experience with 74 Secondary 4 students in Hong Kong. The experimental group used SRLbot to maintain a regular study habit and facilitate their SRL, while the control group utilized rule‐based AI chatbots. Results showed that SRLbot effectively enhanced students' science knowledge, behavioural engagement and motivation. Quantile regression analysis indicated that the number of interactions significantly predicted variations in SRL. Students appreciated the personalized recommendations and flexibility of SRLbot, which adjusted responses based on their specific learning and SRL scenarios. The ChatGPT‐enhanced instructional design reduced learning anxiety and promoted learning performance, motivation and sustained learning habits. Students' feedback on learning challenges, psychological support and self‐regulation behaviours provided insights into their progress and experience with this technology. SRLbot's adaptability and personalized approach distinguished it from rule‐based chatbots. The findings offer valuable evidence for AI developers and educators to consider generative AI settings and chatbot design, facilitating greater success in online science learning. Practitioner notes: What is already known about this topic AI technologies have been used to support student self‐regulated learning (SRL) across subjects. SRL has been identified as an important aspect of student learning that can be developed through technological support. Generative AI technologies like ChatGPT have shown potential for enhancing student learning by providing personalized guidance and feedback. What this paper adds This paper reports on a case study that specifically examines the effectiveness of ChatGPT in promoting SRL among secondary students. The study provides evidence that ChatGPT can enhance students' science knowledge, motivation and SRL compared to a rule‐based AI chatbot. The study offers insights into how ChatGPT can be used as a tool to facilitate SRL and promote sustained learning habits. Implications for practice and/or policy The findings of this study suggest that educators should consider the potential of ChatGPT and other generative AI technologies to support student learning and SRL. Educators and students should be aware of the limitations of AI technologies and ensure that they are used appropriately to generate desired responses. It is also important to equip teachers and students with AI competencies to enable them to use AI for learning and teaching.

Transmit power in wireless cellular networks is a key degree of freedom in the management of interference, energy, and connectivity. Power control in both the uplink and downlink of a cellular network has been extensively studied, especially over the last 15 years, and some of the results have enabled the continuous evolution and significant impact of the digital cellular technology. This survey provides a comprehensive discussion of the models, algorithms, analysis, and methodologies in this vast and growing literature. It starts with a taxonomy of the wide range of power control problem formulations, and progresses from the basic formulation to more sophisticated ones. When transmit power is the only set of optimization variables, algorithms for fixed SIR are presented first, before turning to their robust versions and joint SIR and power optimization. This is followed by opportunistic and non-cooperative power control. Then joint control of power together with beamforming pattern, base station assignment, spectrum allocation, and transmit schedule is surveyed one-by-one. Throughout the survey, we highlight the use of mathematical language and tools in the study of power control, including optimization theory, control theory, game theory, and linear algebra. Practical implementations of some of the algorithms in operational networks are discussed in the concluding section. As illustrated by the open problems presented at the end of most chapters, in the area of power control in cellular networks, there are still many under-explored directions and unresolved issues that remain theoretically challenging and practically important.

This paper provides a comprehensive review of the literature concerning the utilization of Natural Language Processing (NLP) techniques, with a particular focus on transformer-based large language models (LLMs) trained using Big Code, within the domain of AI-assisted programming tasks. LLMs, augmented with software naturalness, have played a crucial role in facilitating AI-assisted programming applications, including code generation, code completion, code translation, code refinement, code summarization, defect detection, and clone detection. Notable examples of such applications include the GitHub Copilot powered by OpenAI’s Codex and DeepMind AlphaCode. This paper presents an overview of the major LLMs and their applications in downstream tasks related to AI-assisted programming. Furthermore, it explores the challenges and opportunities associated with incorporating NLP techniques with software naturalness in these applications, with a discussion on extending AI-assisted programming capabilities to Apple’s Xcode for mobile software development. This paper also presents the challenges of and opportunities for incorporating NLP techniques with software naturalness, empowering developers with advanced coding assistance and streamlining the software development process.

Electric vehicles (EVs) offer an attractive long-term solution to reduce the dependence on fossil fuel and greenhouse gas emission. At the same time, charging a large fleet of EVs distributed across the residential area poses a challenge for the distribution network. In this paper, we formulate this problem by building on the optimal power flow (OPF) framework to model the network constraints that arises from charging EVs at different locations. To overcome the computational challenge when the control horizon is long, we study a nested optimization approach to decompose the joint OPF and EV charging problem. We characterize the optimal EV charging schedule to be a valley-filling profile, which allows us to develop an efficient offline algorithm with significantly lower computational complexity compared to centralized interior point solvers. Furthermore, we propose a decentralized online algorithm that dynamically tracks the valley-filling profile. Our algorithms are evaluated on the IEEE 14 bus system with real residential load profiles, and the simulations show that our online algorithm performs almost optimally under different settings.

In the dynamic landscape of contemporary education, the evolution of teaching strategies such as blended learning and flipped classrooms has highlighted the need for efficient and effective generation of multiple-choice questions (MCQs). To address this, we introduce MCQGen , a novel generative artificial intelligence framework designed for the automated creation of MCQs. MCQGen uniquely integrates a large language model (LLM) with retrieval-augmented generation and advanced prompt engineering techniques, drawing from an extensive external knowledge base. This integration significantly enhances the ability of the LLM to produce educationally relevant questions that align with both the goals of educators and the diverse learning needs of students. The framework employs innovative prompt engineering, combining chain-of-thought and self-refine prompting techniques, to enhance performance of the LLM. This process leads to the generation of questions that are not only contextually relevant and challenging but also reflective of common student misconceptions, contributing effectively to personalized learning experiences and enhancing student engagement and understanding. Our extensive evaluations showcase the effectiveness of MCQGen in producing high-quality MCQs for various educational needs and learning styles. The framework demonstrates its potential to significantly reduce the time and expertise required for MCQ creation, marking its practical utility in modern education. In essence, MCQGen offers an innovative and robust solution for the automated generation of MCQs, enhancing personalized learning in the digital era.