Siegfried Mercelis

Context-Aware Optimization of Distributed Resources in Internet of Things Using Key Performance Indicators

The recent advancements in Internet of Things (IoT) show us a glimpse of a future in which all our devices are connected to the internet, providing users with services that make life easier, more comfortable and safer. Although this interconnectivity seems simple, in practice management of the IoT hardware and the enormous amounts of data it generates is challenging. To bring the connected future into reality and build advanced and useful services, better resource usage estimation (memory, bandwidth, storage etc.) and resource management is required. We propose a IoT optimization methodology, where resources are estimated at each level of the IoT architecture (i.e. nodes, edges and cloud). Using these estimates, the executed code is redistributed across the network in order to optimize the cost and efficiency of the IoT environment, while taking into a specific context (e.g. environment). Initially, we aim to apply this methodology for statically defined contexts. In our future research we aim to perform the optimization at runtime, redistributing tasks across the IoT network dynamically as the context of the nodes changes.

Towards a distributed real-time hybrid simulator for autonomous vehicles

To thoroughly test and validate algorithms and systems of autonomous vehicles, a large number of vehicles, many tests and a multitude of datasets are needed. This way of developing and testing is difficult, expensive and sometimes even dangerous. To combine the benefits of real world testing with the scalability and lower cost of simulation based testing, we present a novel methodology for a real-time hybrid simulator that is capable of handling real and simulated vehicles simultaneously with full interaction, in real time. We validated our methodology by assessing its overall performance and real-time capabilities using an F1/10 scale vehicle. The results effectively show the viability of this approach for validation of autonomous vehicles in a cost-efficient and safe manner.

Testing IoT systems using a hybrid simulation based testing approach

This paper presents an extensive overview of the challenges that arise when testing large IoT applications at the system level. In order do that we start from analyzing behavior of local entities such as IoT devices or people interacting with the IoT system. The interactions of these local entities eventually leads to an emergent behavior. Both the emergent behavior and the local behavior need to be taken into account when testing IoT systems. Therefore, we present a novel hybrid simulation based testing approach that is able to effectively facilitate interactions of these local entities. Furthermore, we introduce various solutions to the challenges that arise when implementing this hybrid methodology. These challenges are mainly related to the IoT development pipeline, synchronization between real-life and simulation environment and the scalability constraints of modern simulation techniques.

Towards a Scalable Distributed Real-Time Hybrid Simulator for Autonomous Vehicles

The rising popularity of autonomous cars is asking for a safe testbed, but real-world testing is costly and dangerous while simulation-based testing is too abstract. Therefore, a hybrid simulator is needed in which a real car can interact with many simulated cars. Such a simulator already exists, but is far from scalable due to a centralised architecture, thus not deployable on many vehicles. Therefore, this paper presents a more distributed and scalable architecture that solves this problem. We assessed the overall performance and scalability of the new system by conducting four experiments using a 1/10th scale car. The results show that this new distributed architecture outperforms the previous approach in terms of overall performance and scalability, thus paving the way to a safe, cost-efficient and hyper scalable testing environment.

Introduction of Deep Neural Network in Hybrid WCET Analysis

Safe and responsive hard real-time systems require the Worst-Case Execution Time (WCET) to determine the schedulability of each software task. Not meeting planned deadlines could result in fatal consequences. During development, system designers have to make decisions without any insight in the WCET of the tasks. Early WCET estimates will help us to perform design space exploration of feasible hardware and thus lowering the overall development costs. This paper proposes to extend the hybrid WCET analysis with deep learning models to support early predictions. Two models are created in TensorFlow to be compatible with our COBRA framework. The framework provides datasets based on hybrid blocks to train each model. The feed-forward neural network has a high convergence rate and is able to learn a trend in the features. However, the error of the models are currently too large to predict meaningful upper bounds. To conclude, we summarise the problems we need to tackle to improve the accuracy and convergence issues.

Challenges of Modeling and Simulating Internet of Things Systems

With the rise of complex Internet of Things systems we see an increasing need for testing and evaluating these systems. Especially, when we expect emergent complex adaptive behavior to arise. Agent based simulation is an often used technique to do this. However, the effectiveness of a simulation depends on the quality of individual models. In this work we look in depth what the characteristics are of Internet of Things devices, actors and environments. We look at how these characteristics can be used to find appropriate, performance optimized modeling techniques and formalisms. During the course of this work we will extensively refer to a custom-developed Internet of Things simulation framework and to relevant related literature.

Context-Aware Distribution In Constrained IoT Environments

The increased adoption of the IoT paradigm requires us to take a good look at the network weight it creates. As adoption increases, so does the network load and server cost, causing a jump in required expenses. A solution for this is Fog Computing, where we distribute the cloud load over the network devices so that the tasks get pre-processed before reaching the cloud level, or might not even have to reach the cloud level. To aid with this research, we wrote a simulator that calculates the optimal spread of the application over the network devices, and shows us how this spread will occur. This spread will be based on context, where for example processor-bound machines get smaller tasks and energy-bound machines get energy-efficient tasks. We use this simulator to compare algorithms used for placing the application.

Distributed Uniform Streaming Framework: Towards an Elastic Fog Computing Platform for Event Stream Processing

More and more devices are connected to the internet. These devices could be used to help execute applications that otherwise would need to be executed on the cloud or on a system with more computational resources. To execute the application on multiple devices, we will split it up into multiple application components that stream events to each other. In this paper we present a framework that allows application components to stream events to each other. On top of this we present a coordinator system to move application components to other devices. This elasticity allows the coordinators to run application components on different devices based on the context, in order to optimize resources such as network usage, response times and battery life. The coordinators use an adapted version of the Contract Net Protocol which allows them to find a local minima in resource consumption. In order to verify this, three use cases are implemented.

Towards Real-time Smart Road Construction: Efficient Process Management through the Implementation of Internet of Things

The introduction of the internet had a significant impact on peoples lives. The next generation internet, also called the Internet of Things (IoT), takes automation to a whole new level. Currently, IoT is still in its infancy. Still, IoT architectures are being implemented for specific applications and/or use-cases. However, as the number of applications and IoT frameworks increases, so does the number of interfaces, protocols and standards. As a result, interoperability between different platforms has become significantly more complex. In road construction and logistical processes, analysis of collected data and real-time communication between the various actors (e.g. paver, contractor, transporter) allows them to operate their machines more effectively, which significantly improves road quality. However, the significant differences in communication protocols and data storage prevent a straightforward interconnection and interoperability between the various applications involved. This paper introduces a middleware that allows on-demand communication, authentication and information exchange in such heterogeneous environments. We demonstrate and validate this middleware with a use-case and field test in road construction in the context of the RoadIT Technology Transformation (TETRA) project.

Cost and Energy Efficient Indoor and Outdoor Localization of Rail Cars in a Confined Maintenance Site

The Belgian national railway company (NMBS) is interested in localization of rail cars in their maintenance sites, which consist of both indoor and outdoor environments. Several localization technologies and methods are currently available, each with its benefits and drawbacks regarding accuracy, range, cost, power consumption, and environmental characteristics. After examining several localization systems with regards to application requirements, power and cost constraints, we found passive RFID to be the most suitable and developed a proof of concept for a real workshop environment.