Architectural Decision Records (ADRs) play a central role in maintaining software architecture quality, yet many decision violations go unnoticed because projects lack both systematic documentation and automated detection mechanisms. Recent advances in Large Language Models (LLMs) open up new possibilities for automating architectural reasoning at scale. We investigated how effectively LLMs can identify decision violations in open-source systems by examining their agreement, accuracy, and inherent limitations. Our study analyzed 980 ADRs across 109 GitHub repositories using a multi-model pipeline in which one LLM primary screens potential decision violations, and three additional LLMs independently validate the reasoning. We assessed agreement, accuracy, precision, and recall, and complemented the quantitative findings with expert evaluation. The models achieved substantial agreement and strong accuracy for explicit, code-inferable decisions. Accuracy falls short for implicit or deployment-oriented decisions that depend on deployment configuration or organizational knowledge. Therefore, LLMs can meaningfully support validation of architectural decision compliance; however, they are not yet replacing human expertise for decisions not focused on code.

The rapid diffusion of Large Language Models (LLMs) is fundamentally changing how Mining Software Repositories (MSR) research is conducted, particularly for studies that rely on unstructured textual artifacts such as commit messages, issue discussions, pull request reviews, and practitioner-generated content. While recent work has demonstrated the potential of LLMs to support classification, summarization, and qualitative analysis tasks, the majority of existing approaches rely on interactive or API-based executions [2, 3, 6]. Such execution models are poorly suited for large-scale empirical MSR studies, where thousands or hundreds of thousands of artifacts must be processed in a controlled, reproducible, and cost-aware manner.

The rapid integration of Artificial Intelligence (AI) – including Machine Learning (ML) and Generative AI – into software systems is reshaping the software development lifecycle. As AI-driven systems become more dynamic and complex, traditional approaches to Technical Debt (TD) management face increasing limitations. Simultaneously, AI-assisted development introduces new forms of TD, particularly in relation to maintainability, explainability, and data governance. This study aims to explore how Technical Debt Management (TDM) must adapt in the context of AI-enhanced software development. It investigates (1) the evolution of TD in AI-driven systems, and (2) the implications of using AI technologies within the software engineering process. We conducted a multivocal literature review, combining insights from both peer-reviewed research and industry sources. Following established guidelines, we systematically analyzed 61 primary sources, categorized TD types and management activities, and identified key challenges and practices emerging in the AI era. Our findings reveal that data-related, infrastructure, and pipeline-related TD are particularly prevalent in ML systems. Machine Learning Operations (MLOps) practices are increasingly recognized as essential for managing such debt, especially in relation to dynamic data dependencies and model retraining. In parallel, AI-generated artifacts and automated pipelines introduce new governance and maintainability challenges. Technical Debt in AI systems demands continuous, automated, and cross-functional management strategies. As software evolves in response to data and usage, new operational paradigms – grounded in practices like MLOps and Small Language Model Operations (SLMOps) – will be vital to ensure long-term software sustainability. This study provides a foundational map for researchers and practitioners navigating the intersection of AI and TD management. • Data-centric AI systems introduce new forms of TD in data, infrastructure, and governance. • MLOps is often assumed in research, while its practices and security concerns are overlooked. • Gray literature captures real-world data debt practices absent in academic sources. • Prompt and explainability debt are rising issues in GenAI with little formal support. • SLMOps may offer future-ready frameworks for managing lightweight AI pipelines.

The Orthogonal Frequency Division Multiplexing (OFDM) has been very successful in terrestrial communication, but it is challenged in underwater acoustic (UWA) channels. This is because OFDM was engineered for under-spread channels, while UWA channels are intrinsically over-spread. In this paper, we will represent the channel in the scale-delay (S-D) domain and propose the Mellin-Frequency Division Multiplexing (MFDM) modulation scheme. When represented in the S-D domain, the received signal is equal to the$\omega$-convolution of the transmitted signal and the channel response. We can then convert the signals and channel response to the Mellin-Frequency (M-F) domain through the Mellin-Fourier transform. The$\omega$-convolution thus becomes point-wise multiplication, leading to low-complexity detection. Based on this motivation, we model the system in a continuous form and then propose a discretized approach for practical system implementation. The symbol error rate (SER) performance of MFDM is compared with two existing modulation schemes: Orthogonal Delay Scale Space (ODSS) and Cyclic Prefix-OFDM (CP-OFDM). It is not surprising to see that CP-OFDM performs the worst. As for the other two modulation schemes, ODSS is based on scale-delay domain equalization, and the interference among symbols resulting from the$\omega$-convolution was ignored. In comparison, the equalization was conducted in the Mellin-Frequency domain for the proposed MFDM, leading to improved SER performance.

Fast underwater localization serves as a critical application of Internet-of-Underwater-Things (IoUT), such as underwater search and rescue. Narrowband sinusoidal pulses are very commonly used in locator beacons installed on flight recorders. A mobile anchor (e.g., autonomous underwater vehicle (AUV)) has to keep receiving the signal and measuring Doppler shift for a long time, so that enough information can be collected for reliable localization. The need of a long observation window is deeply rooted in the fact that the Doppler shift measurements are highly correlated when they are taken at closely located spots. In this paper, we will show that by replacing the narrowband beacon signal with a wideband one, high-accuracy positioning can be achieved within a short period of time. The basic idea is to simultaneously measure Doppler shift and time of arrival (ToA) from wideband signals, and the errors spaces corresponding to these two measurements are complementary even when they are taken at the same position. The Cramér-Rao bound (CRB) will be derived for such a system. In low-signal-to-noise ratio (SNR) regime, the observation window has to be prolonged for effective information extraction, and we will see that the wideband signals still have an edge over the narrowband signals in positioning accuracy. Efficient algorithms are designed for positioning and the closed-form positioning error is derived. We also show that the localization accuracy will experience a significant drop when ToA is replaced by time difference of arrival (TDoA), because the perfect complementation no longer holds. The performance of the proposed algorithm, along with corresponding comparisons, is verified through simulations over various parameters such as SNR, number of measurements, and length of observation window, etc.

In the era of 6G, developing and managing software requires cutting-edge software engineering (SE) theories and practices tailored for such complexity across a vast number of connected edge devices. Our project aims to lead the development of sustainable methods and energy-efficient orchestration models specifically for edge environments, enhancing architectural support driven by AI for contemporary edge-to-cloud continuum computing. This initiative seeks to position Finland at the forefront of the 6G landscape, focusing on sophisticated edge orchestration and robust software architectures to optimize the performance and scalability of edge networks. Collaborating with leading Finnish universities and companies, the project emphasizes deep industry-academia collaboration and international expertise to address critical challenges in edge orchestration and software architecture, aiming to drive significant advancements in software productivity and market impact.

Researchers in the microservices community often resort to demonstrating the impact of their proposed advancements on custom-made microservices projects. This is a possible source of bias that can reduce the trustworthiness of the results. Moreover, it is hard to compare advances in small projects, often developed due to lack of time. It is common across disciplines to recognize benchmarks that mitigate bias and unify the advancements' impact. To facilitate the identification of available open-source microservice projects (OSS-MS), we performed a comprehensive study to identify, curate, and catalog OSS-MS. We started with 389559 projects and filtered them down to 3804 projects that we manually labeled. After manual labeling, our dataset contains 378 projects with three or more microservices and with over 100 commits. We document the projects from many perspectives, including project size, platform, number of contributors, project purpose, and foundation support. This dataset can serve researchers as a roadmap to identify benchmarks, as our dataset can be used to answer questions such as whether the number of services impacts the issue count.

Internet of underwater things (IoUT) attracts many interests in these years both in academia and industry, such as marine data collection, pollution monitoring, and offshore exploration. As a fundamental issue of IoUT, the underwater acoustic channel experiences long delay and temporal-spatial uncertainty compared with terrestrial communications and networks. It is difficult to capture full characteristics of the underwater acoustic channel by statistical models. In this paper, we investigate the properties of acoustic propagation in seawater and different underwater acoustic channel models. Moreover, we survey five underwater acoustic channel models, including ray-theoretical model, normal mode model, multipath expansion model, fast-field model, and parabolic equation model, which are the corresponding solutions of the wave equation. We conclude the paper with the characteristics of each model in terms of different aspects.

Wireless sensor network (WSN), consisting of several sensor nodes, is one of the most promising technologies emerged in the past decade. The positioning system for WSN is particularly meaningful and widely used in the military surveillance, air-sea rescue, traffic monitoring, and etc. However, the traditional positioning system always suffers from deployment and maintenance of anchors. In this paper, we propose a positioning system employing a Raspberry Pi platform attached to a DJI drone as a mobile anchor. The DJI drone can serve as multiple virtual anchors by moving and broadcasting its location information periodically. Thus, it is possible to localize sensor node by itself when the sensor node collects the drone's position. A Gauss-Newton method is applied to improve the accuracy of the proposed positioning system. We also elaborate the adaption of the Gauss-Newton method with the geodetic coordinates. The goal of the proposed positioning system is to achieve higher accuracy and higher coverage at lower cost.

As the data traffic in future wireless communications will explosively grow up to 1000-fold by the deployment of 5G, several technologies are emerging to satisfy this demand, including multiple-input multiple-output (MIMO), millimeter wave communications, Non-orthogonal Multiple Access (NOMA), etc. Millimeter wave communication is a promising solution since it can provide tens of GHz bandwidth by fundamentally exploring higher unoccupied spectrum resources. As the wavelength of higher frequency shrinks, it is possible to design more compact antenna array with large number of antennas with independent RF (Radio Frequency) chains, causing high cost and complexity. By exploring the spatial sparsity of the millimeter wave channels, lens antenna array has been investigated recently as a promising choice with limited RF chains and low complexity. In this paper, we investigate the outage probability for highway communication systems with lens antenna array, under overtaking scenario, where high mobility of users is expected. When a vehicle is trying to pass another one, the channels between these two vehicles and the RSU (Road Side Unit) are unresolvable, thus causing outage for at least tens of symbol durations. We apply power-domain NOMA in this scenario, where these two users are paired by a threshold derived with the QoS of each user, to alleviate this problem and achieve low outage probability.

In this paper, we investigate low-complexity data detection scheme for massive multiple-input multiple-output (MIMO) uplink transmission. We propose to utilize the stair matrix, instead of diagonal matrix in existing proposals, for the development, and achieve near linear minimum mean-square error detection performance. We first demonstrate the applicability of the proposed method by showing that the probability (that the convergence conditions are met) approaches one as long as sufficiently large number of antennas are equipped at the base station. We then propose an iterative method to perform data detection and show that much improved performance can be achieved with the computational complexity remaining at the same level of existing iterative methods, where the diagonal matrix is adopted. Furthermore, we conduct numerical simulations, and the results validate the significant performance enhancement of using the stair matrix over the diagonal matrix in all performance aspects. Moreover, we apply the proposed scheme to a massive MIMO system, where the extended vehicular A channel data are generated. The performance improvement of the proposed scheme over existing proposals is also validated.

Abstract In addition to the requirements of the terrestrial sensor network where performance metrics such as throughput and packet delivery delay are often emphasized, energy efficiency becomes an even more significant and challenging issue in underwater acoustic sensor networks, especially when long‐term deployment is required. In this paper, we tackle the problem of energy conservation in underwater acoustic sensor networks for long‐term marine monitoring applications. We propose an asynchronous wake‐up scheme based on combinatorial designs to minimize the working duty cycle of sensor nodes. We prove that network connectivity can be properly maintained using such a design even with a reduced duty cycle. We study the utilization ratio of the sink node and the scalability of the network using multiple sink nodes. Simulation results show that the proposed asynchronous wake‐up scheme can effectively reduce the energy consumption for idle listening and can outperform other cyclic difference set‐based wake‐up schemes. More significantly, high performance is achieved without sacrificing network connectivity. Copyright © 2015 John Wiley & Sons, Ltd.

In recent years, underwater acoustic sensor networks (UWSNs) are envisioned for different potential applications, ranging from long-term marine environmental monitoring, industrial instrumentation control, to military surveillance and security. Compared to wireless sensor networks (WSNs), energy-efficient data transmission becomes more critical in UWSNs due to non-rechargeable batteries of sensor nodes with limited amount of energies in long-term marine monitoring applications. Besides, in underwater acoustic communications, transmitting and receiving power levels dominate the energy consumption during the data transfer. Data packet retransmission caused by network deployment error increases the energy consumption and reduce the network lifetime. In this paper, we investigate a two-dimensional deployment strategy of UWSNs with a square grid topology. We present a mathematical model to study the deployment error of UWSNs. Based on this model, a parameter, i.e., α, is introduced to balance the network robustness and the energy consumption of sensor nodes. α is defined as the ratio of the transmission range of a sensor node to the distance between two closest adjacent sensor nodes. We report optimal values of α corresponding to this balance for different sizes of UWSNs.

Underwater acoustic sensor network (UWSN) has many important future applications in environmental, natural resources development, and geological oceanography. The recent advancement in underwater acoustic networking technologies and autonomous underwater vehicle (AUV) technologies enables the new underwater networking paradigm of using the AUV as the mobile sink for data collection. In this paper, we propose a coordination scheme for data gathering in UWSN using AUV. The mobility of AUV makes communication between AUV and sensor nodes challenging and it is difficult to guarantee that sensor nodes will be able to come out of sleep mode for communication. With different energy consumption requirements and constraints, the operation of AUV includes sending beacon messages, channel detection, and data reception, whereas that for the underwater sensor nodes involves prolonged sleep mode with sporadic data transmission. Therefore, different cycle periods would be required for AUV and sensor nodes. A new scheme is proposed and the shortest wakeup time and optimal amount of sleeping time for sensor nodes are investigated. Moreover, we demonstrate the effectiveness of our scheme when time synchronization does not exist between AUV and sensor nodes. Furthermore, transmission power control by using received signal strength (RSS) is introduced in order to decrease energy consumption and increase communication reliability. Simulation results show that the proposed scheme with power control leads to relative low energy consumption during communication under the harsh underwater environment compared with that without power control.

Underwater acoustic communications in the ocean is complicated as the acoustic signals may be attenuated, distorted and delayed. In this paper, we review the underwater acoustic signal propagation properties in terms of sound speed profile, spreading loss and absorption loss. We study and compare different approaches on the calculation of signal transmission loss in the water, more specifically, the ray theory model approach and the semi-empirical formula approach. Using the Acoustic Toolbox, we compare their performance under different environmental parameters, including the sound source depth, bathymetry data, and the horizontal distance between the sound source and receiver. Furthermore, in order to obtain how the acoustic propagation characteristics will affect the performance of medium access control (MAC) protocol, we adopt pure ALOHA protocol and use network simulator ns-2 to study the throughput performance under both shallow water and deep ocean conditions. Our results indicate that the transmission loss in the shallow water is close to the result of semi-empirical formula with transition region (k = 1.5), which is close to the result of semi-empirical formula with spherical spreading loss (k=2) in deep water.