Search

Container clusters play an increasingly important role in cloud computing for processing dynamic computing tasks. The resource manager (i.e., orchestrater) of the cluster automates the scheduling of the dynamic requests, effectively manages the resources’ utilization across distributing infrastructure resources. For many applications, the requests to the cluster are often with restricted deadlines. The scheduling of container clusters is often tricky, especially when the cluster’s size is large and the load of the requests is dynamically changing. Machine learningbased approaches such as reinforcement learning have attracted lots of research attention during the past years; However, those approaches suffer from low robustness when the requests in an operational environment are changing and different from the training data sets. This paper investigates this problem by quantifying the robustness and proposing meta-gradient reinforcement learning to improve the robustness of classical reinforcement learning-based approaches. The proposed approach can lead to better deadline guarantees and faster adaptation for timecritical task scheduling under dynamic environments. We then empirically test the benefits of our method using both real-world and synthetic data sets. The evaluation results show that the proposed method outperforms the compared RL methods in scheduling performance and robustness.

International audience; This double volumes LNCS 11229-11230 constitutes the refereed proceedings of the Confederated International Conferences: Cooperative Information Systems, CoopIS 2018, Ontologies, Databases, and Applications of Semantics, ODBASE 2018, and Cloud and Trusted Computing, C&TC, held as part of OTM 2018 in October 2018 in Valletta, Malta. The 64 full papers presented together with 22 short papers were carefully reviewed and selected from 173 submissions. The OTM program every year covers data and Web semantics, distributed objects, Web services, databases, informationsystems, enterprise workflow and collaboration, ubiquity, interoperability, mobility, grid and high-performance computing.

Distributing software processes by software producing organizations (SPOs) is emerging increasingly due to benefits that global software engineering (GSE) brings in terms of cost reduction, leveraging competencies, and market expansion. However, these organizations are facing communication and project control issues that can slow down the overall organization performance. Therefore, SPOs should be able to manage inter-dependencies among the tasks distributed to the globally dispersed teams. We analyze existing works and product software companies’ best practices in coordinating tasks in GSE. This paper specifically focuses on constructing methodological support for task coordination that can be influenced by the situational factors at the companies. The support comprises a framework and a method developed by using a method engineering approach. We introduce the framework that depicts the aspects that should be examined by companies and the method that elaborates the practices to guide companies to coordinate tasks in GSE projects. The validation results show that the framework and the method are accepted by experts regarding completeness and applicability to help SPOs in managing coordination among globally distributed teams.

The competition in globalized markets forces organizations to provide mass-customized integrated solutions for customers. Mass-customization of integrated solutions by business network requires adaptive interactions between parties to address emerging requirements of customers. These adaptive interactions need to be enabled by dynamic networked business processes (DNBP) that are supported by high quality information. However, the dynamic collaboration between parties can result in information quality (IQ) issues such as information syntactic and semantic misalignment, information leakage, and unclear information ownership. To counter negative consequences of poor IQ on the performance, the orchestrator of business network needs to clearly recognize these IQ issues. In this paper, we develop and evaluate a framework to address potential IQ issues related to DNBP. The development of the framework is based on a three step methodology that includes the characterization of dynamism of networked business processes, the characterization of IQ dimensions, and the exploration of IQ issues. To evaluate the practical significance of the explored IQ issues, we conduct a case study in a service ecosystem that is shaped by a car leasing organization to provide integrated mobility solutions for customers.

The separation design and fabrication process in the semiconductor industry leads to potential threats such as trojan side-channels (TSCs). In this paper we design a new family of TSCs from physical unclonable functions (PUFs). In particular, a dedicated attack on the PRESENT block cipher is described by using our PUF-based TSCs. Finally we analyze the performance of our PUF-based TSCs and discuss other potential applications.

Social Sciences identify similarity and mutual trust as main criteria to consider in group formation processes. On this basis, we present a group formation technique which exploits measures of both similarity and trust, in order to improve the compactness of groups in Online Social Networks. Similarity and trust have been jointly exploited to design two algorithms designed to match groups and users, in order to capture the gain of a user who desires to join with a group and the benefit of the group itself. Experimental results show that trust is more valuable than similarity in forming groups and that the two proposed algorithms are capable to deal with large networks.