Scalability Of System Research Articles

Enhancing generalization in Federated Learning with heterogeneous data: A comparative literature review

Federated Learning (FL) is a collaborative training paradigm whereby a global Machine Learning (ML) model is trained using typically private and distributed data sources without disclosing the raw data. The approach paves the way for better privacy guarantees, improved overall system scalability, and sustainability. In this context, Federated Averaging (FedAvg) is a representative FL algorithm adopting a client–server protocol that operates in synchronous rounds, where selected learners contribute to the global model via local model updates, trained using their private data, while a server entity aggregates the local contributions, producing the new-generation global model as a weighted average of the local ones. However, when clients possess (highly) dissimilar data, the FedAvg technique becomes ineffective due to divergence in client models. Consequently, FedAvg-trained models struggle to generalize when presented with unseen data from the global distribution. In this research paper, we conduct a systematic review of state-of-the-art approaches proposed to counteract global model performance degradation in the presence of heterogeneous data.To this end, we compile an original taxonomy, highlighting the main algorithmic approaches and mechanisms behind each identified category. Advancing the current body of knowledge, we empirically evaluate the generalization performance on visual tasks of various methods under moderate and significant levels of data heterogeneity, as common practice within the surveyed literature. In addition, the paper benchmarks the performance of hybrid techniques, resulting as a combination of client- and server-side algorithmic tweaks, by shedding light on some associated performance tradeoffs. While recognizing other relevant issues in FL, such as device heterogeneity and energy consumption, which have a non-negligible impact on the learning process, these well-investigated topics are not the main focus of this article.

Implementation of Service Oriented Architecture in Mobile Applications to Improve System Flexibility, Interoperability, and Scalability

This research aims to apply Service Oriented Architecture (SOA) to mobile applications to improve the flexibility, interoperability, and scalability of the system. SOA is an approach that enables better integration between different components in an application architecture. Mobile applications, as an integral part of everyday life, require high flexibility and adaptability to changing user needs. The research method used involves application requirements analysis, SOA architecture design, and implementation on an existing mobile application. Through the implementation of SOA, resource management and service distribution become more efficient, making it easy to add or change features without affecting the entire system. Test results show improved application performance and responsiveness, along with the ability to easily integrate services from multiple sources. In addition, the SOA architecture provides flexibility in the development and maintenance of mobile applications, accelerating the development process and minimizing the impact of changes. This research contributes to the development of mobile applications by introducing the SOA approach as an architectural foundation. The implementation of SOA in mobile applications is expected to guide developers in creating systems that are more resilient, adaptive, and easy to integrate with rapidly evolving software environments.

Solar-driven interfacial evaporation: materials design and device assembly

Solar-driven interfacial evaporation (SIE) is an emerging research topic that is gaining attention due to its potential in addressing global water scarcity issues. This review provides a comprehensive overview of base materials, recent innovations in photothermal materials and the design of evaporators for effective water desalination and purification. The recent development of SIE is meticulously discussed, providing a deep understanding of the key performance indicators and state-of-the-art materials. Additionally, this review examines novel strategies that have been reported in the literature for enhancing the efficiency and scalability of SIE systems. These strategies involve using photothermal materials and exploring innovative device configurations. Finally, we discuss the existing challenges and future research directions, emphasizing the potential of SIE in addressing global water scarcity and contributing to a sustainable future.

Starling: An I/O-Efficient Disk-Resident Graph Index Framework for High-Dimensional Vector Similarity Search on Data Segment

High-dimensional vector similarity search (HVSS) is gaining prominence as a powerful tool for various data science and AI applications. As vector data scales up, in-memory indexes pose a significant challenge due to the substantial increase in main memory requirements. A potential solution involves leveraging disk-based implementation, which stores and searches vector data on high-performance devices like NVMe SSDs. However, implementing HVSS for data segments proves to be intricate in vector databases where a single machine comprises multiple segments for system scalability. In this context, each segment operates with limited memory and disk space, necessitating a delicate balance between accuracy, efficiency, and space cost. Existing disk-based methods fall short as they do not holistically address all these requirements simultaneously. In this paper, we present Starling, an I/O-efficient disk-resident graph index framework that optimizes data layout and search strategy within the segment. It has two primary components: (1) a data layout incorporating an in-memory navigation graph and a reordered disk-based graph with enhanced locality, reducing the search path length and minimizing disk bandwidth wastage; and (2) a block search strategy designed to minimize costly disk I/O operations during vector query execution. Through extensive experiments, we validate the effectiveness, efficiency, and scalability of Starling. On a data segment with 2GB memory and 10GB disk capacity, Starling can accommodate up to 33 million vectors in 128 dimensions, offering HVSS with over 0.9 average precision and top-10 recall rate, and latency under 1 millisecond. The results showcase Starling's superior performance, exhibiting 43.9x higher throughput with 98% lower query latency compared to state-of-the-art methods while maintaining the same level of accuracy.

Implementing Internet of Things Service Platforms with Network Function Virtualization Serverless Technologies

The need for adaptivity and scalability in telecommunication systems has led to the introduction of a software-based approach to networking, in which network functions are virtualized and implemented in software modules, based on network function virtualization (NFV) technologies. The growing demand for low latency, efficiency, flexibility and security has placed some limitations on the adoption of these technologies, due to some problems of traditional virtualization solutions. However, the introduction of lightweight virtualization approaches is paving the way for new and better infrastructures for implementing network functions. This article discusses these new virtualization solutions and shows a proposal, based on serverless computing, that uses them to implement container-based virtualized network functions for the delivery of advanced Internet of Things (IoT) services. It includes open source software components to implement both the virtualization layer, implemented through Firecracker, and the runtime environment, based on Kata containers. A set of experiments shows that the proposed approach is fast, in order to boost new network functions, and more efficient than some baseline solutions, with minimal resource footprint. Therefore, it is an excellent candidate to implement NFV functions in the edge deployment of serverless services for the IoT.

The Effective Fragment Molecular Orbital Method: Achieving High Scalability and Accuracy for Large Systems.

The effective fragment molecular orbital (EFMO) method has been developed to predict the total energy of a very large molecular system accurately (with respect to the underlying quantum mechanical method) and efficiently by taking advantage of the locality of strong chemical interactions and employing a two-level hierarchical parallelism. The accuracy of the EFMO method is partly attributed to the accurate and robust intermolecular interaction prediction between distant fragments, in particular, the many-body polarization and dispersion effects, which require the generation of static and dynamic polarizability tensors by solving the coupled perturbed Hartree-Fock (CPHF) and time-dependent HF (TDHF) equations, respectively. Solving the CPHF and TDHF equations is the main EFMO computational bottleneck due to the inefficient (serial) and I/O-intensive implementation of the CPHF and TDHF solvers. In this work, the efficiency and scalability of the EFMO method are significantly improved with a new CPU memory-based implementation for solving the CPHF and TDHF equations that are parallelized by either message passing interface (MPI) or hybrid MPI/OpenMP. The accuracy of the EFMO method is demonstrated for both covalently bonded systems and noncovalently bound molecular clusters by systematically examining the effects of basis sets and a key distance-related cutoff parameter, Rcut. Rcut determines whether a fragment pair (dimer) is treated by the chosen ab initio method or calculated using the effective fragment potential (EFP) method (separated dimers). Decreasing the value of Rcut increases the number of separated (EFP) dimers, thereby decreasing the computational effort. It is demonstrated that excellent accuracy (<1 kcal/mol error per fragment) can be achieved when using a sufficiently large basis set with diffuse functions coupled with a small Rcut value. With the new parallel implementation, the total EFMO wall time is substantially reduced, especially with a high number of MPI ranks. Given a sufficient workload, nearly ideal strong scaling is achieved for the CPHF and TDHF parts of the calculation. For the first time, EFMO calculations with the inclusion of long-range polarization and dispersion interactions on a hydrated mesoporous silica nanoparticle with explicit water solvent molecules (more than 15k atoms) are achieved on a massively parallel supercomputer using nearly 1000 physical nodes. In addition, EFMO calculations on the carbinolamine formation step of an amine-catalyzed aldol reaction at the nanoscale with explicit solvent effects are presented.

Scalable evaluation methods for autonomous vehicles

Effective intelligent driving test and evaluation methods can improve the development and deployment process of autonomous vehicles (AVs). However, due to the extreme complexity and high dimensionality of driving behavior, how to objectively and efficiently evaluate the multi-dimensional performance of AVs in simulation and real-world environments is a long-standing problem. This paper proposes an objective multi-dimensional comprehensive evaluation (OMDCE) method that divides the intelligent driving test into four modules: test scenarios, scenario complexity model, simulation test platform, and automated evaluation system. The scenario complexity model is proposed to bridge the gap between the test scenarios and the automated evaluation system, thus enabling adaptive scaling of the evaluation scale for test scenarios with different difficulty levels. Besides, based on the existing four evaluation metrics for ego performance, altruism performance evaluation is first proposed to comprehensively portray AVs’ intelligence degree. The OMDCE method was evaluated by testing two different intelligent driving algorithms and conducting real vehicle testing. The experimental results demonstrated that the OMDCE method can effectively evaluate AVs in various scenarios and quantitatively measure their multi-dimensional performance. The proposed method reduces the subjectivity of manual evaluation and speeds up the evaluation. Moreover, it improves the universality and scalability of the evaluation metrics system.

A scalable MapReduce-based design of an unsupervised entity resolution system.

Traditional data curation processes typically depend on human intervention. As data volume and variety grow exponentially, organizations are striving to increase efficiency of their data processes by automating manual processes and making them as unsupervised as possible. An additional challenge is to make these unsupervised processes scalable to meet the demands of increased data volume. This paper describes the parallelization of an unsupervised entity resolution (ER) process. ER is a component of many different data curation processes because it clusters records from multiple data sources that refer to the same real-world entity, such as the same customer, patient, or product. The ability to scale ER processes is particularly important because the computation effort of ER increases quadratically with data volume. The Data Washing Machine (DWM) is an already proposed unsupervised ER system which clusters references from diverse data sources. This work aims at solving the single-threaded nature of the DWM by adopting the parallelization nature of Hadoop MapReduce. However, the proposed parallelization method can be applied to both supervised systems, where matching rules are created by experts, and unsupervised systems, where expert intervention is not required. The DWM uses an entropy measure to self-evaluate the quality of record clustering. The current single-threaded implementations of the DWM in Python and Java are not scalable beyond a few 1,000 records and rely on large, shared memory. The objective of this research is to solve the major two shortcomings of the current design of the DWM which are the creation and usage of shared memory and lack of scalability by leveraging on the power of Hadoop MapReduce. We propose Hadoop Data Washing Machine (HDWM), a MapReduce implementation of the legacy DWM. The scalability of the proposed system is displayed using publicly available ER datasets. Based on results from our experiment, we conclude that HDWM can cluster from 1,000's to millions of equivalent references using multiple computational nodes with independent RAM and CPU cores.

Multi-Level Split Federated Learning for Large-Scale AIoT System Based on Smart Cities

In the context of smart cities, the integration of artificial intelligence (AI) and the Internet of Things (IoT) has led to the proliferation of AIoT systems, which handle vast amounts of data to enhance urban infrastructure and services. However, the collaborative training of deep learning models within these systems encounters significant challenges, chiefly due to data privacy concerns and dealing with communication latency from large-scale IoT devices. To address these issues, multi-level split federated learning (multi-level SFL) has been proposed, merging the benefits of split learning (SL) and federated learning (FL). This framework introduces a novel multi-level aggregation architecture that reduces communication delays, enhances scalability, and addresses system and statistical heterogeneity inherent in large AIoT systems with non-IID data distributions. The architecture leverages the Message Queuing Telemetry Transport (MQTT) protocol to cluster IoT devices geographically and employs edge and fog computing layers for initial model parameter aggregation. Simulation experiments validate that the multi-level SFL outperforms traditional SFL by improving model accuracy and convergence speed in large-scale, non-IID environments. This paper delineates the proposed architecture, its workflow, and its advantages in enhancing the robustness and scalability of AIoT systems in smart cities while preserving data privacy.

Unified ensemble federated learning with cloud computing for online anomaly detection in energy-efficient wireless sensor networks

Anomaly detection in Wireless Sensor Networks (WSNs) is critical for their reliable and secure operation. Optimizing resource efficiency is crucial for reducing energy consumption. Two new algorithms developed for anomaly detection in WSNs—Ensemble Federated Learning (EFL) with Cloud Integration and Online Anomaly Detection with Energy-Efficient Techniques (OAD-EE) with Cloud-based Model Aggregation. EFL with Cloud Integration uses ensemble methods and federated learning to enhance detection accuracy and data privacy. OAD-EE with Cloud-based Model Aggregation uses online learning and energy-efficient techniques to conserve energy on resource-constrained sensor nodes. By combining EFL and OAD-EE, a comprehensive and efficient framework for anomaly detection in WSNs can be created. Experimental results show that EFL with Cloud Integration achieves the highest detection accuracy, while OAD-EE with Cloud-based Model Aggregation has the lowest energy consumption and fastest detection time among all algorithms, making it suitable for real-time applications. The unified algorithm contributes to the system's overall efficiency, scalability, and real-time response. By integrating cloud computing, this algorithm opens new avenues for advanced WSN applications. These promising approaches for anomaly detection in resource constrained and large-scale WSNs are beneficial for industrial applications.

CROWDMATCH: Optimizing Crowdsourcing Matching through the Integration of Matching Theory and Coalition Games

This paper tackles the challenges inherent in crowdsourcing dynamics by introducing the CROWDMATCH mechanism. Aimed at enabling crowdworkers to strategically select suitable crowdsourcers while contributing information to crowdsourcing tasks, CROWDMATCH considers incentives, information availability and cost, and the decisions of fellow crowdworkers to model the utility functions for both the crowdworkers and the crowdsourcers. Specifically, the paper presents an initial Approximate CROWDMATCH mechanism grounded in matching theory principles, eliminating externalities from crowdworkers’ decisions and enabling each entity to maximize its utility. Subsequently, the Accurate CROWDMATCH mechanism is introduced, which is initiated by the outcome of the Approximate CROWDMATCH mechanism, and coalition game-theoretic principles are employed to refine the matching process by accounting for externalities. The paper’s contributions include the introduction of the CROWDMATCH system model, the development of both Approximate and Accurate CROWDMATCH mechanisms, and a demonstration of their superior performance through comprehensive simulation results. The mechanisms’ scalability in large-scale crowdsourcing systems and operational advantages are highlighted, distinguishing them from existing methods and highlighting their efficacy in empowering crowdworkers in crowdsourcer selection.

Load Balancing using Weight Based scheme in AWS

This paper presents research on load balancing methods for cloud computing platforms. Load balancing is an important element of cloud systems since it allows for optimal resource utilization and consistent performance levels. The study investigates several load balancing technologies and evaluates their effectiveness in coping with the dynamic and heterogeneous nature of cloud workloads. The study conducts experiments and analyses to discover significant parameters influencing load balancing performance, such as workload allocation, server capacity, and network latency. With the weight-based technique, different servers are dynamically given requests based on their weights, which are determined by the processing capacities of the servers. By distributing the work proportionately, this approach aims to decrease user request response times and avoid server overload. The effectiveness of the proposed approach is evaluated via simulations, demonstrating its potential to improve system performance and scalability relative to traditional load balancing methods. Generally speaking, this paper presents an optimized load distribution approach that raises the efficiency and stability of cloud computing systems. The findings enable to improve load balancing methods matched to the particular requirements and issues of cloud computing, eventually enhancing system scalability, reliability, and user

A comparative study between zero-knowledge proof (ZKP) and ring signatures targeting the legal and regulatory implications with the general data protection law (GDPL) and general data protection regulation (GDPR)

Privacy protection ensures that individuals have control over personal data, preventing abuse and preserving trust in the use of online services. In the “Digital Era”, where the collection, storage and processing of personal information have become ubiquitous, data privacy emerges as a relevant topic. In this sense, laws were created, such as the General Data Protection Law (LGPD) in Brazil and the General Data Protection Regulation (GDPR) in Europe, to control privacy and the processing of personal data. The article presents a comparative analysis of 2 (two) data privacy mechanisms, the Zero-Knowledge Proof (ZKP) and Ring Signatures, used in Blockchain, aiming at the legal and regulatory implications with the LGPD and GDPR. The comparative study between ZKP and Ring Signatures highlights the flexibility of ZKP in various contexts, including voting and secure authentication systems, while Ring Signatures offer significant advantages in terms of scalability and efficiency in systems where subscriber anonymity is considered fundamental. Furthermore, the legal and regulatory implications of the ZKP are discussed, mainly in relation to LGPD and GDPR. Finally, the article concludes that the comparative analysis offers insights into applications, challenges and legal and regulatory implications, particularly in relation to data privacy and compliance with regulations such as LGPD and GDPR.

A 0.67 μV-IIRN super-T Ω-Z IN 17.5 μW/Ch Active Electrode With In-Channel Boosted CMRR for Distributed EEG Monitoring.

We present the design, development, and experimental characterization of an active electrode (AE) IC for wearable ambulatory EEG recording. The proposed architecture features in-AE double common-mode (CM) rejection, making the recording's CMRR independent of typically-significant AE-to-AE gain variations. Thanks to being DC coupled and needless of chopper stabilization for flicker noise suppression, the architecture yields a super-T Ω input impedance. Such a large input impedance makes the AE's CMRR practically immune to electrode-skin interface impedance variations across different recording channels, a critical feature for dry-electrode ambulatory systems. Signal quantization and serialization are also performed in-AE, which enables a distributed system in which all AEs use a single data bus for data/command communication to the backend module, thus significantly improving the system's scalability. Additionally, the presented AE hosts auxiliary modules for (i) detection of an unstable electrode-skin connection through continuous interface impedance monitoring, (ii) dynamic measurement and adjustment of input DC level, and (iii) a CM feedback loop for further CMRR enhancement. The article also presents the development of printed (extrusion) tattoo electrodes and their experimental characterization results with the proposed AE architecture. Besides bio-compatibility, low-cost, pattern flexibility, and quick fabrication process, the printed electrodes offer a very stable electrode-skin connection, conform to scalp shape, and exhibit consistent performance under various bending curvatures. Analog circuit blocks of the presented AE architecture are designed and fabricated using a standard 180 nm CMOS technology, and the [Formula: see text] IC is integrated with off-chip low-power digital modules on a PCB to form the AE. Our measurement results show a CMRR of 82.2 dB (at 60 Hz), amplification voltage gain of 52.8 dB, a bandwidth of 0.2-400 Hz, ±500 mV input DC offset tolerance, An input impedance [Formula: see text], and 0.67 μV RMS integrated input referred noise (0.5-100 Hz), while consuming 17.5 μW per channel. All auxiliary modules are tested experimentally, and the entire system is validated in-vivo, for both ECG and EEG recording.

Optimizing passenger vehicle travel time with Model Predictive Control in multi-region traffic networks

Abstract This study investigates the impact of truck traffic on passenger vehicles in an urban network. Utilizing the Macroscopic Fundamental Diagram (MFD), a methodology to calculate the travel time spent (TTS) by passenger vehicles has been developed. To address this issue, an optimal control problem was formulated and solved using a Model Predictive Control (MPC) approach. The MPC framework has been applied in a centralized manner, to manage accumulation for various modes. To explore different traffic management strategies, the centralized MPC technique was implemented in two distinct configurations: region-based and vehicle-based approaches. It has been tested for various vehicle mixes and multiple control scenarios to assess the effectiveness in reducing passenger travel time spent (PTTS) and vehicle accumulation. The results demonstrate that the vehicle-based MPC approach tends to minimize the number of vehicles more effectively compared to the region-based approach. However, in terms of reducing passenger travel time, the region-based approach outperforms the vehicle-based strategy. This is attributed to enhanced coordination among traffic flow controllers, highlighting the importance of strategic controller interactions in urban traffic management systems. This research enhances both the theoretical framework for optimizing traffic flow and provides valuable practical insights for city planners and engineers aiming to deploy advanced traffic management strategies. Future studies could explore the scalability of these control systems and their capability to integrate real-time traffic data.

Suppression of electromagnetic crosstalk by differential excitation for SAW generation

Surface acoustic waves (SAWs) hold a vast potential in various fields such as spintronics, quantum acoustics, and electron-quantum optics, but an electromagnetic wave emanating from SAW generation circuits has often been a major hurdle. Here, we investigate a differential excitation method of interdigital transducers to generate SAWs while reducing the electromagnetic wave. The results show that electromagnetic waves are suppressed by more than 90 % in all directions. This suppression overcomes the operating limits and improves the scalability of SAW systems. Our results promise to facilitate the development of SAW-based applications in a wide range of research fields.

Dynamic relay node selection and routing for cloud-native Software Defined WANs

Cloud Native Software Defined Wide Area Networks (CN-WANs) are proposed to optimize the network performance based on the application requirement. In this paper, we propose a Relay Node Selection and Routing (RNSR) approach to minimize the number of relay nodes in CN-WANs under the end-to-end (E2E) transmission latency constraint. By doing so, we can not only ensure the Service Level Agreement (SLA) to applications, but also reduce the network operation expenses. To improve the system scalability, several algorithms are proposed for networks with different sizes. According to experiments on a prototype built on EC2 and extensive simulations in a custom in-house simulator, we demonstrate that RNSR has a superior performance on reducing the number of required relay nodes, outperforming the state-of-the-art benchmarks by up to 92.48% specifically. In addition, the algorithms designed for large scale networks will not suffer too much performance degradation but significantly improve the system scalability.

A Scalable Monitoring System for Software Defined Networks

Monitoring functionality is an essential element of any network system. Traditional monitoring solutions are mostly used for manual and infrequent network management tasks. Software-defined networks (SDN) have emerged with enabled automatic and frequent network reconfigurations. In this paper, a scalable monitoring system for SDN is introduced. The proposed system monitors small, medium, and large-scale SDN. Multiple instances of the proposed monitoring system can run in parallel for monitoring many SDN slices. The introduced monitoring system receives requests from network management applications, collects considerable amounts of measurement data, processes them, and returns the resulting knowledge to the network management applications. The proposed monitoring system slices the network (switches and links) into multiple slices. The introduced monitoring system concurrently monitors applications for various tenants, with each tenant's application running on a dedicated network slice. Each slice is monitored by a separate copy of the proposed monitoring system. These copies operate in parallel and are synchronized. The scalability of the monitoring system is achieved by enhancing the performance of SDN. In this context, scalability is addressed by increasing the number of tenant applications and expanding the size of the physical network without compromising SDN performance.

Implementasi Load Balancing dengan HAProxy di Sistem Informasi Akademik UIN Sunan Kalijaga

Efficiently managing academic information systems (AIS) is essential for educational institutions to provide reliable services to students and faculty. This research explores the integration of HAProxy load balancing and file synchronization techniques to optimize the performance of AIS. HAProxy is employed to distribute incoming requests across multiple backend servers, and the backend will call web service to access the data saved in the database to facilitate seamless data sharing and access. Additionally, file synchronization mechanisms are implemented to maintain consistency across scripts used in the backend system. The study conducts performance evaluations and benchmarks to assess the impact of HAProxy load balancing and file synchronization on AIS responsiveness and reliability. The results reveal significant system scalability and fault tolerance improvements, reducing downtime and enhancing user experience. This research contributes to optimizing academic information systems, enhancing their ability to handle increased loads, and ensuring the efficient delivery of educational services.

A Testbed Platform to Support an IoT City Lab

This paper describes the development activity that has been carried out for a living laboratory for the city of Cagliari aimed at functioning as a learning center for local SMEs willing to improve their skills in IoT and create applications that will be integrated in an open innovation ecosystem. The many users belonging to the various SMEs involved in the project required an ICT laboratory with a platform that could manage them and provide a multi-tenant environment for the development of IoT applications. The architecture also had to be scalable and interoperable, and the resulting platform had to collect many kinds of data from sensors or other data sources, elaborate them, and show georeferenced information on a 3D satellite interactive view along with statistics on side panels. This work was based on a platform already developed by CRS4 for a previous project. Preserving the concept of the decision-making tool for Smart Cities, almost every component was redesigned, and, in this paper, we describe the new solutions that have been implemented. Starting from the former structure, further features were added in a novel way in order to offer an enhanced framework that can deal with the activities of the laboratory, exploiting the scalability of the open-source systems involved, their robustness and flexibility, and leveraging domain standards. In this article, the main challenges involved in the development of the platform are described, as well as the solutions that have been implemented so far.