Performance Evaluation最新文献_第4页

Discrete Bayesian Optimization via Machine Learning 基于机器学习的离散贝叶斯优化

IF 1 4区计算机科学 Q4 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Performance Evaluation

Pub Date : 2025-05-21 DOI: 10.1016/j.peva.2025.102487

Roberto Sala, Bruno Guindani, Danilo Ardagna, Alessandra Guglielmi

Bayesian Optimization (BO) is a family of powerful algorithms designed to solve complex optimization problems involving expensive black-box functions. These sequential algorithms iteratively update a surrogate model of the objective function (OF), effectively balancing exploration and exploitation to identify near-optimal solutions within a limited number of iterations. Originally designed for continuous, unconstrained domains, its efficiency has inspired adaptations for discrete, constrained optimization problems. On the other hand, Machine Learning (ML) models allow accurate predictions for black-box functions, although they typically require large amounts of data for training. Leveraging the strengths of BO and ML, research tackles the challenge of identifying optimal configurations in the context of cloud computing. This paradigm has become pervasive due to its ability to provide flexible and scalable resources. Identifying the optimal hardware-software configuration is essential for minimizing costs while meeting Quality of Service constraints. This task involves solving complex optimization problems over multidimensional discrete domains and black-box objective functions and constraints, within a limited number of iterations. To address this challenge, this work introduces d-MALIBOO, a BO-based algorithm that integrates ML techniques to enhance the efficiency of finding near-optimal solutions in discrete and bounded domains. While BO builds the surrogate model of the OF, ML models determine the feasible region of the black-box constraints and guide the BO algorithm toward promising regions of the discrete domain. Furthermore, we introduce an

ɛ

-greedy approach to favor exploration in domains with multiple local optima. Experimental results show that our algorithm outperforms OpenTuner, a popular framework for constrained optimization, by reducing the average regret by 29%, and SVM-CBO, a BO-based algorithm that integrates SVM models to determine the feasible region, by 82%.

贝叶斯优化（BO）是一组功能强大的算法，用于解决涉及昂贵的黑盒函数的复杂优化问题。这些顺序算法迭代地更新目标函数（of）的代理模型，有效地平衡探索和利用，以在有限的迭代次数内确定接近最优的解决方案。最初是为连续的、无约束的领域设计的，它的效率激发了对离散的、有约束的优化问题的适应。另一方面，机器学习（ML）模型允许对黑箱函数进行准确的预测，尽管它们通常需要大量的数据进行训练。利用BO和ML的优势，研究解决了在云计算环境中识别最佳配置的挑战。由于能够提供灵活和可扩展的资源，这种范式已经变得普遍。确定最佳的硬件软件配置对于在满足服务质量约束的同时最小化成本至关重要。该任务涉及在有限的迭代次数内解决多维离散域和黑盒目标函数和约束上的复杂优化问题。为了应对这一挑战，这项工作引入了d-MALIBOO，这是一种基于bo的算法，它集成了ML技术，以提高在离散和有界域中寻找近最优解的效率。BO构建of的代理模型，ML模型确定黑箱约束的可行区域，并引导BO算法走向离散域的有希望区域。此外，我们还引入了一种“贪婪”方法，以便在具有多个局部最优解的领域中进行勘探。实验结果表明，我们的算法比OpenTuner（一种流行的约束优化框架）的平均遗憾率降低了29%，比SVM- cbo（一种基于bo的集成SVM模型来确定可行区域的算法）的平均遗憾率降低了82%。

{"title":"Discrete Bayesian Optimization via Machine Learning","authors":"Roberto Sala, Bruno Guindani, Danilo Ardagna, Alessandra Guglielmi","doi":"10.1016/j.peva.2025.102487","DOIUrl":"10.1016/j.peva.2025.102487","url":null,"abstract":"<div><div>Bayesian Optimization (BO) is a family of powerful algorithms designed to solve complex optimization problems involving expensive black-box functions. These sequential algorithms iteratively update a surrogate model of the objective function (OF), effectively balancing exploration and exploitation to identify near-optimal solutions within a limited number of iterations. Originally designed for continuous, unconstrained domains, its efficiency has inspired adaptations for discrete, constrained optimization problems. On the other hand, Machine Learning (ML) models allow accurate predictions for black-box functions, although they typically require large amounts of data for training. Leveraging the strengths of BO and ML, research tackles the challenge of identifying optimal configurations in the context of cloud computing. This paradigm has become pervasive due to its ability to provide flexible and scalable resources. Identifying the optimal hardware-software configuration is essential for minimizing costs while meeting Quality of Service constraints. This task involves solving complex optimization problems over multidimensional discrete domains and black-box objective functions and constraints, within a limited number of iterations. To address this challenge, this work introduces <em>d-MALIBOO</em>, a BO-based algorithm that integrates ML techniques to enhance the efficiency of finding near-optimal solutions in discrete and bounded domains. While BO builds the surrogate model of the OF, ML models determine the feasible region of the black-box constraints and guide the BO algorithm toward promising regions of the discrete domain. Furthermore, we introduce an <span><math><mi>ɛ</mi></math></span>-greedy approach to favor exploration in domains with multiple local optima. Experimental results show that our algorithm outperforms OpenTuner, a popular framework for constrained optimization, by reducing the average regret by 29%, and SVM-CBO, a BO-based algorithm that integrates SVM models to determine the feasible region, by 82%.</div></div>","PeriodicalId":19964,"journal":{"name":"Performance Evaluation","volume":"169 ","pages":"Article 102487"},"PeriodicalIF":1.0,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134308","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Optimizing latency for caching with delayed hits in non-stationary environment 优化非固定环境中带有延迟命中的缓存的延迟

IF 1 4区计算机科学 Q4 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Performance Evaluation

Pub Date : 2025-05-20 DOI: 10.1016/j.peva.2025.102488

Bowen Jiang, Yubo Yang, Bo Jiang

Caching plays a crucial role in many latency-sensitive systems, including content delivery networks, edge computing, and microprocessors. As the ratio between system throughput and transmission latency increases, delayed hits in cache problems become more prominent. In real-world scenarios, object access patterns often exhibit a non-stationary nature. In this paper, we investigate the latency optimization problem for caching with delayed hits in a non-stationary environment, where object sizes and fetching latencies are both non-uniform. We first find that given known future arrivals, evicting the object with the larger size, a higher aggregate delay due to miss and arriving the farthest in the future brings more gains in reducing latency. Following our findings, we design an online learning framework to make cache decisions more effectively. The first component of this framework utilizes historical data within the training window to estimate the object’s non-stationary arrival process, modeled as a mixture of log-gaussian distributions. Subsequently, we predict future arrivals based on this estimated distribution. According to these predicted future arrivals, we can determine the priority of eviction candidates using our defined rank function. Experimental results on four real-world traces show that our algorithm consistently reduces latency by

2 % - 10 %

on average compared to state-of-the-art algorithms.

缓存在许多对延迟敏感的系统中起着至关重要的作用，包括内容交付网络、边缘计算和微处理器。随着系统吞吐量与传输延迟之比的增大，缓存中的延迟命中问题变得更加突出。在实际场景中，对象访问模式通常表现出非固定的性质。在本文中，我们研究了在对象大小和获取延迟都不均匀的非固定环境下，具有延迟命中的缓存的延迟优化问题。我们首先发现，给定已知的未来到达，驱逐具有较大尺寸的对象，由于错过和到达最远的未来而导致的更高的总延迟可以在减少延迟方面获得更多收益。根据我们的发现，我们设计了一个在线学习框架来更有效地做出缓存决策。该框架的第一个组成部分利用训练窗口内的历史数据来估计目标的非平稳到达过程，建模为对数高斯分布的混合。随后，我们根据这一估计分布预测未来的到达人数。根据这些预测的未来到达，我们可以使用我们定义的排名函数确定驱逐候选人的优先级。在四个真实世界轨迹上的实验结果表明，与最先进的算法相比，我们的算法平均可以减少2% - 10%的延迟。

{"title":"Optimizing latency for caching with delayed hits in non-stationary environment","authors":"Bowen Jiang, Yubo Yang, Bo Jiang","doi":"10.1016/j.peva.2025.102488","DOIUrl":"10.1016/j.peva.2025.102488","url":null,"abstract":"<div><div>Caching plays a crucial role in many latency-sensitive systems, including content delivery networks, edge computing, and microprocessors. As the ratio between system throughput and transmission latency increases, delayed hits in cache problems become more prominent. In real-world scenarios, object access patterns often exhibit a non-stationary nature. In this paper, we investigate the latency optimization problem for caching with delayed hits in a non-stationary environment, where object sizes and fetching latencies are both non-uniform. We first find that given known future arrivals, evicting the object with the larger size, a higher aggregate delay due to miss and arriving the farthest in the future brings more gains in reducing latency. Following our findings, we design an online learning framework to make cache decisions more effectively. The first component of this framework utilizes historical data within the training window to estimate the object’s non-stationary arrival process, modeled as a mixture of log-gaussian distributions. Subsequently, we predict future arrivals based on this estimated distribution. According to these predicted future arrivals, we can determine the priority of eviction candidates using our defined rank function. Experimental results on four real-world traces show that our algorithm consistently reduces latency by <span><math><mrow><mn>2</mn><mtext>%</mtext><mo>−</mo><mn>10</mn><mtext>%</mtext></mrow></math></span> on average compared to state-of-the-art algorithms.</div></div>","PeriodicalId":19964,"journal":{"name":"Performance Evaluation","volume":"169 ","pages":"Article 102488"},"PeriodicalIF":1.0,"publicationDate":"2025-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144124925","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

The Multiserver Job Queuing Model with two job classes and Cox-2 service times 具有两个作业类和Cox-2服务时间的多服务器作业队列模型

IF 1 4区计算机科学 Q4 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Performance Evaluation

Pub Date : 2025-05-19 DOI: 10.1016/j.peva.2025.102486

Adityo Anggraito , Diletta Olliaro , Andrea Marin , Marco Ajmone Marsan

Datacenters comprise a variety of resources (processors, memory, input/output modules, etc.) that are shared among requests for the execution of computing jobs submitted by datacenter users. Jobs differ in their frequency of arrivals, demand for resources, and execution times. Resource sharing generates contention, especially in heavily loaded systems, that must therefore implement effective scheduling policies for incoming jobs. The First-In First-Out (FIFO) policy is often used for batch jobs, but may produce under-utilization of resources, in terms of wasted servers. This is due to the fact that a job that requires many resources can block jobs arriving later that could be served because they require fewer resources. The mathematical construct often used to study this problem is the Multiserver Job Queuing Model (MJQM), where servers represent resources which are requested and used by jobs in different quantities. Unfortunately, very few explicit results are known for the MJQM, especially at realistic system loads (i.e., before saturation), and hardly any considers the case of non-exponential service time distributions. In this paper, we propose the first exact analytical model of the non-saturated MJQM in case of two classes of customers with service times having 2-phase Coxian distribution. Our analysis is based on the matrix geometric method. Our results provide insight into datacenter dynamics, thus supporting the design of more complex schedulers, capable of improving performance and energy consumption within large datacenters.

数据中心包含各种资源（处理器、内存、输入/输出模块等），这些资源在执行数据中心用户提交的计算作业的请求之间共享。作业的到达频率、资源需求和执行时间各不相同。资源共享会产生争用，特别是在负载沉重的系统中，因此必须为传入的作业实现有效的调度策略。先进先出（FIFO）策略通常用于批处理作业，但就浪费服务器而言，可能会导致资源利用率不足。这是因为需要大量资源的作业可能会阻塞稍后到达的作业，因为这些作业需要较少的资源。通常用于研究此问题的数学结构是多服务器作业排队模型（MJQM），其中服务器表示作业以不同数量请求和使用的资源。不幸的是，很少有关于MJQM的明确结果，特别是在实际系统负载下（即，在饱和之前），而且几乎没有考虑到非指数服务时间分布的情况。本文提出了两类服务时间具有两相协差分布的客户的非饱和MJQM的第一个精确解析模型。我们的分析是基于矩阵几何方法。我们的结果提供了对数据中心动态的洞察，从而支持更复杂的调度器的设计，能够提高大型数据中心的性能和能耗。

{"title":"The Multiserver Job Queuing Model with two job classes and Cox-2 service times","authors":"Adityo Anggraito , Diletta Olliaro , Andrea Marin , Marco Ajmone Marsan","doi":"10.1016/j.peva.2025.102486","DOIUrl":"10.1016/j.peva.2025.102486","url":null,"abstract":"<div><div>Datacenters comprise a variety of resources (processors, memory, input/output modules, etc.) that are shared among requests for the execution of computing jobs submitted by datacenter users. Jobs differ in their frequency of arrivals, demand for resources, and execution times. Resource sharing generates contention, especially in heavily loaded systems, that must therefore implement effective scheduling policies for incoming jobs. The First-In First-Out (FIFO) policy is often used for batch jobs, but may produce under-utilization of resources, in terms of wasted servers. This is due to the fact that a job that requires many resources can block jobs arriving later that could be served because they require fewer resources. The mathematical construct often used to study this problem is the Multiserver Job Queuing Model (MJQM), where servers represent resources which are requested and used by jobs in different quantities. Unfortunately, very few explicit results are known for the MJQM, especially at realistic system loads (i.e., before saturation), and hardly any considers the case of non-exponential service time distributions. In this paper, we propose the first exact analytical model of the non-saturated MJQM in case of two classes of customers with service times having 2-phase Coxian distribution. Our analysis is based on the matrix geometric method. Our results provide insight into datacenter dynamics, thus supporting the design of more complex schedulers, capable of improving performance and energy consumption within large datacenters.</div></div>","PeriodicalId":19964,"journal":{"name":"Performance Evaluation","volume":"169 ","pages":"Article 102486"},"PeriodicalIF":1.0,"publicationDate":"2025-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144131426","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Controller synthesis in timed Büchi automata: Robustness and punctual guards 定时<s:1> chi自动机的控制器综合：鲁棒性和准时保护

IF 1 4区计算机科学 Q4 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Performance Evaluation

Pub Date : 2025-04-05 DOI: 10.1016/j.peva.2025.102483

Benoît Barbot , Damien Busatto-Gaston , Catalin Dima , Youssouf Oualhadj

We consider the synthesis problem on timed automata with Büchi objectives, where delay choices made by a controller are subjected to small perturbations. Usually, the controller needs to avoid punctual guards, such as testing the equality of a clock to a constant. In this work, we generalize to a robustness setting that allows for punctual transitions in the automaton to be taken by controller with no perturbation. In order to characterize cycles that resist perturbations in our setting, we introduce a new structural requirement on the reachability relation along an accepting cycle of the automaton. This property is formulated on the region abstraction, and generalizes the existing characterization of winning cycles in the absence of punctual guards. We show that the problem remains within

PSPACE

despite the presence of punctual guards.

考虑具有 chi目标的时间自动机的综合问题，其中控制器所作的延迟选择受到小扰动的影响。通常，控制器需要避免守时保护，例如测试时钟与常数的相等性。在这项工作中，我们推广到一个鲁棒设置，该设置允许控制器在没有扰动的情况下采取自动机中的准时过渡。为了描述在我们的设置中抵抗扰动的周期，我们引入了一个新的结构要求，即自动机在一个接受周期上的可达关系。这一性质在区域抽象的基础上得到了表述，并推广了在没有准时守卫时获胜循环的现有特征。我们表明，尽管有准时的警卫，问题仍然存在于PSPACE。

引用次数: 0

A generalized result for the discrete-time two-queue randomly alternating service system 离散时间双队列随机交变服务系统的一个推广结果

IF 1 4区计算机科学 Q4 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Performance Evaluation

Pub Date : 2025-04-03 DOI: 10.1016/j.peva.2025.102484

Herwig Bruneel, Arnaud Devos, Joris Walraevens

In this paper, we revisit the discrete-time two-queue randomly alternating service system, where one common server is shared by two queues by allocating the server, independently from slot to slot, with fixed probabilities to either queue. Arrivals of new customers into the two-queue system occur independently from slot to slot, but may be mutually dependent within a slot. They are characterized by the joint probability generating function (pgf)

A (z_{1}, z_{2})

of the numbers of arrivals in both queues during one time slot. The service times of all customers are equal to exactly one time slot.

We extend various existing results with respect to the queueing behavior of this system. Specifically, we show that the exact solutions that were previously found for the steady-state joint pgf

U (z_{1}, z_{2})

of the system contents in both queues for the scenarios of independent Bernoulli arrivals, identical Bernoulli arrivals, global geometric arrivals, global geometric group arrivals, and the superposition of identical Bernoulli arrivals and global geometric (group) arrivals, are all special cases of a more general result, which is valid for a whole class of arrival pgfs

A (z_{1}, z_{2})

that (among others) encompasses the aforementioned specific arrival scenarios. However, the defined class is much broader than this, and our new result allows the solution for entirely new arrival pgfs as well. We illustrate this abundantly with a large number of detailed examples. The proof of the general result is a mainly algebraic one and, unlike earlier studies, does not require intricate techniques from complex-function analysis.

本文重新研究了离散时间双队列随机交替服务系统，其中一个公共服务器由两个队列共享，独立于插槽分配服务器，每个队列具有固定的概率。进入双队列系统的新客户在每个时段独立出现，但在一个时段内可能相互依赖。它们的特征是一个时隙内两个队列到达人数的联合概率生成函数（pgf） A（z1,z2）。所有客户的服务时间正好等于一个时隙。我们扩展了关于该系统的排队行为的各种现有结果。具体来说，我们证明了在独立伯努利到达、相同伯努利到达、全局几何到达、全局几何群到达以及相同伯努利到达和全局几何（群）到达叠加的情况下，两个队列中系统内容的稳态联合pgf U（z1,z2）的精确解都是一个更一般结果的特殊情况。它适用于包含前面提到的特定到达场景的整个到达pgfs a （z1,z2）类。但是，所定义的类要比这广泛得多，并且我们的新结果也允许全新到达pgfs的解决方案。我们用大量详细的例子来充分说明这一点。一般结果的证明主要是代数证明，与早期的研究不同，它不需要复杂的复函数分析技术。

{"title":"A generalized result for the discrete-time two-queue randomly alternating service system","authors":"Herwig Bruneel, Arnaud Devos, Joris Walraevens","doi":"10.1016/j.peva.2025.102484","DOIUrl":"10.1016/j.peva.2025.102484","url":null,"abstract":"<div><div>In this paper, we revisit the discrete-time two-queue <em>randomly alternating service</em> system, where one common server is shared by two queues by allocating the server, independently from slot to slot, with fixed probabilities to either queue. Arrivals of new customers into the two-queue system occur independently from slot to slot, but may be mutually dependent within a slot. They are characterized by the joint probability generating function (<em>pgf</em>) <span><math><mrow><mi>A</mi><mrow><mo>(</mo><msub><mrow><mi>z</mi></mrow><mrow><mn>1</mn></mrow></msub><mo>,</mo><msub><mrow><mi>z</mi></mrow><mrow><mn>2</mn></mrow></msub><mo>)</mo></mrow></mrow></math></span> of the numbers of arrivals in both queues during one time slot. The service times of all customers are equal to exactly one time slot.</div><div>We extend various existing results with respect to the queueing behavior of this system. Specifically, we show that the <em>exact</em> solutions that were previously found for the steady-state joint pgf <span><math><mrow><mi>U</mi><mrow><mo>(</mo><msub><mrow><mi>z</mi></mrow><mrow><mn>1</mn></mrow></msub><mo>,</mo><msub><mrow><mi>z</mi></mrow><mrow><mn>2</mn></mrow></msub><mo>)</mo></mrow></mrow></math></span> of the system contents in both queues for the scenarios of <em>independent Bernoulli</em> arrivals, <em>identical Bernoulli</em> arrivals, <em>global geometric</em> arrivals, <em>global geometric group</em> arrivals, and the superposition of <em>identical Bernoulli</em> arrivals and <em>global geometric (group)</em> arrivals, are all special cases of a more general result, which is valid for a whole class of arrival pgfs <span><math><mrow><mi>A</mi><mrow><mo>(</mo><msub><mrow><mi>z</mi></mrow><mrow><mn>1</mn></mrow></msub><mo>,</mo><msub><mrow><mi>z</mi></mrow><mrow><mn>2</mn></mrow></msub><mo>)</mo></mrow></mrow></math></span> that (among others) encompasses the aforementioned specific arrival scenarios. However, the defined class is much broader than this, and our new result allows the solution for entirely new arrival pgfs as well. We illustrate this abundantly with a large number of detailed examples. The proof of the general result is a mainly <em>algebraic</em> one and, unlike earlier studies, does not require intricate techniques from complex-function analysis.</div></div>","PeriodicalId":19964,"journal":{"name":"Performance Evaluation","volume":"168 ","pages":"Article 102484"},"PeriodicalIF":1.0,"publicationDate":"2025-04-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143785629","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Editorial: Special issue on timed and stochastic approaches to system evaluation: Selected papers from QESTFORMATS 2024 社论：关于系统评估的定时和随机方法的特刊：来自QESTFORMATS 2024的论文选集

IF 1 4区计算机科学 Q4 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Performance Evaluation

Pub Date : 2025-04-02 DOI: 10.1016/j.peva.2025.102485

引用次数: 0

Certificates and witnesses for multi-objective queries in Markov decision processes 马尔可夫决策过程中多目标查询的证书和见证

IF 1 4区计算机科学 Q4 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Performance Evaluation

Pub Date : 2025-03-24 DOI: 10.1016/j.peva.2025.102482

Christel Baier, Calvin Chau, Sascha Klüppelholz

Probabilistic model checking is a technique for formally verifying the correctness of probabilistic systems w.r.t. given specifications. Typically, a model checking procedure outputs whether a specification is satisfied or not, but does not provide additional insights on the correctness of the result, thereby diminishing the trustworthiness and understandability of the verification process. In this work, we consider certifying verification algorithms that also provide an independently checkable certificate and witness in addition to the verification result. The certificate can be used to easily validate the correctness of the result and the witness provides useful diagnostic information, e.g. for debugging purposes. More specifically, we study certificates and witnesses for specifications in the form of multi-objective queries in Markov decision processes. We first consider multi-objective reachability and invariant queries and then extend our techniques to mean-payoff expectation and mean-payoff percentile queries. Thereby, we generalize previous works on certificates and witnesses for single reachability and invariant constraints. In essence, we derive certifying verification algorithms from known linear programming techniques and show that witnesses, both in the form of schedulers and subsystems, can be obtained from the certificates. As a proof-of-concept, we report on an implementation of our certifying verification algorithms and present experimental results, demonstrating the applicability on moderately-sized case studies.

概率模型检查是一种在给定规范下正式验证概率系统正确性的技术。通常，模型检查过程输出规范是否满足，但不提供对结果正确性的额外见解，从而降低了验证过程的可信度和可理解性。在这项工作中，我们考虑认证验证算法，除了验证结果之外，还提供独立可检查的证书和证人。证书可以用来很容易地验证结果的正确性，而见证提供了有用的诊断信息，例如用于调试目的。更具体地说，我们研究了马尔可夫决策过程中多目标查询形式的规范证书和见证。我们首先考虑多目标可达性和不变查询，然后将我们的技术扩展到平均收益期望和平均收益百分位查询。因此，我们推广了以往关于单可达性和不变约束的证书和见证的工作。从本质上讲，我们从已知的线性规划技术中推导出认证验证算法，并表明可以从证书中获得证人，无论是调度程序还是子系统。作为概念验证，我们报告了我们的认证验证算法的实现并提出了实验结果，证明了在中等规模的案例研究中的适用性。

{"title":"Certificates and witnesses for multi-objective queries in Markov decision processes","authors":"Christel Baier, Calvin Chau, Sascha Klüppelholz","doi":"10.1016/j.peva.2025.102482","DOIUrl":"10.1016/j.peva.2025.102482","url":null,"abstract":"<div><div>Probabilistic model checking is a technique for formally verifying the correctness of probabilistic systems w.r.t. given specifications. Typically, a model checking procedure outputs whether a specification is satisfied or not, but does not provide additional insights on the correctness of the result, thereby diminishing the trustworthiness and understandability of the verification process. In this work, we consider certifying verification algorithms that also provide an independently checkable certificate and witness in addition to the verification result. The certificate can be used to easily validate the correctness of the result and the witness provides useful diagnostic information, e.g. for debugging purposes. More specifically, we study certificates and witnesses for specifications in the form of <em>multi-objective</em> queries in Markov decision processes. We first consider multi-objective reachability and invariant queries and then extend our techniques to mean-payoff expectation and mean-payoff percentile queries. Thereby, we generalize previous works on certificates and witnesses for single reachability and invariant constraints. In essence, we derive certifying verification algorithms from known linear programming techniques and show that witnesses, both in the form of schedulers and subsystems, can be obtained from the certificates. As a proof-of-concept, we report on an implementation of our certifying verification algorithms and present experimental results, demonstrating the applicability on moderately-sized case studies.</div></div>","PeriodicalId":19964,"journal":{"name":"Performance Evaluation","volume":"168 ","pages":"Article 102482"},"PeriodicalIF":1.0,"publicationDate":"2025-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143748192","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Optimizing spatial modulation MIMO IoT systems with full-duplex/half-duplex UAVs and enhanced transmit antenna selection 优化空间调制MIMO物联网系统与全双工/半双工无人机和增强的发射天线选择

IF 1 4区计算机科学 Q4 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Performance Evaluation

Pub Date : 2025-03-13 DOI: 10.1016/j.peva.2025.102481

Bao The Phung , Ba Cao Nguyen , Nguyen Van Vinh , Bui Vu Minh , Nguyen Huu Khanh Nhan

Multiple-input multiple-output (MIMO) systems play a crucial role in elevating the efficiency and reliability of communication networks, especially within Internet of Things (IoT) applications. This article introduces a novel approach involving full-duplex (FD) and half-duplex (HD) relays mounted on unmanned aerial vehicles (UAVs) to enhance MIMO systems. Incorporating spatial modulation (SM) and transmit antenna selection (TAS) techniques aims to optimize system performance while reducing computational complexity to meet IoT requirements. The article mathematically formulates outage probabilities (OPs) and system throughputs (STs) for the proposed MIMO-IoT-UAV systems utilizing SM with FD/HD-UAV, both with and without TAS, over practical Nakagami-

m

channels. Numerical illustrations underscore the advantages of employing FD/HD-UAV and TAS in MIMO-IoT-UAV systems. Specifically, OPs with TAS are significantly lower, while STs with TAS are notably higher than their counterparts without TAS. Additionally, TAS with FD-UAV yields greater benefits than HD-UAV, particularly in preventing the error floor associated with residual self-interference (RSI). To mitigate this error floor in MIMO-IoT-UAV systems using FD-UAV, an effective strategy involves increasing the number of transmit/receive antennas. The choice between FD and HD modes hinges on parameters such as transmit power, data rate, and RSI. Depending on these factors, FD-UAV performance may exhibit lower or higher error rates than HD-UAV. Hence, the optimal selection of FD or HD mode, combined with TAS, is essential for enhancing MIMO-IoT-UAV system performance. This optimization process should consider parameters like RSI level, the number of transmit/receive antennas, data rate requirements, and UAV position to ensure efficient and reliable communication across diverse scenarios.

多输入多输出（MIMO）系统在提高通信网络的效率和可靠性方面发挥着至关重要的作用，特别是在物联网（IoT）应用中。本文介绍了一种安装在无人机上的全双工（FD）和半双工（HD）继电器来增强MIMO系统的新方法。结合空间调制（SM）和发射天线选择（TAS）技术旨在优化系统性能，同时降低计算复杂性，以满足物联网要求。本文在数学上计算了在实际的Nakagami-m通道上使用SM和FD/HD-UAV的拟议MIMO-IoT-UAV系统的中断概率（OPs）和系统吞吐量（STs），包括有和没有TAS。数值实例强调了在MIMO-IoT-UAV系统中采用FD/HD-UAV和TAS的优势。具体而言，有TAS的OPs显著低于无TAS的OPs，而有TAS的STs显著高于无TAS的STs。此外，配备FD-UAV的TAS比HD-UAV具有更大的优势，特别是在防止与残余自干扰（RSI）相关的误差方面。为了在使用FD-UAV的MIMO-IoT-UAV系统中减轻这种误差层，一种有效的策略包括增加发射/接收天线的数量。FD和HD模式之间的选择取决于传输功率、数据速率和RSI等参数。取决于这些因素，FD-UAV的性能可能比HD-UAV表现出更低或更高的错误率。因此，优化选择FD或HD模式，并结合TAS，对于提高MIMO-IoT-UAV系统性能至关重要。该优化过程应考虑RSI水平、发射/接收天线数量、数据速率要求和无人机位置等参数，以确保在不同场景下高效可靠的通信。

{"title":"Optimizing spatial modulation MIMO IoT systems with full-duplex/half-duplex UAVs and enhanced transmit antenna selection","authors":"Bao The Phung , Ba Cao Nguyen , Nguyen Van Vinh , Bui Vu Minh , Nguyen Huu Khanh Nhan","doi":"10.1016/j.peva.2025.102481","DOIUrl":"10.1016/j.peva.2025.102481","url":null,"abstract":"<div><div>Multiple-input multiple-output (MIMO) systems play a crucial role in elevating the efficiency and reliability of communication networks, especially within Internet of Things (IoT) applications. This article introduces a novel approach involving full-duplex (FD) and half-duplex (HD) relays mounted on unmanned aerial vehicles (UAVs) to enhance MIMO systems. Incorporating spatial modulation (SM) and transmit antenna selection (TAS) techniques aims to optimize system performance while reducing computational complexity to meet IoT requirements. The article mathematically formulates outage probabilities (OPs) and system throughputs (STs) for the proposed MIMO-IoT-UAV systems utilizing SM with FD/HD-UAV, both with and without TAS, over practical Nakagami-<span><math><mi>m</mi></math></span> channels. Numerical illustrations underscore the advantages of employing FD/HD-UAV and TAS in MIMO-IoT-UAV systems. Specifically, OPs with TAS are significantly lower, while STs with TAS are notably higher than their counterparts without TAS. Additionally, TAS with FD-UAV yields greater benefits than HD-UAV, particularly in preventing the error floor associated with residual self-interference (RSI). To mitigate this error floor in MIMO-IoT-UAV systems using FD-UAV, an effective strategy involves increasing the number of transmit/receive antennas. The choice between FD and HD modes hinges on parameters such as transmit power, data rate, and RSI. Depending on these factors, FD-UAV performance may exhibit lower or higher error rates than HD-UAV. Hence, the optimal selection of FD or HD mode, combined with TAS, is essential for enhancing MIMO-IoT-UAV system performance. This optimization process should consider parameters like RSI level, the number of transmit/receive antennas, data rate requirements, and UAV position to ensure efficient and reliable communication across diverse scenarios.</div></div>","PeriodicalId":19964,"journal":{"name":"Performance Evaluation","volume":"168 ","pages":"Article 102481"},"PeriodicalIF":1.0,"publicationDate":"2025-03-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143642833","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Approximation of cumulative distribution functions by Bernstein phase-type distributions 用Bernstein相型分布逼近累积分布函数

IF 1 4区计算机科学 Q4 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Performance Evaluation

Pub Date : 2025-03-11 DOI: 10.1016/j.peva.2025.102480

András Horváth , Illés Horváth , Marco Paolieri , Miklós Telek , Enrico Vicario

The inclusion of generally distributed random variables in stochastic models is often tackled by choosing a parametric family of distributions and applying fitting algorithms to find appropriate parameters. A recent paper proposed the approximation of probability density functions (PDFs) by Bernstein exponentials, which are obtained from Bernstein polynomials by a change of variable and result in a particular case of acyclic phase-type distributions. In this paper, we show that this approximation can also be applied to cumulative distribution functions (CDFs), which enjoys advantageous properties and achieves similar accuracy; by focusing on CDFs, we propose an approach to obtain stochastically ordered approximations. The use of a scaling parameter in the approximation is also presented, evaluating its effect on approximation accuracy.

在随机模型中包含一般分布的随机变量通常是通过选择一个参数分布族和应用拟合算法来找到合适的参数来解决的。最近的一篇论文提出了用伯恩斯坦指数逼近概率密度函数（pdf）的方法，该方法是由伯恩斯坦多项式通过变量的变化得到的，并导致了非循环相型分布的特殊情况。在本文中，我们证明了这种近似也可以应用于累积分布函数（CDFs），它具有优越的性质并达到类似的精度；通过关注CDFs，我们提出了一种获得随机有序近似的方法。文中还介绍了尺度参数在近似中的应用，并评价了尺度参数对近似精度的影响。

引用次数: 0

Optimizing parallel I/O performance in NVMe SSDs by Dynamic cache partitioning 通过动态缓存分区优化NVMe ssd的并行I/O性能

IF 1 4区计算机科学 Q4 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Performance Evaluation

Pub Date : 2025-03-10 DOI: 10.1016/j.peva.2025.102479

Zecheng Li , Shu Yin , Xiaojun Ruan

Solid State Drive cache, implemented as on-board shared DRAM memory, can significantly enhance I/O performance by caching frequently accessed data. Although SSD caching strategies for single I/O data flows have been extensively explored, studies on cache partitioning to optimize parallel I/O in an SSD are scarce. In this paper, we present a novel dynamic cache partitioning approach designed to improve overall performance of multi-parallel I/O data flows by minimizing performance degradation of cache pollution and resource contention. By dynamically adjusting cache partition sizes for each data flow by considering cache sensitivity on performance, our strategy seeks to determine the optimal cache partition sizes to maximize overall I/O throughput. We implemented the strategy in the SSD simulator MQSim and evaluated its performance using various synthetic and real-world workloads. Our experimental results indicate that our dynamic cache partitioning strategy achieves an overall throughput increase of up to 33.22% compared to shared cache methods and outperforms static cache partitioning strategies by up to 21.19%.

固态硬盘缓存实现为板载共享DRAM内存，可以通过缓存频繁访问的数据来显著提高I/O性能。尽管针对单个I/O数据流的SSD缓存策略已经得到了广泛的研究，但针对SSD中并行I/O优化的缓存分区研究却很少。在本文中，我们提出了一种新的动态缓存分区方法，旨在通过最小化缓存污染和资源争用的性能降低来提高多并行I/O数据流的整体性能。通过考虑缓存对性能的敏感性，动态调整每个数据流的缓存分区大小，我们的策略旨在确定最佳缓存分区大小，以最大化总体I/O吞吐量。我们在SSD模拟器MQSim中实现了该策略，并使用各种合成工作负载和实际工作负载评估了其性能。我们的实验结果表明，与共享缓存方法相比，我们的动态缓存分区策略的总体吞吐量提高了33.22%，比静态缓存分区策略的吞吐量提高了21.19%。

{"title":"Optimizing parallel I/O performance in NVMe SSDs by Dynamic cache partitioning","authors":"Zecheng Li , Shu Yin , Xiaojun Ruan","doi":"10.1016/j.peva.2025.102479","DOIUrl":"10.1016/j.peva.2025.102479","url":null,"abstract":"<div><div>Solid State Drive cache, implemented as on-board shared DRAM memory, can significantly enhance I/O performance by caching frequently accessed data. Although SSD caching strategies for single I/O data flows have been extensively explored, studies on cache partitioning to optimize parallel I/O in an SSD are scarce. In this paper, we present a novel dynamic cache partitioning approach designed to improve overall performance of multi-parallel I/O data flows by minimizing performance degradation of cache pollution and resource contention. By dynamically adjusting cache partition sizes for each data flow by considering cache sensitivity on performance, our strategy seeks to determine the optimal cache partition sizes to maximize overall I/O throughput. We implemented the strategy in the SSD simulator MQSim and evaluated its performance using various synthetic and real-world workloads. Our experimental results indicate that our dynamic cache partitioning strategy achieves an overall throughput increase of up to 33.22% compared to shared cache methods and outperforms static cache partitioning strategies by up to 21.19%.</div></div>","PeriodicalId":19964,"journal":{"name":"Performance Evaluation","volume":"168 ","pages":"Article 102479"},"PeriodicalIF":1.0,"publicationDate":"2025-03-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143609296","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0