François Christophe

A methodology supporting syntactic, lexical and semantic clarification of requirements in systems engineering

Product development is a challenging activity. The process begins with a description and representation of a design problem in form of a requirements document. It involves two phases: elicitation by description in Natural Language (NL) and clarification of the description. NL implies interpretation of terms within a context to avoid later misunderstanding. The paper proposes a methodology to elicit and refine the initial needs. The elicitation is done by finding support information from several sources such as patent databases, encyclopaedias and commercial websites. The refinement supported by a computer-based approach is done on different levels (grammar, words and context selection) to reduce the ambiguity of the requirements descriptions. The initial description is refined by an automatic questioning process. This is followed by an assisted search and selection of answers from different web-based sources. Relevant answers are selected using a similarity metric. A case study is used to demonstrate the approach.

Survey and evaluation of neural computation models for bio-integrated systems

Abstract Integrating neurobiological cultures with computer systems presents an opportunity to enhance computational energy efficiency. These Bio-Integrated Systems (BISs) require knowledge about structure and behavior of neural components and their interfacing. In the early design phases, modeling neurons offers cost, failure-free and retrial benefits compared to laboratory grown neural networks. The usefulness of these models lays in characteristics of being realistic but also computationally efficient. This survey reviews computational models of spiking neurons and their changes in connections, known as plasticity. The review studies models that are faithful to real neural cultures, and are computational efficient for real-time BISs. Also, criteria and methods for comparing models with ‘in-vitro’ experiments are reviewed to conclude on the level of realism of models in comparison with biological setups. Izhikevich’s model of spiking neurons is recommended due to its accuracy in reproducing real neural firing patterns, computational efficiency, and ease of parameter adjustment. The model of Spike-timing dependent plasticity is recommended as current basis for representing neuron changes in connections. For the analysis of network connectivity and connectivity changes in BIS, the Cox method is recommended because it evaluates connections based on activities from all recorded neurons as opposed to pair-wise approaches.

Impact analysis of graph-based requirements models using PageRank algorithm

Managing requirements changes of complex systems and the potential impact of such changes represents a big issue for companies. Currently, commercial modelers propose tools for analyzing the direct impact of requirements changes on system design or code but the analysis of requirement change on other requirements remains seldom studied. This paper proposes an approach for the impact analysis of changes in requirements combined with a ranking of importance of requirements in graph based requirements network. Warshall algorithm is used in this paper for performing the impact analysis. Along with this approach, PageRank algorithm is used for ranking requirements according to their importance. Requirements hierarchy and their textual description of importance are considered as input for calculating their impact as well as their importance within the network of requirements. This combination of Warshall and PageRank algorithms provide significant results for helping designers in decision-making process of modifying requirements for future design versions.

Systematic Search for Design Contradictions in Systems’ Architecture: Toward a Computer Aided Analysis

Time pressure imposed to the engineering design process is one fundamental constraint pushing engineers to rush into known solutions, to avoid analysing properly the environment of a design problem, to avoid modelling design problems and to take decision based on isolated evidences. Early phases in particular have to be kept short despite the large impact of decisions taken at this stage. Significant efforts are currently spent within different engineering communities to develop a model-based design approach adapted to conceptual stages. Developing such type of models is also challenging due to the fuzziness of the information and due to the complexity of the concepts and processes manipulated at this stage. Currently few support tools are really capable of really supporting an analysis of the early design concepts and architectures. Simultaneously the approach should be fast, easy to use and should provide a real added-value to efficiently support the decision and the design process. The present article is presenting a framework based on a progressive transformation of the design concepts. The final material generated by this transformation process is an oriented graph with different types of classified variables. This graph can be processed as described in the article to automatically exhibit the conflicts or contradictions present in the design concept architecture. The article is proposing two main contributions which are a real move toward model development at conceptual stage and the possibility to process those models to detect solution weaknesses. The discussion is presenting further developments and possibilities associated with this method.

A Combined Design Structure Matrix (DSM) and Discrete Differential Evolution (DDE) Approach for Scheduling and Organizing System Development Tasks Modelled using SysML

During a system engineering process there are an important number of tasks that need to be organized, mapped together and recursively considered. The tasks that are mapped together are exchanging different flows of information and material. In this type of iterative processes, significant savings in term of development time can be made by providing a method that is optimizing the amount of feedbacks and iterations to the minimal level simply required for the successful development of the system. Task scheduling in a system engineering process can become extremely complex. Nevertheless it is a crucial step of the early stages of the systems engineering process for time-to-market, cost-efficiency and quality reasons. In this article, the authors are proposing to combine a computational approach (Discrete Differential Evolution) with Model Based Systems Engineering (MBSE) for minimizing iterations and reducing lead-time development. The present article is contributing to recent research works using Design Structure Matrixes (DSM) and computational methods for visualizing and analyzing systems engineering processes. The paper is proposing a framework integrating a model-based approach and a DSM based analysis of the process architecture to assist system engineers in organizing and scheduling tasks. As a result, this framework allows engineers to automatically populate DSMs generated from MBSE models developed in SysML. A specific stereotype is proposed to represent system development tasks in SysML. The sequencing of the engineering tasks is optimized with the application of a Discrete Differential Evolution algorithm (DDE) taking into account the different constraints. The practical use of the proposed framework is demonstrated on the case study of a mobile robot developed for the Eurobot competition. The article also discusses the possibility to use the current framework to analyze the impact of requirement changes on the scheduling of development tasks.

Building wireless sensor networks with biological cultures: components and integration challenges

Abstract The development of wireless sensor networks (WSNs) struggles with limited computing, communication, and energy resources. Bio-Integrated Systems (BISs) contain cultured cells that perform certain tasks. For instance, neurons can work as brains for robots. BISs could provide solutions to overcome WSNs resource constraints. Biological entities integrating in computing hardware gains interest especially in robotics. This interest is due to the potentials of such integration in deep learning capacity, massive parallelism, energy savings, and communication and integration with living subjects. This paper collects existing BISs research and provides research motivations for Bio-Integrated WSNs (BI-WSNs). BI-WSNs solutions for improving energy preserving, sensing, processing, and communication are proposed and supported with existing examples. Further, on-going research on integrating neural networks in WSNs is presented. Challenges related to protection of biological entities from external environment are discussed. Finally, a prospective model of BI-WSNs consisting in optogenetic communication combined with neural network processing is given. Graphical abstract This graphical abstract represents the different components of a node of a sensor networks. This paper studies the possibilities to replace each of these components with a biological counterpart enabling similar functionalities. The benefits and challenges for the development of such biological sensors, named Bio-Integrated Wireless Sensor Network (BI-WSN), are discussed in this study.

Data correction for seven activity trackers based on regression models.

Using an activity tracker for measuring activity-related parameters, e.g. steps and energy expenditure (EE), can be very helpful in assisting a persons fitness improvement. Unlike the measuring of number of steps, an accurate EE estimation requires additional personal information as well as accurate velocity of movement, which is hard to achieve due to inaccuracy of sensors. In this paper, we have evaluated regression-based models to improve the precision for both steps and EE estimation. For this purpose, data of seven activity trackers and two reference devices was collected from 20 young adult volunteers wearing all devices at once in three different tests, namely 60-minute office work, 6-hour overall activity and 60-minute walking. Reference data is used to create regression models for each device and relative percentage errors of adjusted values are then statistically compared to that of original values. The effectiveness of regression models are determined based on the result of a statistical test. During a walking period, EE measurement was improved in all devices. The step measurement was also improved in five of them. The results show that improvement of EE estimation is possible only with low-cost implementation of fitting model over the collected data e.g. in the app or in corresponding service back-end.

Graph Based Representation and Analyses for Conceptual Stages

What is the fundamental similarity between investing in stock of a company, because you like the products of this company, and selecting a design concept, because you have been impressed by the esthetic quality of the presentation made by the team developing the concept?Except that both decisions are based on a surface analysis of the situations, they both reflect a fundamental human’s cognitive feature. Human brain is profoundly trying to minimize the efforts required to solve a cognitive task and is using when possible an automatic mode relying on recognition, memory, and causality. This mode is even used in some occasion without the engineer being conscious of it. Such type of tendencies are naturally pushing engineers to rush into known solutions, to avoid analyzing the context of a design problem, to avoid modelling design problems and to take decision based on isolated evidences. Those behaviors are familiar to experience teachers and engineers. This tendency is magnified by the time pressure imposed to the engineering design process. Early phases in particular have to be kept short despite the large impact of decisions taken at this stage. Few support tools are capable of supporting a deep analysis of the early design conditions and problems regarding the fuzziness and complexity of the early stage. The present article is hypothesizing that the natural ability of humans to deal with cause-effects relations push toward the massive usage of causal graphs analysis during the design process and specifically during the early phases. A global framework based on graphs is presented in this paper to efficiently support the early stages. The approach used to generate graphs, to analyze them and to support creativity based on the analysis is forming the central contribution of this paper.Copyright

Evaluation of Visual Tracking Algorithms for Embedded Devices

Today’s embedded platforms enable executing difficult tasks such as visual tracking. However, such resource-constrained systems are still facing challenges regarding the performance and accuracy in executing these tasks. This paper presents the evaluation of 5 open-source visual tracking implementations available from the contributions branch of the Open Computer Vision (OpenCV) library. This evaluation is performed based on the performance and accuracy of these implementations when embedded in a Raspberry Pi. The algorithms evaluated are On-Line Boosting, Multiple Instance Learning (MIL), Median Flow, Tracking-Learning-Detection (TLD), and Kernelized Correlation Filters (KCF). Even if commercial implementations of these algorithms perform better than their open-source version, the popularity of OpenCV motivates this evaluation. Tests are based on a benchmark of 100 video streams from which the tracking implementations should follow moving objects. The algorithms are evaluated for accuracy using averaged Jaccard indices and for performance by measuring their frame rate. We want to find an open-source implementation that performs well on these two criteria when tested on an embedded platform. Results show Median Flow being the fastest but its accuracy is the lowest. We therefore recommend KCF as it is the second fastest and the most accurate.

On electrophysiological signal complexity during biological neuronal network development and maturation

Developing neuronal populations are assumed to increase their synaptic interactions and generate synchronized activity, such as bursting, during maturation. These effects may arise from increasing interactions of neuronal populations and increasing simultaneous intra-population activity in developing networks. In this paper, we investigated the neuronal network activity and its complexity by means of self-similarity during neuronal network development.