Search

The demand for deep learning frameworks capable of running in edge computing environments is rapidly increasing due to the exponential growth of data volume and the need for real-time processing. However, edge computing environments often have limited resources, necessitating the distribution of deep learning models. Distributing deep learning models can be challenging as it requires specifying the resource type for each process and ensuring that the models are lightweight without performance degradation. To address this issue, we propose the Microservice Deep-learning Edge Detection (MDED) framework, designed for easy deployment and distributed processing in edge computing environments. The MDED framework leverages Docker-based containers and Kubernetes orchestration to obtain a pedestrian-detection deep learning model with a speed of up to 19 FPS, satisfying the semi-real-time condition. The framework employs an ensemble of high-level feature-specific networks (HFN) and low-level feature-specific networks (LFN) trained on the MOT17Det dataset, achieving an accuracy improvement of up to AP50 and AP0.18 on MOT20Det data. [ABSTRACT FROM AUTHOR]

The privacy of electricity consumers has become one of the most critical subjects in designing smart meters and their proliferation. In this work, a multilayer architecture has been proposed for anonymous data collection from smart meters, which provides: (1) The anonymity of information for third‐party data consumers; (2) Secure communication to utility provider network for billing purposes; (3) Online control of data sharing for end‐users; (4) Low communication costs based on available Internet of things (IoT) communication protocols. The core elements of this architecture are, first, the digital twin equivalent of the cyber‐physical system and, second, the Tangle distributed ledger network with IOTA cryptocurrency. In this architecture, digital twin models are updated in real‐time by information received from trusted nodes of the Tangle distributed network anonymously. A small‐scale laboratory prototype based on this architecture has been developed using the dSPACE SCALEXIO real‐time simulator and open‐source software tools to prove the feasibility of the proposed solution. The numerical results confirm that after a few seconds of anomaly detection, the microgrid was fully stabilized around its operating point with less than 5% deviation during the transition time. [ABSTRACT FROM AUTHOR]

The medical sector actively changes and implements innovative features in response to technical development and revolutions. Many of the most crucial elements in IoT-connected health services are safeguarding critical patient records from prospective attackers. As a result, BlockChain (BC) is gaining traction in the business sector owing to its large implementations. As a result, BC can efficiently handle everyday life activities as a distributed and decentralized technology. Compared to other industries, the medical sector is one of the most prominent areas where the BC network might be valuable. It generates a wide range of possibilities and probabilities in existing medical institutions. So, throughout this study, we address BC technology's widespread application and influence in modern medical systems, focusing on the critical requirements for such systems, such as trustworthiness, security, and safety. Furthermore, we built the shared ledger for blockchain-based healthcare providers for patient information, contractual between several other parties. The study's findings demonstrate the usefulness of BC technology in IoHT for keeping patient health data. The BDSA-IoHT eliminates 2.01 seconds of service delay and 1.9 seconds of processing time, enhancing efficiency by nearly 30%. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Signal Processing is the property of Journal of Signal Processing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Several unique characteristics of Internet of Things (IoT) devices, such as distributed deployment and limited storage, make it challenging for standard centralized access control systems to enable access control in today’s large-scale IoT ecosystem. To solve these challenges, this study presents an IoT access control system called Ether-IoT based on the Ethereum Blockchain (BC) infrastructure with Attribute-Based Access Control (ABAC). Access Contract (AC), Cache Contract (CC), Device Contract (DC), and Policy Contract (PC) are the four central smart contracts (SCs) that are included in the proposed system. CC offers a way to save user characteristics in a local cache system to avoid delays during transactions between BC and IoT devices. AC is the fundamental program users typically need to run to build an access control technique. DC offers a means for storing the resource data created by devices and a method for querying that data. PC offers administrative settings to handle ABAC policies on users’ behalf. Ether-IoT, combined with ABAC and the BC, enables IoT access control management that is decentralized, fine-grained and dynamically scalable. This research gives a real-world case study to illustrate the suggested framework’s implementation. In the end, a simulation experiment is performed to evaluate the system’s performance. To ensure data integrity in dispersed systems, the results show that Ether-IoT can sustain high throughput in contexts with a large number of requests. [ABSTRACT FROM AUTHOR]

Indonesia is in the top 5 country that generates solid waste. It is the 14th largest country in the world in managing waste, but unfortunately, Indonesia is not known for its diligence to waste management. In 2019, the coordinating minister for Maritime Affairs and Investment of the Republic of Indonesia, Luhut Binsar Pandjaitan, stated that it is a top priority in the national agenda to keep improving solid waste management due to the growing number of people living in the urban areas and the estimated 105.000 tons of solid waste a day. This paper aims to create an IT blueprint for this waste management through a distributed system that allows an optimum flow of waste out of dense urban centers and into the proper waste disposal facilities. It involves smarter waste surveillance, a consolidated fleet of collection agents of varying mobility and capacity and includes people participation to control unnecessary waste generation. The system should be scalable while using existing resources and systems to manage the problem in the long term. It should maintain the three principles of Integrated Solid Waste Management (ISWM): waste prevention, recycling, and disposal. New technologies, such as a sensor network to monitor waste generation close to the source, can help burgeon localized SWM techniques such as community composting. The country can take the steps necessary to mitigate the problem. [ABSTRACT FROM AUTHOR]

In order to meet the requirements of task scheduling in on-board distributed computing environment, an evolutionary optimization scheduling method for on-board tasks was proposed based on the on-board dynamic heterogeneous computing resource model and task model. In this method, task scheduling priority coding was used to adapt to dynamic changes of computing resources. Heuristic critical path for comprehensive index balance was adopted to implement the evaluation, so as to achieve a balance between task makespan, power consumption and reliability. Multi-group neighborhood search strategy was applied to avoid the algorithm falling into local optimization and simplify the complexity of scheduling algorithm. Fault-tolerant strategy of convergence monitoring and stop loss reconfiguration were designed for scheduling unit fault tolerance. Scheduling strategy based on active and backup redundant subtasks was used for computing unit fault tolerance. The simulation results showed the efficiency of this method to deal with the scenario of on-board distributed computing resources and implement the optimal scheduling of on-board tasks. [ABSTRACT FROM AUTHOR]

Countless endeavors have been undertaken to address the Byzantine Generals Problem, a generalization of the Two Generals Problem. The emergence of proof of work (PoW) for Bitcoin has led to various consensus algorithms diverging, and comparable existing consensus algorithms are being gradually utilized interchangeably, or only developed for each specific application domain. Our approach employs an evolutionary phylogeny method to classify blockchain consensus algorithms based on their historical development and current usage. To demonstrate the relatedness and lineage of distinct algorithms, as well as to support the recapitulation theory, which posits that the evolutionary history of its mainnets is mirrored in the development of an individual consensus algorithm, we present a taxonomy. We have created a comprehensive classification of past and present consensus algorithms that serves to organize this swift consensus algorithm evolution period. By recognizing similarities, we have compiled a list of different verified consensus algorithms and performed clustering on over 38 of these. Our new taxonomic tree presents five taxonomic ranks, including the evolutionary process and decision-making method, as a technique for analyzing correlation. Through the examination of the evolution and utilization of these algorithms, we have developed a systematic and hierarchical taxonomy that enables the grouping of consensus algorithms into distinct categories. The proposed method classifies various consensus algorithms according to taxonomic ranks and aims to reveal the direction of research on the application of blockchain consensus algorithms for each domain. [ABSTRACT FROM AUTHOR]

Over the last few decades, distributed generation (DG) has become the most viable option in distribution systems (DSs) to mitigate the power losses caused by the substantial increase in electricity demand and to improve the voltage profile by enhancing power system reliability. In this study, two metaheuristic algorithms, artificial gorilla troops optimization (GTO) and Tasmanian devil optimization (TDO), are presented to examine the utilization of DGs, as well as the optimal placement and sizing in DSs, with a special emphasis on maximizing the voltage stability index and minimizing the total operating cost index and active power loss, along with the minimizing of voltage deviation. The robustness of the algorithms is examined on the IEEE 33-bus and IEEE 69-bus radial distribution networks (RDNs) for PV- and wind-based DGs. The obtained results are compared with the existing literature to validate the effectiveness of the algorithms. The reduction in active power loss is 93.15% and 96.87% of the initial value for the 33-bus and 69-bus RDNs, respectively, while the other parameters, i.e., operating cost index, voltage deviation, and voltage stability index, are also improved. This validates the efficiency of the algorithms. The proposed study is also carried out by considering different voltage-dependent load models, including industrial, residential, and commercial types. [ABSTRACT FROM AUTHOR]

In this paper, a massively parallel implementation of Boltzmann's thermally activated molecular transport model is presented. This models allows taking into account potential energy barriers in molecular simulations and thus modeling thermally activated diffusion processes in liquids. The model is implemented as an extension to the basic Dynamic Lattice Liquid (DLL) algorithm on ARUZ, a massively parallel FPGA-based simulator located at BioNanoPark Lodz. The advantage of this approach is that it does not use any exponentiation operations, minimizing resource usage and allowing one to perform simulations containing up to 4,608,000 nodes. [ABSTRACT FROM AUTHOR]

Copyright of Control Theory & Applications / Kongzhi Lilun Yu Yinyong is the property of Editorial Department of Control Theory & Applications and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Electric vehicles (EVs) are environment friendly and are expected to find tremendous growth in the near future. However, the short voyaging reach and the lack of charging lattice have impeded the large-scale application of EVs. A choice of a particular route for EV is largely dependent on the availability of charging stations on that route as well as the ease of charging facility offered by the charging station (including state of the charging station, user convenience, etc.). Hence, it is vital to delve into the challenge of locating a charging station that offers the utmost user convenience in charging EVs. This paper attempts to present a solution to this very problem. The proposed approach is an improvised distributed system, namely, S2RC (Smart Search of Route and Charging) which plans an energy efficient EV route considering EV’s state-of-charge (SoC) level, traffic conditions, the frequency of charging stations and the state of charging station (the charging resources of the charging station along with their occupancy levels, etc.). This distributed architecture employs the proposed agile charging slot reservation approach for the EVs that wish to get recharged at a particular charging station. Besides, to delineate the interactive mechanism of EVs recharging, three non-identical objective functions are formulated to minimize the overall energy consumption of EVs, waiting time at charging station, and total charging expenditure. After performing the extensive simulations on the weighted directed real transportation graph of Chandigarh, India, the proposed S2RC system recommends a charging station in accordance to the preferences given by the EV user. Case studies verify the significance and validity of the proposed model in terms of energy efficiency and user convenience.

Discussing the common scenarios in modern data warehouse systems that need to receive a large amount of streaming data, connect it with the existing data on the disk, and then store it in the warehouse.By rationally setting disk paging and applying cache modules to disperse the disk I/O pressure, a more efficient data receiving scheme was proposed based on the existing research, and a consistent Hash function was introduced and extended to distributed environment and a D-CACHEJOIN algorithm applied to distributed environment was proposed.The cost model of the algorithm was calculated by theory and simulation experiment was performed using data that obey the Zipfian distribution.The experiment results show that the proposed algorithm has higher efficiency than existing algorithms in practical application scenarios close to reality, and can be quickly and easily extended to distributed environments.

Since the advent of Bitcoin in 2008, blockchain has gradually become a research hotspot in academia.As the key technology of blockchain, consensus algorithm has also attracted more attention from researchers.It's easy to introduce Byzantine fault nodes in blockchain system because of its complex and variable runtime, so the blockchain Byzantine fault tolerant consensus algorithm is a difficulty that must be overcome.This paper systematically summarizes the research progress of the blockchain Byzantine fault tolerant consensus algorithm, in order to provide a reference for the innovation of consensus algorithms in the future.Firstly, sorting out the four major factions of the existing blockchain Byzantine fault tolerant consensus algorithms and introducing the BFT consensus algorithm.Secondly, reviewing several important values in the classic PBFT algorithm and its correctness proof.Thirdly, putting forward the four optimization goals of the BFT consensus algorithm:decentralization, efficiency, fault tolerance rate and security.Then, based on the dimensions of consensus rounds, number of consensus nodes, underlying hardware, communication mode or encryption algorithm, probability of fault nodes, five optimization ideas of BFT consensus algorithm are summarized.Finally, analysising 10 classic BFT consensus algorithms in detail and making performance comparison.

Abstract Unsupervised federated domain adaptation uses the knowledge from several distributed unlabelled source domains to complete the learning on the unlabelled target domain. Some of the existing methods have limited effectiveness and involve frequent communication. This paper proposes a framework to solve the distributed multi-source domain adaptation problem, referred as self-supervised federated domain adaptation (SFDA). Specifically, a multi-domain model generalization balance is proposed to aggregate the models from multiple source domains in each round of communication. A weighted strategy based on centroid similarity is also designed for SFDA. SFDA conducts self-supervised training on the target domain to tackle domain shift. Compared with the classical federated adversarial domain adaptation algorithm, SFDA is not only strong in communication cost and privacy protection but also improves in the accuracy of the model.

Polling system is not trusted everywhere around the world it is very important in this modern world to replace the traditional polling system with the new technology. Some countries like United States, Japan, and India suffer from corrupted polling system. Major issues are faced by current polling systems like system hacking, vote rigging, vote manipulation, distributed denial of service attack, and online polling booth capturing. This paper will lead to the problems faced by the traditional polling system and how the new technology will provide the solution to that problem. Also, our purpose is to check the feasibility of the system by recording the transaction fees and evaluate the right way to spend the amount of gas in the transaction. This will highlight blockchain frameworks including blockchain as a service and polling system which is on blockchain that addresses all constraint introducing ethereum which is a blockchain-based distributed computing platform. Ethereum is open source, and publicly available with a system featuring smart contracts. It provides the cryptocurrency wallets that let you make cheap, instant payments with gas in the form of ethers. The ethereum community is the most active and largest blockchain community in the world. There is no centralized organization that controls ethereum.

Tasks allocation problem in heterogeneous and distributed-multiprocessing computing environments is a nonlinear multi-objective NP-hard problem and from a research perspective, it is considereda major issue. The underlying objective of any allocation mechanism to execute the specified tasks set and enumerating processors is to minimize the overall cost. To solve the tasks allocation problems numerous meta-heuristic techniques have been tried and tested but still, plenty of scopes are there for optimal strategies. In the presented article, a comprehensive task assignment model based on particle swarm optimization (PSO) is developed, which optimizes response time, flowtime, and cost of the distributed computing system. In the present technique, the ‘n’ number of cluster centroids of ‘r’ tasks, is updated by the PSO technique to form the ‘n’ task clusters to minimize the communication costs then their allocation is made by the newly proposed heuristic method. Considering the fact, in the given model objective functions conflict with each other and the good thing about the PSO algorithm are its accuracy and speed. And to come up with the most favorable solution, authors have developed PSO based algorithm.The PSO-based technique that is proposed in this article to solve a given assignment problem in a distributed computing system is able to give better results in a manner of convergence rate as thePSO method integrates local and global search methods in an attempt to strike a balance between research and utilization. To examine the functioning of the developed technique, the well-demonstrated scheduling policies based on different techniques have been compared and obtained finer outcomes. The developed mechanism is acceptable for an erratic number of tasks and processors.

Called on to refine service-oriented architecture, a new architectural pattern named microservices emerged in the early two-thousands promising to speed up the delivery of new features. All of this is achieved by allowing many teams to work on separate autonomous parts of an application independently, though steepening the learning curve and introducing infrastructure-related issues. Overtime, while gaining more popularity, the initial meaning, as well as peculiarities of distributed systems development and reason for the use of the pattern, has turned out to be less prominent, instead becoming a sliver bullet for monolithic applications, which irreversibly leads to a tremendous increase in complexity of a system as well as other inherent problems. The absence of a single definition as well as the misleading name of the microservices pattern has all contributed to the development of the notoriously known distributed monolith. In this document, a review was conducted to resurrect the notion of microservices, understand their nature, and how and when the pattern should be applied. This is done by firstly understanding the difference between monolithic and distributed systems including their strengths and weaknesses, and defining the single most important reason for microservices to be used. The main part focuses on reducing the coupling in the system, which is the main obstacle during the development of a reliable and agile distributed system. Firstly, by modeling boundaries following domain teams and architectural needs, we ensure autonomous deployable units are created that provide independence during development, isolation, and reliability. Later by analyzing approaches to sharing data and communication, which are two major forces structuring a distributed system, we secure the previously established boundaries.

Many graph-related applications face the challenge of managing excessive and ever-growing graph data in a distributed environment. Therefore, it is necessary to consider a graph partitioning algorithm to distribute graph data onto multiple machines as the data comes in. Balancing data distribution and minimizing edge-cut ratio are two basic pursuits of the graph partitioning problem. While achieving balanced partitions for streaming graphs is easy, existing graph partitioning algorithms either fail to work on streaming workloads, or leave edge-cut ratio to be further improved. Our research aims to provide a better solution that fits the need of streaming graph partitioning in a distributed system, which further reduces the edge-cut ratio while maintaining rough balance among all partitions. We exploit the similarity measure on the degree of vertices to gather structuralrelated vertices in the same partition as much as possible, this reduces the edge-cut ratio even further as compared to the state-of-the-art streaming graph partitioning algorithm-FENNEL. Our evaluation shows that our streaming graph partitioning algorithm is able to achieve better partitioning quality in terms of edge-cut ratio (up to 20% reduction as compared to FENNEL) while maintaining decent balance between all partitions, and such improvement applies to various real-life graphs.

Recently, deep neural networks (DNNs) have shown great promise in many fields while their parameter sizes are rapidly expanding. To break through the computation and memory limitation of a single machine, pipeline model parallelism is proposed for large-scale DNN training by fully utilizing the computation and storage power of the distributed cluster. Cloud data centers can also provide sufficient computing, storage and bandwidth resources. However, most existing approaches apply layer-wise partitioning, which is difficult to obtain an even model partition result because of the large computational overhead discrepancy between DNN layers, resulting in degraded efficiency. To tackle this issue, we propose "Bi-Partition", a novel partitioning method based on bidirectional partitioning for forward propagation (FP) and backward propagation (BP), which improves the efficiency of the pipeline model parallelism system. By deliberated designing distinct cut positions for FP and BP of DNN training, workers in the pipeline get nearly equal computational loads, and the balanced pipeline fully utilizes the computing resources. Experiments on various DNN models and datasets validate the efficiency of our mechanism, e.g., the training efficiency achieving up to 1.9 × faster than the state-of-the-art method PipeDream. [ABSTRACT FROM AUTHOR]

In this paper, the concept of a distributed wireless system of weigh-in-motion is proposed. A wireless sensor for weigh-in-motion (WIM) based on piezoelectric materials is designed. The corresponding prototype for indoor testing is made according to the design. It aims to realize the main functions of dynamic pressure sensing, charge signal amplification and conversion, wireless signal transmission and reception, etc. The material properties of the mechanical properties and the piezoelectric properties in the analyses are provided in detail. Through the indoor test platform, the feasibility of the wireless sensor for WIM designed in this paper is preliminarily verified, which provides a basic tool for the realization of the distributed self-powered wireless system of WIM in the next step. [ABSTRACT FROM AUTHOR]

Over the last few decades, graphs have become increasingly important in many applications and domains for managing Big data. Big data analysis in a graph database is described as an analysis of exponentially increasing massive interconnected data concerning time. However, analyzing big connected data in social networks and synthetic identity detection is challenging. In previous approaches, fraud detection has been done on the complete graph data, which is a time-consuming process and will create bottlenecks while query execution. To overcome the issue, this paper proposes a new fraud detection technique to unveil synthetic identities involved in the Panama Paper leak dataset (unprecedented leak of 11.5 m data from the database of the world's fourth-biggest offshore law arm, Mossack Fonseca) using a Node rank-based fraud detection algorithm by integrating distributed data profiling techniques on a minimized graph by minimizing the least influential nodes. The proposed model is verified on the three nodes cluster to improve data scalability, reduce the query execution time by an average of 30–36% and finally reduce the fraud detection time by 18.2%. [ABSTRACT FROM AUTHOR]

On-demand charging of sensor nodes (SNs) in wireless rechargeable sensor networks has garnered immense attention. Existing works with multiple mobile chargers (MCs) overlooked the benefits of partial charging and distributed control in large-scale wireless rechargeable sensor networks. In addition, most of them have not considered the complete heterogeneity of SNs’ energy profiles and idleness of the MCs. In this article, we address the aforesaid issues and present a game theory based distributed mobile charging protocol, called DMCP. We formulate the mobile charging problem as a series of repeated games played among the MCs and represent a game as a 0 − 1 integer linear programming to maximize the total profit of the MCs. In each game, DMCP dynamically determines the factor of charging of the energy-deficit SNs based on their relative criticality and selects the next-to-be-recharged SNs using an efficient bi-objective payoff function. In addition, it utilizes an innovative strategy and a self-enforcing agreement to promote cooperation among the MCs. Through extensive simulations and hypothesis testing, we show that DMCP reduces the charging delay up to 44.04% and enhances the charging coverage and survival rate up to 45.12% and 49.97%, respectively. [ABSTRACT FROM AUTHOR]

The pre-trained model (PTM) is revolutionizing Artificial Intelligence (AI) technology. However, the hardware requirement of PTM training is prohibitively high, making it a game for a small proportion of people. Therefore, we proposed PatrickStar system to lower the hardware requirements of PTMs and make them accessible to everyone. PatrickStar uses the CPU-GPU heterogeneous memory space to store the model data. Different from existing works, we organize the model data in memory chunks and dynamically distribute them in the heterogeneous memory. Guided by the runtime memory statistics collected in a warm-up iteration, chunks are orchestrated efficiently in heterogeneous memory and generate lower CPU-GPU data transmission volume and higher bandwidth utilization. Symbiosis with the Zero Redundancy Optimizer, PatrickStar scales to multiple GPUs on multiple nodes. The system can train tasks on bigger models and larger batch sizes, which cannot be accomplished by existing works. Experimental results show that PatrickStar extends model scales 2.27 and 2.5 times of DeepSpeed, and exhibits significantly higher execution speed. PatricStar also successfully runs the 175B GPT3 training task on a 32 GPU cluster. Our code is available at https://github.com/Tencent/PatrickStar. [ABSTRACT FROM AUTHOR]

Eventual consistency is a consistency model that favors liveness over safety. It is often used in large-scale distributed systems where models ensuring a stronger safety incur performance that are too low to be deemed practical. Eventual consistency tends to be uniformly applied within a system, but we argue a demand exists for differentiated eventual consistency, e.g. in blockchain systems. We propose update-query consistency with primaries and secondaries (UPS) to address this demand. UPS is a novel consistency mechanism that works in pair with our novel two-phase epidemic broadcast protocol gossip primary-secondary (GPS) to offer differentiated eventual consistency and delivery speed. We propose two complementary analyses of the broadcast protocol: a continuous analysis and a discrete analysis based on compartmental models used in epidemiology. Additionally, we propose the formal definition of a scalable consistency metric to measure the consistency trade-off at runtime. We evaluate UPS in two simulated worldwide settings: a one-million-node network and a network emulating that of the Ethereum blockchain. In both settings, UPS reduces inconsistencies experienced by a majority of the nodes and reduces the average message latency for the remaining nodes. [ABSTRACT FROM AUTHOR]

The smart city system, which is a type of enterprise management system (EMS), automatically manages cities and schedules resources efficiently based on spatial data generated by devices, such as the Internet of Things and mobile. However, with the increasing deployment of technologies, including sensor and location-based services, their ever-growing spatial data are no longer managed efficiently by traditional EMS. To overcome this issue, we present SeFrame, which is a spatially enabled framework for improving the efficiency of smart city EMS based on a distributed architecture. The framework supports a set of spatial queries, including: The range query, k-nearest neighbors query, and spatial join query. It benefits greatly from using the buffer-enabled partition method to eliminate duplicate results. In each partition, the local index based on combination of the quad-tree and grid index (CQG) significantly improves the spatial query efficiency in memory. CQG manages complex spatial objects, including a point, polygon, and polyline. By taking full advantage of the local index, SeFrame accesses skewed spatial data in constant time. In experiments, we demonstrated that the proposed method delivered superior performance in terms of scalability and query efficiency, in most cases. [ABSTRACT FROM AUTHOR]

Chemical agents are one of the major threats to soldiers in modern warfare, so it is so important to detect chemical agents rapidly and accurately on battlefields. Raman spectroscopy-based detectors are widely used but have many limitations. The Raman spectrum changes unpredictably due to various environmental factors, and it is hard for detectors to make appropriate judgments about new chemical substances without prior information. Thus, the existing detectors with inflexible techniques based on determined rules cannot deal with such problems flexibly and reactively. Artificial intelligence (AI)-based detection techniques can be good alternatives to the existing techniques for chemical agent detection. To build AI-based detection systems, sufficient amounts of data for training are required, but it is not easy to produce and handle fatal chemical agents, which causes difficulty in securing data in advance. To overcome the limitations, in this paper, we propose the distributed Raman spectrum data augmentation system that leverages federated learning (FL) with deep generative models, such as generative adversarial network (GAN) and autoencoder. Furthermore, the proposed system utilizes various additional techniques in combination to generate a large number of Raman spectrum data with reality along with diversity. We implemented the proposed system and conducted diverse experiments to evaluate the system. The evaluation results validated that the proposed system can train the models more quickly through cooperation among decentralized troops without exchanging raw data and generate realistic Raman spectrum data well. Moreover, we confirmed that the classification model on the proposed system performed learning much faster and outperformed the existing systems. [ABSTRACT FROM AUTHOR]