Seamless Model-Based Design and Deployment of Wireless Networked Systems Tobias Schwalb, Tobias G¨ adeke, Johannes Schmid and Klaus D. M¨ uller-Glaser Karlsruhe Institute of Technology, Institute for Information Processing Technologies Engesserstr. 5, 76131 Karlsruhe, Germany Email: {tobias.schwalb, tobias.gaedeke, johannes.schmid, klaus.mueller-glaser}@kit.edu Abstract—To allow for rapid abstract development of algo- rithms, model-based tools are often used nowadays. These tools support design using predefined blocks, adaptation using param- eters, simulation in a virtual environment and also generation of code for programming embedded devices. However, there is usually no direct connection between the embedded system and the model for programming and runtime control and monitoring. In this paper, we present a seamless tool chain for model- based algorithm development, simulation and integration on an embedded system as well as runtime control and monitoring from model level. Based on our concept, the development of algorithms can largely be separated from the execution platform and its control and monitoring possibilities. Yet, it is possible to feedback monitored data to support the simulation. The concept is integrated and evaluated and its applicability for the development of an algorithm is demonstrated by means of an example from the field of wireless senor networks localization. We conclude that the outlined approach offers a simple opportunity to reduce development overhead. I. I NTRODUCTION The design of complex systems becomes more and more challenging, especially for distributed and networked embed- ded platforms. As an exemplary application we consider a wireless sensor network (WSN) for indoor localization. Such a WSN consists of a number of nodes equipped with a microcon- troller unit (MCU), a wireless transceiver and different sensors. For localization, reference nodes with known positions are deployed in the area of interest. Mobile nodes, carried by persons or attached to equipment, are localized based on distance estimation to reference nodes within range. For the localization and tracking problem a vast number of approaches and algorithms with different performance and complexity measures have been proposed. To manage the complexity in the development of new algorithms, a model-based rapid prototyping approach is often considered based on simulated input data. To account for the demands in a specific application scenario real-world input data are needed for optimization and further development. To avoid reimplementation overhead automatic code-generation has been adopted in many applica- tion domains. There is typically a straightforward development path from the initial algorithm to a working system under real-world conditions, given that real-world data are present. However, the link from the real deployment back to the model is still missing in most tool chains (cf. Figure 1). This complicates system development because the system designer has to acquire real-world data manually. The user needs a deep ? Real World On Off Switch Model Figure 1. Motivation for Seamless Tool-Chain for Networked Systems. understanding of the mechanisms for control and monitoring of each used platform, which is time-consuming and thus costly. In this paper we present a concept for a seamless tool chain for model-based algorithm development and simulation as well as model-based runtime control and monitoring of a deployed system. This reduces the costly overhead by acquisition of real-world data from a distributed wireless system into the model to close the missing design loop. In extension to previ- ous work we integrate a model-based description of platforms and parameterizable recording setups [1]. In addition, we implement a loop to the algorithm development to allow white- box access with abstract parameters. As a result, it becomes possible to configure the recording according to the algorithm on model level during the integration phase. We close the gap by control and monitoring on the target platform for further algorithm optimizations. The recorded data can also be feedback into the simulation process for repeatable scenarios and comparisons of algorithms. The paper focuses on the concepts and tool integration aspects for rapid prototyping of WSN following a model-based design process. We first describe the state of the art in model- based methods for algorithm design, control and monitoring on embedded systems in Section II. Section III outlines an overview of the proposed concept and describes the flow using the integrated tool chain. Section IV illustrates the differences between simulation and real-world scenarios and points out the advantages of seamless integration. The developed tool chain for integration of a distributed system is described in Section V, while Section VI concentrates on runtime control and monitoring. Tests and results concerning the localization algorithm within the wireless sensor network are presented in Section VII. We close with conclusions and an outlook on future work in Section VIII. 978-1-4673-2789-3/12/$31.00 c 2012 IEEE 100