Flow delegation is a flexible technique to mitigate flow table capacity bottlenecks in Software-defined Networks (SDN). Such bottlenecks occur when SDN switches provide insufficient flow table capacity which leads to performance degradation and network failures. Flow delegation addresses this problem by automatically relocating flow rules from a bottlenecked switch to neighboring switches with spare capacity. This paper introduces a new algorithm to efficiently perform flow delegation based on a novel delegation template abstraction and multi-period multi-objective optimization. Different from existing work, our approach can include estimated knowledge about future network situations and deal with different optimization criteria such as link and control overhead. We discuss the problem decomposition for the new algorithm and introduce an efficient two-step heuristic. Results show, that our approach performs significantly better than the simple greedy algorithm used in earlier work and is capable of handling flow delegation for networks with hundreds of switches.

This paper deals with synchronized multicast reception while considering Service Function Chaining (SFC), i.e., all data packets sent towards a multicast group are first processed by the required service functions and then received by the group members at approximately the same time. This is an important requirement for several distributed applications, e.g., emergency push-to-talk services like TETRA or DMR. The presented TopoSync-SFC approach exploits recent advancements in the fields of Software-defined Networking (SDN) and Network Function Virtualization (NFV) to address synchronization directly inside the network, i.e., no dedicated support for synchronization is required by the applications. It is based on the idea of application-specific SFC-embedding multicast distribution trees that exploit paths of different length in the topology to reduce the variance in packet reception times. We present the architecture for TopoSync-SFC and an ILP-based problem formulation that respects synchronization requirements. The evaluation of a push-to-talk use-case shows that TopoSync-SFC is a significant improvement with respect to synchronized reception over approaches that do not consider synchronization.

Steam boilers are widely used in industrial applications ranging from air conditioning to power plants. Advanced control schemes for such boilers require the knowledge of internal state variables, which are not always measurable. This paper proposes a new observer for steam boilers whose construction builds on a special state variable choice and a Lyapunov-based design of the observer gains. Tuning insight is gained from an oscillator-like structure of the (linearised) observer error dynamics. Comparisons to an extended Kalman filter in simulations and on experimental data from a small-scale automotive application demonstrate the performance of the proposed approach.

Despite its significance, only few mathematical models for backlash and gear play have been published. This contribution discusses the most notable modeling approaches in current literature, and employs them to simulate the drive line of a test bed that was intentionally equipped with a pre-designed backlash spot. A comparison with measurements shows that the established models suffer from similar drawbacks. Based on these findings, a new concept that includes damping in situations without classical backlash contact is introduced. It is shown that in the given setting the new approach significantly improves the quality of the simulations. All considerations are validated using experimental data.

Procedures for measuring the emissions of automotive vehicles typically include a speed trace that the driver has to track within prescribed tolerances. For development purposes, following this trace by means of automatic control is desirable in order to minimize costs. In this contribution, an iterative learning scheme is proposed that iteratively improves a feed-forward control signal. This is done by means of an optimization problem that takes the speed tolerances into account in the form of constraints. Experimental results obtained with a vehicle on a Road-to-Rig (R2R) test bed for a part of the Worldwide Harmonized Light Vehicle Test Procedure (WLTP) are presented and compared to results of a pure PI control scheme. After very few iterations, both tolerance violations and sudden changes of the pedal position are eliminated, yielding a significantly improved driving behavior.

This paper presents GCMI, a novel API for generic control message interception that can be used to programmatically access control messages exchanged between controllers and switches in a software-defined network. The approach allows it to program GCMI apps that register for specific control messages, e.g., all monitoring replies or flow table modification requests for a certain subnet. If such a control message is sent, it is redirected to the GCMI app where it may (or may not) be altered prior to forwarding to the original destination. This intermediate processing is widely used in the SDN domain, e.g., for transparent optimizations, verification or debugging purposes. In practice, however, the same interception functionality is re implemented over and over again because there currently is no generic solution to access control messages. In this paper, we show that there is a strong demand for GCMI, present a proxy-based OpenFlow prototype and demonstrate the features and benefits of our approach by re-implementing different use cases that rely on control message interception.

Software Defined Networking aims to separate network control and data plane by moving the control logic from network elements into a logically-centralized controller. Using a well-defined, unified control-channel protocol, such as OpenFlow, the controller is able to configure the forwarding behavior of data plane devices. Here, the OpenFlow protocol is translated to vendor- and device-specific instructions that, for instance, manipulate the flow table entries of a switch. In practice, SDN-enabled switches often feature different hardware capabilities and configurations with respect to the number of flow tables, their implementation, and which kind of data plane features they support. This leads to device heterogeneity within the SDN landscape, thereby obstructing the increased scalability and flexibility promised by the SDN paradigm. To overcome this challenge we propose TableVisor, a transparent proxy-layer for the SDN control channel that enables the flexible abstraction of heterogeneous data plane devices into a single emulated data plane switch. In this paper, we extend our previous work by introducing features to integrate modern P4 devices into an existing SDN environment and perform a detailed performance evaluation to quantify the overhead induced by our approach.

Hierarchical Heavy Hitters (HHHs) identify frequent items in streaming data. Finding these items has several applications to network monitoring, particularly in distributed denial-of-service (DDoS) mitigation and anomaly detection. Several algorithms are available to compute HHHs, each with different performance characteristics in terms of resource consumption, speed and accuracy. These characteristics determine which HHH algorithm may be best suited for a given network situation (e.g., because it offers sufficient accuracy for fine-grained traffic analysis). However, since the situation can evolve over time, the best choice for an HHH algorithm may also change. Simply replacing a chosen HHH algorithm has the drawback of losing all previously acquired monitoring information. This paper introduces the novel concept of HHH-transitions that transfer monitoring information between HHH variants and consequently allows it to adopt new performance characteristics by switching algorithms at runtime. For example, this enables a DDoS mitigation system to adapt to evolving network situations and therefore increase overall Return-on-Mitigation. We present explicit transition rules for common one-dimensional HHH variants and evaluate our approach based on real traffic from MAWILab. Results indicate that trade-offs between performance characteristics can be realized at runtime and that it is possible to increase overall post-transition accuracy by retaining monitoring information.

Torsional vibrations are a troubling issue in many powertrain applications. To suppress them, a hydrodynamic steel spring coupling with suitable damping characteristics is often incorporated into the affected driveline. Its behaviour is commonly described by the so-called Kelvin-Voigt model, which is valid only for a fixed excitation frequency. Such a model proves to be unsuitable for use in most transient simulations or for model based control purposes. Therefore, this paper presents a novel control-oriented model for the coupling's transient behaviour. Starting from physical first principles, a second order state-space model is derived. Its validity is verified using experimental data obtained on a complex combustion engine test bed.

This paper conducts a systematic analysis of different conceptual choices on implementing ICN core functionalities (routing, forwarding, caching) in Software-defined networks. The analysis considers recent developments in the area of Software-defined Networking and Network Functions Virtualization. Based on this, four abstract models of deployment options are presented and their requirements, advantages and disadvantages are discussed. The paper also presents a preliminary coarse-grained performance evaluation of the different options.

This paper presents an approach for developing transient strategies for multi-cylinder gas engines using only a single cylinder in a hardware-in-the-loop setup. Engine components that are physically not present are replaced by a real-time model. The model generates pressure and speed trajectories, which are applied to the single-cylinder engine to ensure that its behavior is identical to its operation as part of the multi-cylinder engine. Design and analysis of two of the controllers used for that purpose are discussed. For the control of the charge air pressure, a concept based on nonlinear decoupling and a strategy maximizing the actuator range available for disturbance rejection is presented. To control the fuel gas pressure, a linear plant model is obtained and a controller motivated by stability conditions is proposed.

This paper introduces SDN Cockpit, an easy-to-use and open ecosystem for teaching network softwarization based on mininet and the Ryu controller. The ecosystem allows candidates to gain hands-on-experience with SDN in prefabricated scenarios without having to deal with potentially complex details such as traffic generation. It provides useful tooling for instructors and automated evaluation for assignments. The paper discusses the design goals, the architecture and the workflow of the ecosystem. First experiments with SDN Cockpit show that the approach can improve the motivation and the learning experience of the candidates.

Software-Defined Networking (SDN) allows fine granular control by applications running in the network control plane, facilitating the management, orchestration, and deployment of network services. However, the diversity of application, protocols, and switches makes the task of developing applications for such networks very complex. Besides, such heterogeneity makes it harder to support the manifold requirements that may arise from different control plane applications and to verify if the underlying infrastructure satisfies the requirements from these applications. In this paper, we propose a two-phase solution for this problem, extending the Model-Driven Networking (MDN) framework for: i) enabling it to model infrastructure capabilities, so that we can verify if these capabilities could satisfy applications requirements; and ii) applying a flow delegation technique to leverage the set of network capabilities in order to support applications requirements. Our experiments demonstrate that our flow delegation mechanism not only improves the network compatibility but also achieves better bandwidth usage and jitter ratios ^22% lower when considering QoS requirements.

In this paper, we present TREX, a novel tenant-driven network traffic extraction platform for SDN-based cloud environments. With TREX, tenants can define templates that consist of extraction point, fine-grained network traffic filter and a target identifier representing a tenant-operated analysis application. All packets that match on the filter will be extracted in a lightweight fashion at a virtual switch and forwarded to the analysis application in near real time. Therefore, TREX enables analysis and debugging based on proprietary applications that are operated directly by the tenant. Furthermore, original network traffic remains unaffected and migration of extraction points to different physical cloud nodes, which is typical for cloud environments, will not lead to a gap in the extraction.

Reproducing research results, as it is required for peer review, can be a time-consuming and difficult task. In this work, we propose three approaches to improve the way of how research results can be substantiated and discuss their applicability. Our proposals are based on a brief study on evaluation methods (for SDN research) and insights from a comprehensive discussion on reproducibility.

Modern Software Defined Networking (SDN) applications rely on sophisticated packet processing. However, there is a mismatch between control plane requirements and data plane capabilities caused by increasing hardware heterogeneity. To overcome this challenge, we propose TableVisor, a proxy-layer for the OpenFlow control channel that enables the flexible and scalable abstraction of multiple physical devices into one emulated data plane switch that meets the requirements of the control plane application. TableVisor registers with the SDN controller as a single switch with use-case specific capabilities. It translates the instructions and rules from the control application towards the appropriate physical device where they are executed. In this paper, we present the updated architecture and functionality of TableVisor as well as first evaluation results based on testbed experiments.

Flow delegation can achieve scalability in a software-defined environment where switches experience heterogeneous utilization of flow tables. Following this approach, so called delegation relationships are used to distribute the load between neighboring switches in order to cope with local capacity bottlenecks. This paper presents the Flow Delegation Algorithm (FDA) that calculates delegation relationships in such a way, that existing spare capacity of the network is efficiently utilized while simultaneously taking into account flow table capacity constraints, link utilization, the amount of required control operations and the current and estimated future traffic situation. Simulation results show, that the proposed algorithm outperforms the straw-man approach presented in earlier work for a range of different efficiency metrics.

While many DDoS mitigation approaches utilize the flexibility of software-defined infrastructures in a decentralized fashion, most of them assume that the infrastructure as a whole is willing and capable of mitigating all incoming packets of an attack. Those approaches cannot be used out of the box, if the attack overburdens the infrastructure or the monetary budget for mitigation is limited. Therefore we present EarlyDrop, a trade-off driven DDoS defense mechanism based on transparent blackbox monitoring. With EarlyDrop, operators can choose to drop undifferentiated traffic before it is forwarded to the mitigation system in order to reduce the load onto the infrastructure - a trade-off between mitigation cost (resources, money) and return of mitigation.