Pub Date : 2024-03-23DOI: 10.1177/00375497241236968
Tao Dai, Mingyu Yu, Yong Wu
In queueing systems where queues are invisible, it is critical for companies to make decisions about the timing of announcing the anticipated delay to customers. In this paper, a simulation model is built to simulate an invisible queueing system, and several queueing scenarios are considered, including different system congestion and different company goals, to explore the impact of different announcement timings. In the modeling of customer behavior, we argue that it is difficult for companies to announce perfectly accurate delay and rarely have customers fully trust the announcement, so the quantal-response model is included to model customers’ probabilistic choice behavior due to the bounded rationality. Meanwhile, we consider customer heterogeneity, assign customers different initial patience and, as an extension, also assume that patience will be updated. We perform simulation experiments and analyze the experimental data to dissect the underlying reasons, and then give sound management suggestions. The experiments show that the optimal announcement timing is different for different scenarios, which shows that in practical decisions, companies should adopt different announcement strategies for different scenarios. What’s more, in some scenarios, delayed announcement at specific time is better than on-arrival announcement, which suggests that when we judge the value of announcement, we should add the definite word of specific timing to the announcement. The breakthrough point of this paper is to consider customers’ bounded rationality and the dynamic patience; meanwhile, it fills the gap of announcement timing research and explores the value of additional announcements after the initial announcement.
{"title":"When to announce the queueing information for bounded rationality customers: a discrete-event–based simulation model","authors":"Tao Dai, Mingyu Yu, Yong Wu","doi":"10.1177/00375497241236968","DOIUrl":"https://doi.org/10.1177/00375497241236968","url":null,"abstract":"In queueing systems where queues are invisible, it is critical for companies to make decisions about the timing of announcing the anticipated delay to customers. In this paper, a simulation model is built to simulate an invisible queueing system, and several queueing scenarios are considered, including different system congestion and different company goals, to explore the impact of different announcement timings. In the modeling of customer behavior, we argue that it is difficult for companies to announce perfectly accurate delay and rarely have customers fully trust the announcement, so the quantal-response model is included to model customers’ probabilistic choice behavior due to the bounded rationality. Meanwhile, we consider customer heterogeneity, assign customers different initial patience and, as an extension, also assume that patience will be updated. We perform simulation experiments and analyze the experimental data to dissect the underlying reasons, and then give sound management suggestions. The experiments show that the optimal announcement timing is different for different scenarios, which shows that in practical decisions, companies should adopt different announcement strategies for different scenarios. What’s more, in some scenarios, delayed announcement at specific time is better than on-arrival announcement, which suggests that when we judge the value of announcement, we should add the definite word of specific timing to the announcement. The breakthrough point of this paper is to consider customers’ bounded rationality and the dynamic patience; meanwhile, it fills the gap of announcement timing research and explores the value of additional announcements after the initial announcement.","PeriodicalId":501452,"journal":{"name":"SIMULATION","volume":"204 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140202845","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-03-23DOI: 10.1177/00375497241232432
Ishan Honhaga, Claudia Szabo
Cooperative multiagent reinforcement learning approaches are increasingly being used to make decisions in contested and dynamic environments, which tend to be wildly different from the environments used to train them. As such, there is a need for a more in-depth understanding of their resilience and robustness in conditions such as network partitions, node failures, or attacks. In this article, we propose a modeling and simulation framework that explores the resilience of four c-MARL models when faced with different types of attacks, and the impact that training with different perturbations has on the effectiveness of these attacks. We show that c-MARL approaches are highly vulnerable to perturbations of observation, action reward, and communication, showing more than 80% drop in the performance from the baseline. We also show that appropriate training with perturbations can dramatically improve performance in some cases, however, can also result in overfitting, making the models less resilient against other attacks. This is a first step toward a more in-depth understanding of the resilience c-MARL models and the effect that contested environments can have on their behavior and toward resilience of complex systems in general.
{"title":"A simulation and experimentation architecture for resilient cooperative multiagent reinforcement learning models operating in contested and dynamic environments","authors":"Ishan Honhaga, Claudia Szabo","doi":"10.1177/00375497241232432","DOIUrl":"https://doi.org/10.1177/00375497241232432","url":null,"abstract":"Cooperative multiagent reinforcement learning approaches are increasingly being used to make decisions in contested and dynamic environments, which tend to be wildly different from the environments used to train them. As such, there is a need for a more in-depth understanding of their resilience and robustness in conditions such as network partitions, node failures, or attacks. In this article, we propose a modeling and simulation framework that explores the resilience of four c-MARL models when faced with different types of attacks, and the impact that training with different perturbations has on the effectiveness of these attacks. We show that c-MARL approaches are highly vulnerable to perturbations of observation, action reward, and communication, showing more than 80% drop in the performance from the baseline. We also show that appropriate training with perturbations can dramatically improve performance in some cases, however, can also result in overfitting, making the models less resilient against other attacks. This is a first step toward a more in-depth understanding of the resilience c-MARL models and the effect that contested environments can have on their behavior and toward resilience of complex systems in general.","PeriodicalId":501452,"journal":{"name":"SIMULATION","volume":"160 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140202741","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-03-22DOI: 10.1177/00375497241237831
Hung Tuan Trinh, Sang-Hoon Bae, Duy Quang Tran
In the future, mixed traffic flow will include two types of vehicles: connected autonomous vehicles (CAVs) and human-driven vehicles (HDVs). CAVs emerge as new solutions to disrupt the traditional transportation system. This new solution shares real-time data with each other and the roadside units (RSU) for network management. Reinforcement learning (RL) is a promising approach for traffic signal management in complex urban areas by leveraging information gathered from CAVs. In particular, coordinating signal management at many intersections is a critical challenge in multi-agent reinforcement learning (MARL). According to this vision, we propose an approach that combines an actor–critic network–based multi-agent deep deterministic policy gradient (MADDPG) model and a rerouting technique (RT) to increase traffic performance in vehicular networks. This algorithm overcomes the inherent non-stationary of Q-learning and the high variance of policy gradient (PG) algorithms. Based on centralized learning with decentralized execution, the MADDPG model employs one actor and one critic for each agent. The actor network uses local information to execute actions, while the critic network is trained with extra information, including the states and actions of other agents. Through a centralized learning process, agents can coordinate with each other, diminishing the influence of an unstable environment. Unlike previous studies, we not only manage traffic light systems but also consider the effect of platooning vehicles on increasing throughput. Experimental results show that our model outperforms other models in terms of traffic performance in different scenarios.
{"title":"Combining multi-agent deep deterministic policy gradient and rerouting technique to improve traffic network performance under mixed traffic conditions","authors":"Hung Tuan Trinh, Sang-Hoon Bae, Duy Quang Tran","doi":"10.1177/00375497241237831","DOIUrl":"https://doi.org/10.1177/00375497241237831","url":null,"abstract":"In the future, mixed traffic flow will include two types of vehicles: connected autonomous vehicles (CAVs) and human-driven vehicles (HDVs). CAVs emerge as new solutions to disrupt the traditional transportation system. This new solution shares real-time data with each other and the roadside units (RSU) for network management. Reinforcement learning (RL) is a promising approach for traffic signal management in complex urban areas by leveraging information gathered from CAVs. In particular, coordinating signal management at many intersections is a critical challenge in multi-agent reinforcement learning (MARL). According to this vision, we propose an approach that combines an actor–critic network–based multi-agent deep deterministic policy gradient (MADDPG) model and a rerouting technique (RT) to increase traffic performance in vehicular networks. This algorithm overcomes the inherent non-stationary of Q-learning and the high variance of policy gradient (PG) algorithms. Based on centralized learning with decentralized execution, the MADDPG model employs one actor and one critic for each agent. The actor network uses local information to execute actions, while the critic network is trained with extra information, including the states and actions of other agents. Through a centralized learning process, agents can coordinate with each other, diminishing the influence of an unstable environment. Unlike previous studies, we not only manage traffic light systems but also consider the effect of platooning vehicles on increasing throughput. Experimental results show that our model outperforms other models in terms of traffic performance in different scenarios.","PeriodicalId":501452,"journal":{"name":"SIMULATION","volume":"364 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140202846","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-03-21DOI: 10.1177/00375497241235202
Ying Zheng, Langchao Ji, Zhengxiang Xu, Yanyan Chen, Xingang Li
In recent years, terrorist attacks all over the world caused many civilian casualties seriously. Under the knife and axe terrorist attack, there are usually four groups of event-related persons, that is, pedestrians, terrorists, police, and safety guards in the scene. The behavior of them is different, but they are rarely considered together in existing litterateurs, especially the police and safety guards. In this paper, we propose the FFPS (extended floor-field model combined with on-site control actions of police and safety guards) model to study pedestrian evacuation dynamics considering terrorist attack on-site control action differences between police and security guards. The FFPS model consists of three aspects. First, pedestrian evacuation dynamics in terrorist attack environment is modeled. Second, the behavior of terrorists, police and safety guards is studied. Third, the differences of on-site control actions between police and safety guards are studied. We set up the scenario for simulation. The influence of terrorists’ number and location, police and security guards’ different actions, police entering time, police shooting distance, police and safety guards’ spatial position, and police control strategy on evacuation dynamics are analyzed in detail. Those results provide valuable insights to quickly control terrorists, and sharply reduce casualties in terrorist attack environment.
{"title":"Pedestrian evacuation dynamics considering terrorist attack on-site control action differences between police and security guards","authors":"Ying Zheng, Langchao Ji, Zhengxiang Xu, Yanyan Chen, Xingang Li","doi":"10.1177/00375497241235202","DOIUrl":"https://doi.org/10.1177/00375497241235202","url":null,"abstract":"In recent years, terrorist attacks all over the world caused many civilian casualties seriously. Under the knife and axe terrorist attack, there are usually four groups of event-related persons, that is, pedestrians, terrorists, police, and safety guards in the scene. The behavior of them is different, but they are rarely considered together in existing litterateurs, especially the police and safety guards. In this paper, we propose the FFPS (extended floor-field model combined with on-site control actions of police and safety guards) model to study pedestrian evacuation dynamics considering terrorist attack on-site control action differences between police and security guards. The FFPS model consists of three aspects. First, pedestrian evacuation dynamics in terrorist attack environment is modeled. Second, the behavior of terrorists, police and safety guards is studied. Third, the differences of on-site control actions between police and safety guards are studied. We set up the scenario for simulation. The influence of terrorists’ number and location, police and security guards’ different actions, police entering time, police shooting distance, police and safety guards’ spatial position, and police control strategy on evacuation dynamics are analyzed in detail. Those results provide valuable insights to quickly control terrorists, and sharply reduce casualties in terrorist attack environment.","PeriodicalId":501452,"journal":{"name":"SIMULATION","volume":"7 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140202797","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
As digitalization is permeating all sectors of society toward the concept of “smart everything,” and virtual technologies and data are gaining a dominant place in the engineering and control of intelligent systems, the Digital Twin (DT) concept has surfaced as one of the top technologies to adopt. This paper discusses the DT concept from the viewpoint of Modeling and Simulation (M&S) experts. It both provides literature review elements and adopts a commentary-driven approach. We first examine the DT from a historical perspective, tracing the historical development of M&S from its roots in computational experiments to its applications in various fields and the birth of DT-related and allied concepts. We then approach DTs as an evolution of M&S, acknowledging the overlap in these different concepts. We also look at the M&S workflow and its evolution toward a DT workflow from a software engineering perspective, highlighting significant changes. Finally, we look at new challenges and requirements DTs entail, potentially leading to a revolutionary shift in M&S practices. In this way, we hope to foster the discussion on DTs and provide the M&S expert with innovative perspectives.
{"title":"From modeling and simulation to Digital Twin: evolution or revolution?","authors":"Zeeshan Ali, Raheleh Biglari, Joachim Denil, Joost Mertens, Milad Poursoltan, Mamadou Kaba Traoré","doi":"10.1177/00375497241234680","DOIUrl":"https://doi.org/10.1177/00375497241234680","url":null,"abstract":"As digitalization is permeating all sectors of society toward the concept of “smart everything,” and virtual technologies and data are gaining a dominant place in the engineering and control of intelligent systems, the Digital Twin (DT) concept has surfaced as one of the top technologies to adopt. This paper discusses the DT concept from the viewpoint of Modeling and Simulation (M&S) experts. It both provides literature review elements and adopts a commentary-driven approach. We first examine the DT from a historical perspective, tracing the historical development of M&S from its roots in computational experiments to its applications in various fields and the birth of DT-related and allied concepts. We then approach DTs as an evolution of M&S, acknowledging the overlap in these different concepts. We also look at the M&S workflow and its evolution toward a DT workflow from a software engineering perspective, highlighting significant changes. Finally, we look at new challenges and requirements DTs entail, potentially leading to a revolutionary shift in M&S practices. In this way, we hope to foster the discussion on DTs and provide the M&S expert with innovative perspectives.","PeriodicalId":501452,"journal":{"name":"SIMULATION","volume":"14 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140196318","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-03-19DOI: 10.1177/00375497241233783
Mostafa D Fard, Hessam S Sarjoughian
Understanding the dynamics of complex systems requires developing and combining different kinds of models that can be simulated separately and together. Modeling the interactions as separate models contributes to building flexible hybrid simulation frameworks. In this research, a Discrete Event System Specification–based Interaction Model (DEVS-IM) framework is developed based on the Knowledge Interchange Broker (KIB) approach. This KIB-based RESTful modeling composition framework is shown to enable systematic modeling and simulation of interactions between disparate simulatable models. It supports storing IMs developed for componentized Water Evaluation and Planning System (WEAP) and Low Emissions Analysis Platform (LEAP) tools. It generates the skeleton of DEVS-IMs stored in database for DEVS-Suite simulator. An exemplar model consisting of water, energy, and IMs demonstrates this methodology for developing nexus models of water–energy systems.
{"title":"A knowledge interchange broker composition modeling framework for simulating water, energy, and water-energy nexus systems","authors":"Mostafa D Fard, Hessam S Sarjoughian","doi":"10.1177/00375497241233783","DOIUrl":"https://doi.org/10.1177/00375497241233783","url":null,"abstract":"Understanding the dynamics of complex systems requires developing and combining different kinds of models that can be simulated separately and together. Modeling the interactions as separate models contributes to building flexible hybrid simulation frameworks. In this research, a Discrete Event System Specification–based Interaction Model (DEVS-IM) framework is developed based on the Knowledge Interchange Broker (KIB) approach. This KIB-based RESTful modeling composition framework is shown to enable systematic modeling and simulation of interactions between disparate simulatable models. It supports storing IMs developed for componentized Water Evaluation and Planning System (WEAP) and Low Emissions Analysis Platform (LEAP) tools. It generates the skeleton of DEVS-IMs stored in database for DEVS-Suite simulator. An exemplar model consisting of water, energy, and IMs demonstrates this methodology for developing nexus models of water–energy systems.","PeriodicalId":501452,"journal":{"name":"SIMULATION","volume":"27 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140170855","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-03-16DOI: 10.1177/00375497241235199
Peter M Maurer
Differential search trees can be used for selection with replacement and for a form of selection without replacement. We show that they can be extended to many different types of selection, both with and without replacement. In addition, virtually every aspect of a differential search tree can be modified dynamically. We provide algorithms for making these modifications. Virtually all differential search tree algorithms are straightforward and easy to implement, especially with our preferred implementation, which is both simple and efficient. Differential search tree operations are virtually all logarithmic with the exception of building the tree and dynamically adding leaves to the tree, which are both linear.
{"title":"Discrete random variates with finite support using differential search trees","authors":"Peter M Maurer","doi":"10.1177/00375497241235199","DOIUrl":"https://doi.org/10.1177/00375497241235199","url":null,"abstract":"Differential search trees can be used for selection with replacement and for a form of selection without replacement. We show that they can be extended to many different types of selection, both with and without replacement. In addition, virtually every aspect of a differential search tree can be modified dynamically. We provide algorithms for making these modifications. Virtually all differential search tree algorithms are straightforward and easy to implement, especially with our preferred implementation, which is both simple and efficient. Differential search tree operations are virtually all logarithmic with the exception of building the tree and dynamically adding leaves to the tree, which are both linear.","PeriodicalId":501452,"journal":{"name":"SIMULATION","volume":"22 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140154857","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-03-15DOI: 10.1177/00375497241233284
Andrew Gibson, Manuel D Rossetti
This paper presents a massively parallel, cloud-computing framework for the ad hoc evaluation of discrete-event simulation (DES) models to enable broad exploration of the design space for model parameters. Parallel evaluation is enabled through use of a serverless computing environment allowing thousands of simultaneous experiments, on demand, without the need to explicitly provision or manages hardware. A standard Simulation Evaluation application programming interface (API) was designed for evaluating simulation functions that enables language independence between client application and simulation model, encouraging reuse of simulation models for multiple purposes (what-if analysis, ranking and selection, sensitivity analysis, or optimization). Extensions to the Java Simulation Library (JSL)27 enable rapid deployment of models built with the JSL as parameterized serverless functions implementing the Simulation Evaluation API. New Java packages facilitate the calling of any serverless functions that implement the Simulation Evaluation API.
本文介绍了一种大规模并行云计算框架，用于对离散事件仿真（DES）模型进行临时评估，以广泛探索模型参数的设计空间。并行评估是通过使用无服务器计算环境实现的，该环境允许按需同时进行数千次实验，而无需明确提供或管理硬件。为评估仿真功能设计了一个标准的仿真评估应用编程接口（API），使客户端应用程序和仿真模型之间不依赖语言，鼓励为多种目的（假设分析、排序和选择、灵敏度分析或优化）重复使用仿真模型。通过对 Java 仿真库（JSL）27 的扩展，可以将使用 JSL 构建的模型快速部署为实现仿真评估 API 的参数化无服务器函数。新的 Java 包有助于调用任何实现仿真评估 API 的无服务器函数。
{"title":"Enabling massively parallel, ad hoc exploration of the design space for simulation models within a serverless environment","authors":"Andrew Gibson, Manuel D Rossetti","doi":"10.1177/00375497241233284","DOIUrl":"https://doi.org/10.1177/00375497241233284","url":null,"abstract":"This paper presents a massively parallel, cloud-computing framework for the ad hoc evaluation of discrete-event simulation (DES) models to enable broad exploration of the design space for model parameters. Parallel evaluation is enabled through use of a serverless computing environment allowing thousands of simultaneous experiments, on demand, without the need to explicitly provision or manages hardware. A standard Simulation Evaluation application programming interface (API) was designed for evaluating simulation functions that enables language independence between client application and simulation model, encouraging reuse of simulation models for multiple purposes (what-if analysis, ranking and selection, sensitivity analysis, or optimization). Extensions to the Java Simulation Library (JSL)<jats:sup>27</jats:sup> enable rapid deployment of models built with the JSL as parameterized serverless functions implementing the Simulation Evaluation API. New Java packages facilitate the calling of any serverless functions that implement the Simulation Evaluation API.","PeriodicalId":501452,"journal":{"name":"SIMULATION","volume":"2 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140154930","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：