Companies increasingly integrate Artificial Intelligence (AI) into their applications to stay competitive. However, the efficient and successful development and deployment of AI applications requires complex setups. Providing developers with all the required resources, applications, and services for developing and deploying AI applications without reinventing the wheel remains challenging. Therefore, we propose a domainand workflow-agnostic reference architecture (RA) for an onpremises AI Platform that supports teams throughout the entire AI lifecycle and is reusable across multiple projects. Additionally, we present an evaluation strategy to validate the RA.

Software testing is vital for strengthening the security of embedded systems by identifying and rectifying code errors, flaws and vulnerabilities. This is particularly significant when addressing vulnerabilities associated with side-channel attacks, given that they introduce a distinctive class of vulnerabilities, primarily subject to manual testing procedures. Manual testing remains prevalent despite advances in automation, posing challenges, particularly for complex environments. This research aims to automate embedded software testing on hardware in a modular and scalable manner, addressing the limitations of manual testing. We present a system designed to automate testing, including Side-Channel Analysis (SCA), in Continuous Integration (CI) environments, emphasizing accessibility and collaboration through open-source tools. Our evaluation setup based on GitLab, Jenkins and the ChipWhisperer framework shows that automating and integrating Side-Channel Analysis (SCA) in Continuous Integration (CI) environments is possible in an efficient way.

As soon as Artificial Intelligence (AI) projects grow from small feasibility studies to mature projects, developers and data scientists face new challenges, such as collaboration with other developers, versioning data, or traceability of model metrics and other resulting artifacts. This paper suggests a data-centric AI project with an Active Learning (AL) loop from a developer perspective and presents ”Git Workflow for AL”: A methodology proposal to guide teams on how to structure a project and solve implementation challenges. We introduce principles for data, code, as well as automation, and present a new branching workflow. The evaluation shows that the proposed method is an enabler for fulfilling established best practices.

As of now, intelligent technologies experience a rapid growth. For a reliable adoption of those new and powerful systems into day-to-day life, especially with respect to high-risk settings such as medicine, technical means to realize legal requirements correctly, are indispensible. Our proposed methodology comprises an approach to translate such partly more abstract concepts into concrete instructions - it is based on Quality Gates along the intelligent system’s complete life cycle, which are composed of use-case adapted Criteria that need to be addressed with respect to certification. Also, the underlying philosophy regarding stakeholder inclusion, domain embedding and risk analysis is illustrated. In the present paper, the Quality Gate Metrics is outlined for the application of machine learning performance metrics focused on binary classification.

Quality of system and software architectures plays a major role in today's development to reduce the complexity of systems during design and ensure maintainability and extensibility. Metrics can provide system architects with information about the quality of the system architectures and support them to ensure the quality characteristics defined in ISO 25010. In this position paper, we look on selected metrics from code and object-oriented design analysis to use them for the evaluation of system architectures modeled with EAST-ADL. Therefore, the metrics (collections) Lines of Code, Cyclomatic Complexity, Chidamber and Kemerer and MOOD are examined for their application on EAST-ADL models.

Over the last decades, systems immanent complexity has significantly increased. In order to cope with the emerging challenges during the development of such systems, modeling approaches become an indispensable part. While many process steps are applicable to the model-level, there are no sufficient realizations for test execution yet. As a result, we present a semi-formal approach enabling developers to perform abstract test execution straight on the modeled artifacts to support the overarching objective of a shift left of verification and validation tasks. Our concept challenges an abstract test case (derived from test model) against a system model utilizing an integrated set of domain-specific models, i.e. the omni model. Driven by an optimistic dataflow analysis based on a combined view of an abstract test case and its triggered system behavior, possible test verdicts are assigned. Based on a prototypical implementation of the concept, the proof of concept is demonstrated and further on put in the context of related research.

Network virtualization and network slicing are key-features of the fifth generation (5G) of mobile networks to overcome the challenge of increasingly diverging network requirements emerging from new use-cases like IoT, autonomous driving and Industry 4.0. In particular, low latency network slices require deployment of their services and applications, including Network Functions (NFs), close to the user, i.e., at the edge of the mobile network. Since the users of those services might be widely distributed and mobile, multiple instances of the same application are required to be available on numerous distributed edge clouds. This paper tackles the problem of Network Slice Embedding (NSE) with edge computing. Based on an Integer Linear Program (ILP) formulation of the NSE problem, the optimal set of network slices, in terms of network slice revenue and cost, is determined. The required network slice applications, functions and services are allocated nearly optimally on the 5G end-to-end mobile network infrastructure. The presented solution also provides the optimal number of application instances and their optimal deployment locations on the edge clouds, even for multiple User Equipment (UE) connectivity scenarios. Evaluation shows that the proposed holistic approach for NSE and multiple edge cloud allocation is feasible and efficient for small and medium sized problem instances.

The complexity in automotive systems engineering is increasing over the last decade. In particular, new comfort functions as well as functions towards autonomous driving are reasons for this complexity. A new dimension is introduced by the usage of multi-core processors since there is a shift from sequential to parallel thinking in the different development phases. Therefore, in this paper we present an approach for supporting the development process of distributed systems aligned with the EAST-ADL approach, by using partitioning. We present an extension to EAST-ADL for partitioning and show ways how an automatic partitioning on different levels of abstraction can be achieved. These partitions can support system designers during the design process of functional architectures, by giving a first insight how well the functional components can be distributed on hardware in later stages of the development process.

Nowadays, modern cars can be configured by means of a wide range of software configuration options.In this context, we speak about Software Product Lines (SPLs).In almost every modern automotive vehicle safetycritical components like, e.g., an Adaptive Cruise Control (ACC) are installed.Some SPLs have different safety-critical requirements whereas other SPLs have similar requirements.This paper proposes an approach for the reduction of the complexity of SPLs without loss of safety (aspects) for all participants.For this purpose, a concept has been developed, which clusters products of SPLs with similar safety-critical requirements, i.e., the set of products of an SPLs, which must still be tested, can be reduced immensely.The paper also provides an application example how the reduced set can be used in order to perform a Safety, Security and Timing (SST) based trade-off analysis.

Development of embedded systems depends on both software and hardware design activities. Quality management of development artifacts is of crucial importance for the overall function and safety of the final product. This paper introduces a metamodel and model-based approach and accompanying domain-specific language for validation of electronic circuit designs. The solution does not depend on any particular electronic design automation (EDA) software, instead we propose a workflow that is integrated into a modular, general-purpose model analysis framework. The paper illustrates both the underlying concepts as well as possible application scenarios for the developed validation domain-specific language, MBDRC (Model-Based Design Rule Checking). It also discusses fields for further research and transfer of the initial results.

Internet of Things (IoT) is spreading increasingly in different areas of application. Accordingly, IoT also gets deployed in health care including ambient assisted living, telemedicine or medical smart homes. However, IoT also involves risks. Next to increased security issues also safety concerns are occurring. Deploying health care sensors and utilizing medical data causes a high need for IoT architectures free of vulnerabilities in order to identify weak points as early as possible. To address this, we are developing a safety and security analysis approach including a standardized meta model and an IoT safety and security framework comprising a customizable analysis language.

Modern cars are equipped with a large number of electronic assistance systems such as Adaptive Cruise Control (ACC) to improve road safety and driving comfort.These systems require a complex cross-linking, both inside and outside the vehicle, e.g., by means of bus systems or wireless interfaces like Bluetooth.Thus, safety of road users can endangered if the communication between these systems failed.Communication failures can be affected by hacking attacks, e.g., delayed decelerating of an ACC system, thereby presenting a security and timing vulnerability endangering safety of road users.Hence, in this paper safety is considered as primary goal.Goals that contribute to achieve the primary goal can be in contradiction to each other under certain circumstances.Therefore, an approach is proposed to model Safety, Security and Timing (SST) constraints to guarantee maximum safety.Furthermore, a preventative risk assessment of the individual concerns including a tradeoff analysis is performed to enable the development of Safety-Critical Systems (SCS).

Effective and sufficient testing has always been a challenging task in software development.The ongoing increase of software complexity forces developers and testers to make extensive use of the concept of abstraction, thereby leading to model-based approaches.Further, standardization organizations aim for harmonized process templates to assure a certain quality level of the processes behind.In order to combine the process-level advice as well as the concept-level advice, we aim for a consistent and strict application of model-based methodologies throughout the testing processes, introduced by the ISO/IEC/IEEE 29119 standard for software testing.After a brief introduction to the standards content and a critical view on it, we focus on our model-based interpretation of the postulated processes.Thereby, we extend the original idea of model-based testing, incorporating the separation of concerns on the model-level, to form a broad information basis.Subsequent activities are aligned with these concepts, in order to make sure a purely model-based testing life cycle, with respect to consistency and quality of development artifacts.Following the related work of impacted research areas, we end up with a conclusive statement on the intended combination of approaches.

The increasing complexity of modern embedded systems demands advanced design and development methods.Incremental evolution of model-based engineering practice has led to heterogeneous tool environments without proper integration and exchange of design artifacts.These problems are especially prevalent in tightly regulated domains, where an independent assessment is required for newly developed products, e.g. in automotive or aviation systems.To address these shortcomings of current engineering practice, we propose a holistic model-based approach for the seamless design and development of an integrated system model.We describe an embedding of a variety of domain-specific modeling languages into a common general-purpose modeling language, in order to facilitate the integration between heterogeneous design artifacts.Based on this conceptual modeling approach, we introduce a framework for automated model-based analysis of integrated system models.A case study demonstrates the suitability of this modeling and analysis approach for the design of a safety-critical embedded system, a hypothetical gas heating burner.

Summary The automotive industry seeks to include more and more features in its vehicles. For this purpose, the necessary policy shift towards multi‐core technology is in full swing. To eventually exploit the extra processing power, there is much additional effort needed for coping with the tremendously increased complexity. This is largely due to the elaborate parallelization process that spans a vast search space. Consequently, there is a strong need for innovative methods and appropriate tools for the migration of legacy single‐core software. We use the results of a data dependency analysis performed on AUTOSAR system descriptions to determine advantageous partitions as well as initial task‐to‐core mappings. Afterwards, the extracted information serves as input for the simulation within a multi‐core timing tool suite. Here, the initial solution is evaluated with respect to proper scheduling and metrics like cross‐core communication rates, communication latencies, or core load distribution. A subsequent optimization process improves the initial solution and enables a comparative assessment. To demonstrate the benefit, we substantially expand a previous case study by applying our approach to two complex engine management systems and by showing the advantages compared to a parallelization process without preceding dependency analysis and initial partition/mapping suggestions.

Test automation in distributed systems requires new methods in signal simulation for the stimulation of the distributed system.Increasing complexity of electric electronic (E/E) systems enhances the testing-effort.The main challenge is reducing the time consuming manual stimulation in consideration of improving the quality of testing.Currently used systems for test automation with a software-based approach have to be adapted to each hardware and software version of the system to be tested.The approach represented shows a new approach through the integration of a simulation service in the AUTOSAR software architecture.By integrating a generic software-based simulation module with an interaction point at the basic software driver layer, the execution of tests can be automated and improved under consideration of adaptivity and reproducibility.

Emerging scenarios of vertical industries, such as, adaptive manufacturing, cooperative autonomous driving, and real-time logistics demand for seamless communication among mobile entities, among them ground conveyors, user terminals, cars, or sensors. In practice, this leads to challenging requirements in terms of latency, bandwidth, availability, reliability, etc. that current mobile communications technologies, including cellular networks and IEEE 802-based solutions, do not fully account for. When deployed individually, neither of them fulfills the broad range of requirements, whereas a coordinated co-deployment suffers from the lack of efficient network management solutions as well as from insufficiently defined operating and owner structures for these mixed system deployments. Therefore, this work proposes a concept for an intelligent cross-domain 5G network management system and related optimization functions. In particular, the design of a cognitive, joint management of mobile industrial and cellular networks is outlined. Further, cognitive methods and virtualization techniques are motivated as major enablers. Advantages include an operator-grade network management of local industrial networks as well as a seamless integration with cellular networks. Following 5G design principles, the suggested network management system is also extensible to emerging radio access technologies (RATs) in the 5G context.

The operation of a Self-Organizing Network (SON) such that it achieves the objectives of a cellular network operator is a complex task. The previously presented SON objective manager enables the automatic management of SON functions based on formalized operator objectives on network Key Performance Indicators (KPIs). However, this approach initially did not consider SON coordination which resolves runtime conflicts between concurrently executed SON functions. We present an objective-driven SON coordination which adapts the SON objective manager approach to SON function conflict resolution. Therefore, the SON objective manager concept is based on multiattribute utility theory and the conflict resolution is modeled as a constraint optimization problem. As a result, this approach coordinates the execution of SON functions such that the operator objectives are satisfied as quickly as possible.

As the rise of single-core processing power is exhausted due to technical limitations, the automotive branch is forced to migrate its control unit software to architectures that feature multiple Independent Execution Units (IEUs).This policy shift brings along new problems resulting from the tremendously increased complexity of such systems.Facing these challenges, software engineers have to cope with possible data inconsistencies caused by, e.g., race conditions or cycles.Being an important and standardized software architecture for electronic control units, the Automotive Open System Architecture (AUTOSAR) provides the basis for tools that support the complexity handling when migrating to architectures with multiple IEUs.Our concept is realized by a tool that executes data dependency analyses directly on AUTOSAR models, determines critical dependencies, automatically solves trivial problems and provides semi-automatic resolution of advanced conflicts.To support the actual parallelization of the system, the tool additionally determines groups of executable units that are suitable to run on a common IEU.This appreciably facilitates the validation of AUTOSAR models and the search for a good mapping of the processing tasks to IEUs.

The Self-Organizing Network (SON) paradigm enables automated management of next generation mobile communication networks by introducing a set of autonomous functions that jointly achieve self-configuration, self-optimization, and self-healing. Each SON function is driven by a specific objective to optimize one or more Key Performance Indicators (KPIs). If a function encounters situations in which the given targets cannot be achieved, it can produce undesired behavior, resulting in an impaired SON. This paper presents the idea of SON Operational Troubleshooting function and outlines a concept for the automatic detection, analysis, and mitigation of operational problems.

Policy-based management is a flexible approach for the management of networks as policies make context-sensitive and automated decisions. For their effective development it is desired to specify policies at a high level of abstraction initially and to refine them until they are represented in a machine-executable way. We present an approach that uses models to specify event-condition-action (ECA) policies at different abstraction layers and that uses model transformations to refine them in an automated way. A relational algebra is used to formally validate the models and define the semantics of the refinement process. One benefit of the approach is the automated policy refinement at runtime. Changes at the high-level models are automatically reflected in their low-level implementation through refinement. This allows to manage a system at a high level of abstraction. The approach is applied to the network management domain and demonstrated with policies for physical cell identification (PCI) in a mobile network. It can also be applied to other domains and supports a flexible number of abstraction layers.

Policy-based management allows to adapt systems to changed requirements in a flexible and automated way. Policy development usually starts with the specification of high-level policies, which are then refined into a low-level representation. We use models to specify event-condition-action (ECA) policies at different levels of abstraction and consequently separate domain and policy aspects from each other. Domain-specific concepts are used within policies in their event, condition, and action parts. We present a formal specification of the models by means of a relational algebra. The algebra is used to validate the models at each level. Finally, executable policy code is generated from the low-level models.

Future wireless networks will experience a continuous growth regarding the number of network elements with increasingly complex interrelations between the configuration of multiple network elements. Another trend is the seamless integration of multiple radio technologies into a single wireless network. Both developments increase network management complexity and require new management concepts with a very high degree of automation. For this purpose, a novel management approach based on a combination of workflow and policy technologies is presented. The goal is to simplify and automate management tasks in mobile networks in order to raise the state of self-organization while the network still remains under control of the operator. The approach provides the means for a dynamic system to automatically adapt to context changes. Moreover, an experimental system is presented for the purposes of concept validation and evaluation along with a real world use case.

The trend for future mobile networks is to move away from Network Elements (NEs) delivered with specially tailored configurations towards off-the-shelf products. The configurations of NEs are automatically created with respect to their context including information on location and configuration of neighboring NEs. To minimize time-consuming and error-prone human interaction, automatic behavior is required for all stages of a NE's life cycle. The possibility to pre-assess the effects of configuration changes is inevitable in order to avoid service degradation caused by unnecessary reconfigurations. Graph coloring-based Physical Cell ID (PCID) assignment for LTE networks was introduced previously. The foundation on graph coloring theory allowed to transfer knowledge from this domain to the task of PCID assignment in order to pre-asses if an assignment is possible and how many PCIDs are required. Now the focus lies on adaptations of the basic approach to satisfy additional operator requirements such as safety margins. Those adaptations should provide equally good results in terms of used PCIDs with only minimal impact on costs and operation and maintenance tasks. Variations of the basic PCID assignment approach are discussed to address other types of problems.

Collaborative research in various scientific disciplines requires sup-port for scalable data management enabling the efficient correlation of globally distributed data sources. Motivated by the expected data rates of upcoming projects and a growing number of users, com-munities explore new data management techniques for achieving high throughput. Community-driven data grids deliver such high-throughput data distribution for scientific federations by partition-ing data according to application-specific data and query character-istics. Query hot spots are an important and challenging problem in this environment. Existing approaches to load-balancing from Peer-to-Peer (P2P) data management and sensor networks do not directly meet the requirements of a data-intensive e-science envi-ronment. In this paper, our contributions are partitioning schemes based on multi-dimensional index structures enabling communities to trade off data load balancing and handling query hot spots via splitting and replication. We evaluate the partitioning schemes with two typical kinds of data sets from the astrophysics domain and workloads extracted from Sloan Digital Sky Survey (SDSS) query traces and perform throughput measurements in real and simulated networks. The experiments demonstrate the improved workload distribution capabilities and give promising directions for the de-velopment of future community grids. 1.

In this paper, we present a new approach for service discovery combining semantic web and peer-to-peer techniques. A reference ontology is used to describe and discover services in our approach. We do not need a central point of control at any time. All information required for service description and discovery is completely distributed across the nodes of a peer-to-peer overlay network. We describe the design of a Semantically- enhanced Distributed Discovery System (SDDS) allowing dynamic and efficient registration and discovery of services. We also present performance analysis and discuss open issues of our system.

To gain wider acceptance for the use of agents in industry, it is a necessity to relate it to the nearest antecedent technology (object-oriented software development) and to introduce appropriate artifacts to support the development environment throughout the full system life cycle. We address both of these requirements by presenting AGENT UML, the Agent UML (Unified Modeling Language) — a set of UML idioms and extensions. This paper provides an AGENT UML representation of the internal behavior of an agent; it then relates this internal description to the external behavior of the agent by using and extending UML class diagrams and by describing agent interaction protocols in a new way. Our claim is that by extending the de-facto standard representation for object-oriented modeling to agents, the learning curve for object-oriented developers to adopt aspects of agent-based programming becomes much less steep. Thus, agent-oriented programming as a whole will become more amenable to mainstream software engineering.