Rossby Centre: A World Leading Actor in Climate Research Using MATLAB

Rossby Centre is a climate research group at SMHI in Norrköping that provides Swedish authorities, researchers, and companies with supporting material for questions related to climate change. Rossby Centre is a world leading actor in climate research. The group develops numerical models for climate simulations and performs advanced analysis of model output data. Its climate simulations, performed on Linux clusters at NSC in Linköping, are used worldwide in numerous projects. This session will include examples on climate change hot topics Rossby Centre is involved with and will show how the group uses MATLAB as an important tool for analysis.

How MATLAB Is Addressing Key Technology Trends: Parallel and Multicore Computing, Embedded Software Development, Object-Oriented Programming, and the Need for Optimal Design

In this session, we’ll look at how MATLAB is addressing some of the key technology trends introduced in Jim Tung’s morning keynote address. We will explore recent capabilities introduced to the MATLAB product family that enable you to more easily leverage multicore computers and computing clusters, rapidly develop embedded algorithms directly in MATLAB, and develop optimal designs for larger and larger systems. The session will also cover the object-oriented programming capabilities introduced in MATLAB R2008a, which support the development of large-scale applications.

The World of System Identification

System identification is the art and science of building mathematical models of dynamic systems from observed input-output data. It can be seen as the interface between the real world of applications and the mathematical world of control theory and model abstractions. As such it is a ubiquitous necessity for successful applications.

System identification is a very large topic, with different techniques that depend on the character of the models to be estimated, such as linear, non-linear, hybrid, nonparametric, and so forth. At the same time, the area can be characterized by a small number of leading principles, for example, looking for sustainable descriptions by proper tradeoffs in the triangle of model complexity, information contents in the data, and effective validation. The area has many facets, and there are many approaches and methods. This session will both give an overview of the "science" side (basic principles and results) and illustrate the practical, "art" side of how to approach a real problem.

Introduction to Object-Oriented Programming in MATLAB

The object-oriented programming capabilities in MATLAB® enable you to develop and maintain large applications and data structures more easily. Using engineering examples, this master class will demonstrate how you can define classes and work with objects, highlighting the benefits of this programming approach over traditional procedural techniques. Features covered include class definitions, properties, property attributes, methods, method attributes, and inheritance. No knowledge of object-oriented programming is required.

Power Tips for Using Simulink on Large Projects

In this class, we will examine industry best practices for major capabilities introduced in recent years that support using Simulink® for large-scale applications. We’ll emphasize practical examples and industry workflows. Key topics include using model referencing and buses with component-based modeling techniques; performing version control and configuration management; applying model patterns that yield clear designs and generate efficient code; managing large data sets associated with your model and code; running simulations as efficiently as possible; and adapting Simulink to your environment, including customizing user interfaces and automating model guideline checks.

Designing Embeddable Algorithms and Automatically Generating C Code with MATLAB

In this workshop, we will showcase new capabilities of MathWorks products that enable you to generate C code from your Embedded MATLAB™ code. You will learn about these capabilities by going through an example for the design of a video processing system. Through demonstrations, you will learn how to create and modify your MATLAB algorithms to be compliant with the Embedded MATLAB language subset, generate C code from your Embedded MATLAB code directly from MATLAB desktop, and call your Embedded MATLAB code as a new block within Simulink to integrate and simulate your algorithm as part of a larger system model.

Parallel Computing with MATLAB

This master class will show you how the new products and features for MATLAB enable you to take advantage of recent advances in computer hardware, from multiprocessor machines to computer clusters. You will learn how to utilize multiple cores in your desktop machine through the new parallelism capabilities of MATLAB and Parallel Computing Toolbox™. We will also introduce the use of MATLAB Distributed Computing Server™ on a computer cluster to speed up your algorithms and handle larger data sets.

Advanced Programming Techniques in MATLAB

This master class covers two important MATLAB topics: handling memory efficiently and choosing among the rich set of function types. Participants will gain an understanding of how different MATLAB data types are stored in memory and how to program in MATLAB to use memory efficiently. Recent versions of MATLAB have introduced several new programming concepts, including new function types; this class illustrates and explores the usage and benefits of the various function types under the different conditions. You will learn how using the right function type can lead to more robust and maintainable code. Demonstrations will show how to apply these techniques to problems that arise in typical applications.

Simulation of MIMO Antenna Systems in Simulink

M. Viberg, L. Pettersson, T. Boman, U. Carlberg, E. Arabi, S. Ali, O. Moussa, and M. Bilal

Multi-input/multi-output (MIMO) has emerged as a hot topic in wireless communications during the last decade. This is due to possible dramatic increases in reliability and capacity as compared with single-antenna solutions. However, much of the existing theoretical results are based on very simplistic models of the antennas and transceiver circuitry. Within the Strategic Research Center on Antenna Systems Charmant at Chalmers, we are developing a systems simulator for making more realistic studies of MIMO systems. Models of different complexity can be used for the different components. For the linear components, S-parameters are used as the interface. These can come from theoretical models, electromagnetic field simulations, or directly from measurements. For the nonlinear components, simple memoryless models can be used, as existing in RF Toolbox™. However, wideband applications demand more elaborate models. We have implemented a so-called memory polynomial model, which has been found to match measurements of nonlinear wideband power amplifiers well. Since RF Toolbox cannot handle multiport scattering matrices (yet), the MIMO part is here implemented in Embedded MATLAB™ code. The paper describes an implementation of a transceiver chain including mixers, filters, PA/LNA, matching network, MIMO antennas, and channel models. The simulator can be used for a wide range of wireless communication applications.

Simulator for Optimization of Radio-Frequency Pulses in Magnetic Resonance Imaging

Johanna Öberg, Ph.D. Student, Karolinska Institutet, Stockholm, Sweden
Rouslan Sitnikov, Researcher, Karolinska Institutet, Stockholm, Sweden

The surface coils being driven by radio-frequency pulses are often a primary choice in magnetic resonance imaging for both radio-frequency excitation and reception of weak magnetic resonance imaging signals from the volume of interest. The advantages of the surface coils compared with the volume coils are the minimized radio-frequency power deposition in tissue and the improved detection sensitivity due to its close proximity to the volume of interest. However, a surface coil has a disadvantage of producing a highly inhomogeneous electromagnetic field (B1), which makes it difficult to use for complex multi-pulse magnetic resonance experiments. The adiabatic radio-frequency pulses, consisting of complex amplitude and phase functions, help solve this problem if properly applied. The adiabatic condition needs to be satisfied for every radio-frequency offset and every point in space. A simulation tool is presented that optimizes electromagnetic fields generated by such radio-frequency pulses in frequency, space, and time domains. The obtained electromagnetic excitation profile is experimentally verified on water samples.

Using MATLAB to Aid the Implementation of a Fast RSA Processor on a Xilinx FPGA

Carsten Siggaard, Senior Consultant, Danish Technological Institute, 8000 Århus Denmark

In cryptographic applications such as multiplications modulo, a large number is the core of algorithms such as RSA and El-Gamal. These operations are expensive, making even a Pentium IV unable to perform more than a few thousand cryptographic operations per second. Many algorithmic optimizations have been proposed, for example, by Montgomery and Barrett; however, the operations are still very expensive. The expensive operations cause the need for large server farms just to be able to handle the key exchange in large Web applications. ASIC implementations exist that are very efficient, but ASICs suffer from the lack of flexibility, which is the hallmark of many FPGAs. We present a method to map the multiplications into a Xilinx FPGA, creating a huge speedup. The advantage is that it is possible to upgrade the FPGA, for example, if key sizes have to be increased or the algorithm must be improved. The modeling is done by means of MATLAB® and Simulink®, and the code generation is done by Simulink HDL Coder™.

Pump Analyser: A Versatile Tool for Configuration Analysis of Asena Infusion Pumps

Nasser Hosseini, Ph.D., Sahlgrenska University Hospital, Göteborg, Sweden
Jessica Ylvén, M.Sc., Sahlgrenska University Hospital, Göteborg, Sweden
Christer Blomster, B.Sc., Sahlgrenska University Hospital, Göteborg, Sweden
Anna U Bengtsson, B.Sc., Sahlgrenska University Hospital, Göteborg, Sweden
Wiebke Rosenberg, B.Sc., Sahlgrenska University Hospital, Göteborg, Sweden

Infusion pumps have widespread use in healthcare, and each hospital has hundreds of them in operation. New models are more advanced than earlier ones. An Asena infusion pump can be configured using about 700 parameters, controlling the way the pump is working and presenting a drug list including flow rate limits, drug concentrations, and more. Infusion pumps need periodic preventive maintenance, often once a year. Pumps also need repair or a software update. Manually setting all parameters is very time-consuming, and a mistake could mean a high risk for the patients. Thus, there is an obvious need for a tool that can read the settings from a reference pump to a computer and save them as a reference file. The reference file is then used to check the settings of other pumps by reading their settings and comparing the settings file with the reference file. The MATLAB programming environment was used to develop Pump Analyser, a tool with a graphical user interface (GUI). The data communication between the PC and the infusion pump occurs via a serial RS232 or Infrared Data Association (IrDA) point-to-point type. Using this tool, general options, permitted syringes, and the drug library are copied from the infusion pump to the computer. The result data is saved in binary (MAT) format, which can be compared with the data from earlier results and possible differences can be detected. The use of Pump Analyser can be of advantage to patients, nurses, and medical engineers.

MATLAB/Simulink Implementation of Phi-Transforms – A New Toolbox Only or the Rival of Wavelet Toolbox for the Next Decade?

Peteris Misans, Professor, Riga Technical University, Riga, Latvia
Maris Terauds, Assistant, Riga Technical University, Riga, Latvia
Arturs Aboltins, Doctoral Student, Riga Technical University, Riga, Latvia
Gatis Valters, Doctoral Student, Riga Technical University, Riga, Latvia

The paper gives an introductory description of novel classes of fast orthogonal transforms (Phi-Transforms), which are defined, described, and exploited by the team of authors during the recent three years. The main goal of the paper is the presentation of open collection of MATLAB functions and the library of Simulink blocks (Phi-Transforms Toolbox and Blockset). Currently, the toolbox contains almost 50 units and supplements existing Signal Processing Blockset™, Signal Processing Toolbox™, Communications Blockset™, Communications Toolbox™, Image Processing Toolbox™, and other toolboxes and blocksets for MATLAB and Simulink. The categories of the units include basis function generators, spectrum analyzers, signal synthesizers, parametrical orthogonal filters, parametrical communication engines, GUI-based interactive tools, demo tools, and others. The units are developed for both the 1-D and 2-D case. The preliminary comparison with orthogonal wavelets is provided. Advantages of Phi-Transforms are demonstrated by simple examples from the signal synthesis, compression, and filtering. The authors treat the toolbox as a rich base for teaching students and the development of novel DSP algorithms and devices.

Image Analysis Methods for Determining HER2/neu Status of Breast Cancers

Michele Avoni, BME MSc., DTU – Technical University of Denmark, Copenhagen, Denmark, in collaboration with Visiopharm A/S, Hørsholm, Denmark

Purpose: The purpose of this study is to investigate the reliability of a new unsupervised method to assess the human epidermal growth factor receptor (HER/2) oncogene protein overexpression of breast cancer tissue using the FDA-approved HercepTest and grading system (negative, 0, or 1+; weakly positive, 2+; strongly positive, 3+). In clinical practice, the percentage of overexpressing cells is usually estimated through human visual inspection of the stained specimen, and the result is subjective and/or inconsistent. To alleviate this problem, image analysis techniques are introduced to computerize cell analysis. We have developed a software tool using MATLAB to fully automatically assess HER2/neu status in breast cancer IHC specimens.

Experimental Design: HER/2 status in 30 patients with lymph node-positive breast cancer is investigated by conventional immunohistochemistry and by applying the HercepTest. The true HER/2 status of each subject is known and is the evaluation result of several skilled doctors from the Department of Molecular and Quantitative Pathology, Stavanger University Hospital, Norway.

Results: The developed methods are able to predict the HER/2 status of the subjects with accuracies almost equally as good as the evaluation of the pathologist. The unsupervised method is able to match the pathologist’s assessment by accuracies above 0.9 per single image and to predict the status of the considered subjects with an accuracy of 0.97 if five images for each patient are considered.

Simulink as a Core Tool in Development of Next-Generation Gripen

Henric Andersson, Systems Engineer, Saab Aerosystems, Linköping, Sweden
Anders Weitman, Systems Engineer, Saab Aerosystems, Linköping, Sweden
Johan Ölvander, Department of Mechanical Engineering, Linköping University, Sweden

In the planning and concept study phases of the next-generation Gripen fighter aircraft, methods and tools studies have been performed. Capabilities and limitations of the Simulink toolset have been evaluated to explore how it can support model-based systems/software engineering. This paper presents three approaches of Simulink usage for functional development:

1. The functional-oriented systems modeling and simulation approach, where the function is in focus; complete enough to be simulatable, but abstract from an implementation point of view.
2. An implementation-oriented specification approach that is based on a modeling framework with predefined system architecture, scheduling, data types, and rules for discretization. The resulting embedded software is hand-coded using the model as specification.
3. Similar to approach two but here the embedded software is automatically generated using a high-quality code generator.

The driver for choosing the approach is threefold: high quality, short development time, and low cost. Some experiences based on these prerequisites are presented, mainly concerning the aspects of scalability, such as model architecture, license model, and project ramp-up challenges. The results are also compared with the existing SystemBuild based development environment. When introducing high-end engineering practices and tools such as Simulink in an organization developing safety-critical products, it is important to make sure that basic management practices (e.g., requirements, configuration, and change management) are also thoroughly handled.

Finally, some needs/suggestions for further improvements are discussed.

Simulating the Impact of Topographical Microstructures on Triangulation Measurement Setups Using MATLAB

Jan Thim, Research engineer, Mid Sweden University, Sundsvall, Sweden
Mattias O´Nils, Professor, Mid Sweden University, Sundsvall, Sweden
Anatoliy Manuilskiy, Research engineer, Mid Sweden University, Sundsvall, Sweden
Benny Thörnberg, Assistant professor, Mid Sweden University, Sundsvall, Sweden

The paper manufacturing industry is currently exploring the possibility of measuring microstructural topography online in a paper manufacturing machine, which is intended to lead to a more precise measure of the paper quality reel to reel and a more efficient use of raw material. This paper presents a MATLAB simulation model that can be used to configure such measurement readout systems and includes a demonstration of the model in use. The model will also be used for research purposes in order to assist in gaining a better understanding of both the limitations and possibilities of such measurement systems. In this regard the angular shading of microstructures and center of gravity (CoG) functions are included in the attributes that require further exploration.

Seamless integration of MATLAB/Simulink and Rubus ICE

Kurt-Lennart Lundbäck, Arcticus Systems AB, Stockholm, Sweden
Staffan Sandberg, Arcticus Systems AB, Stockholm, Sweden
Mats Lindberg, Arcticus Systems AB, Stockholm, Sweden
John Lundbäck, Arcticus Systems AB, Stockholm, Sweden

In this paper, we present experiences and ongoing research and development with integration of MATLAB and Simulink and Rubus ICE in the customer’s software development process. Rubus Integrated Component Environment (Rubus ICE) is an IDE for component-based development of real-time systems. The Rubus ICE tool suite provides graphical, state-of-the-art, component-based execution modeling, pre-run-time execution analysis, and generation of the run-time environment for a specific target. By focusing on execution modeling, the real-time requirements can be analyzed and guaranteed.

With a seamless integration to Simulink and code generators, analysis is combined with automatically generated application code, without double workload and glue code. Furthermore, the Simulink model and the Rubus ICE model are kept synchronized (with respect to subsystems, ports, data types, and so forth).

Use of MATLAB for Radar Remote Sensing of Forests

Gustaf Sandberg, Ph.D. Student, Chalmers University of Technology, Göteborg, Sweden

MATLAB provides an excellent tool for the radar remote sensing researcher. Radar remote sensing from air- or space-borne platforms provides a unique view of the earth’s surface. Because radar is an active system, it can operate regardless of light conditions and the radar signals can penetrate cloud cover. A possible application of radar remote sensing is the mapping of forests around the world. Many studies have shown that radar signals, especially at low frequencies, contain information on the amount of biological mass in forests. Using radar satellites, it could be possible to provide maps of the earth’s forest biomass with unprecedented accuracy. This would be an important variable for climate models, which are more important than ever with the ever-increasing threat of climate change. The extensive capabilities of MATLAB enable the researcher to solve most problems encountered in the research, as illustrated by examples in this paper. Although there are some problems for which MATLAB does not provide the best solution, the software is very useful for the researcher in radar remote sensing.

Evaluation of Compilers for MATLAB to C Code Translation

Markus Müllegger, Software Designer, Ericsson AB, Göteborg, Sweden

MATLAB to C code translation is of increasing interest for science and industry. This paper investigates in detail two MATLAB to C compilers: MATLAB to C Synthesis (MCS) and Embedded MATLAB C (EMLC). Three aspects of automatic code generation were studied: generation of reference code, target code generation, and floating-to-fixed-point conversion. The benchmark code used covers simple to more complex code. The compilers are viewed from a theoretical as well as a practical perspective. A fixed-point filter implementation is elaborated. EMLC and MCS offer several fixed-point design tools. MCS provides a better support for C algorithm reference generation, by covering a larger set of the MATLAB language as such. More suitable for direct target implementation is code generated from EMLC, which allocates memory only statically. Functional correctness was generally achieved for each automatic translation.

MATLAB Implementation of an Automated System for High-Power Diode Laser Characterization with GUI and Direct Database Connectivity

Matei Rusu, Application Engineer, Modulight Inc., Tampere, Finland
Kaj Torrkulla, Product Engineer, Modulight Inc., Tampere, Finland

Using the powerful and flexible features built in MATLAB, Modulight engineers assembled a fully automated testing rig for high-power semiconductor lasers. Besides complex testing, the MATLAB application is able to perform full data analysis, as well as direct interfacing with Modulight’s database, MLDB. The MATLAB application uses Instrument Control Toolbox to communicate with a number of laboratory devices (power supply, optical spectrum analyzer, multimeter, optical powermeter, and temperature controller). A GUI complements the functionality of the application, offering easy editing of the test parameters.

Object-Oriented Approach to the Development of Level-2 M-File S-Function

Gianmarco Romano, Assistant Professor, Dipartimento di Ingegneria dell'Informazione, Seconda Università di Napoli, Aversa (CE), Italy

We present an object-oriented approach to the development of custom blocks in Simulink. MATLAB provides an object-oriented programming model, as the M language provides classes and other object-oriented concepts such as inheritance and interfaces, but the Simulink M-file S-function cannot take advantage of such features. Custom (M-file) blocks cannot be implemented as a class, but single (callback) methods must be implemented, and internal state, or any data that must be passed between callbacks in an S-function, can only be represented by simple data type; for example, DWork vectors in Level-2 M-file S-function cannot be of class data type and no analog of PWork, available for C++ S-functions, exists. As the availability of object-oriented MATLAB programs increases, a way to reuse MATLAB classes in S-function is needed in order to avoid a complete rewrite of algorithms as simple functions. We show how to overcome such limitations by providing an example of a synchronous CDMA Simulink model that uses MATLAB classes to implement custom blocks in Simulink. Such an approach can ease and accelerate the development of custom blocks, especially for programmers accustomed to the modularity and abstraction of object-oriented languages and can provide scaling and productivity gains as those achieved by object-oriented languages.

Object-Oriented Approach to the Development of Level-2 M-File S-Function

Gianmarco Romano, Assistant Professor, Dipartimento di Ingegneria dell'Informazione, Seconda Università di Napoli, Aversa (CE), Italy

We present an object-oriented approach to the development of custom blocks in Simulink. MATLAB provides an object-oriented programming model, as the M language provides classes and other object-oriented concepts such as inheritance and interfaces, but the Simulink M-file S-function cannot take advantage of such features. Custom (M-file) blocks cannot be implemented as a class, but single (callback) methods must be implemented, and internal state, or any data that must be passed between callbacks in an S-function, can only be represented by simple data type; for example, DWork vectors in Level-2 M-file S-function cannot be of class data type and no analog of PWork, available for C++ S-functions, exists. As the availability of object-oriented MATLAB programs increases, a way to reuse MATLAB classes in S-function is needed in order to avoid a complete rewrite of algorithms as simple functions. We show how to overcome such limitations by providing an example of a synchronous CDMA Simulink model that uses MATLAB classes to implement custom blocks in Simulink. Such an approach can ease and accelerate the development of custom blocks, especially for programmers accustomed to the modularity and abstraction of object-oriented languages and can provide scaling and productivity gains as those achieved by object-oriented languages.

Simulation of a Wideband Reconfigurable Multi-Antenna System with Space-Time Coding

Nima Seifi, Ph.D. Student, Chalmers University of Technology, Göteborg, Sweden
Ali Soltani Tehrani, Ph.D. Student, Chalmers University of Technology, Göteborg, Sweden
Mats Viberg, Professor, Chalmers University of Technology, Göteborg, Sweden

MIMO systems will play a crucial role in the future of wireless systems; for example, multiple antenna techniques can be a key to boost the performance of modern wireless systems. In order to study the performance and complexity of MIMO systems, a suitable simulation tool is essential. Simulation is important in system architecture exploration, algorithm optimization, and bottleneck detection. Simulation is also a powerful tool when designing various components in an antenna system such as electromagnetic solvers for radiation patterns and circuit simulation for amplifiers, but simulating an entire antenna system is not usual.

In this paper the capabilities of Simulink as a simulation platform for MIMO systems have been investigated and a typical MIMO-WCDMA link has been developed. Some of the blocks were used from the WCDMA library, and others were developed accordingly. The space-time encoder and decoder for a general MIMO link were also developed. Furthermore, in order to be able to test different antenna types in a system, the antenna block is separated from the channel. Henceforth, a double directional channel has been developed based on the spatial channel model (SCM) proposed by the 3rd Generation Partnership Project (3GPP), which is an environmental channel independent from the antenna. The results of the simulation show that Simulink is a capable platform for simulation of an entire communication link.

On a MATLAB Open-Source CAD/CAM/CAQ Software System for 2D NURBS Curve Segments Milling

Wolfgang H. Koch, Prof. Dr., Department of Production and Quality Engineering, Norwegian University of Science and Technology
Jingming Huo, MSc., Department of Production and Quality Engineering, Norwegian University of Science and Technology

Because of increasing demands for higher productivity, flexibility, and quality in part production, activities developing software for more sophisticated data creation, processing, utilization, and integration are accelerating and extending dramatically. Mainly due to business reasons, these developments are almost providing closed-box software systems with nearly no chances for users to interact reasonably whenever things are not running well. This seems to be a strategic drawback – not only in having no control over causes and effects. One solution would be to also offer also open-source software, enabling reasonably and suitably interactive adaptation, correction, and operation.

Making visible what might happen in commercial software systems for computer-aided design, manufacturing, and quality assurance (CAD/CAM/CAQ), the open-source MATLAB program system OpenCM for the engineering design, fabrication, and quality assurance of freeform geometries realized exemplarily with two-dimensional NURBS curve segments is being presented.

The program system imports as input NURBS curve segments (the nominal CAD-information) and generates ready-to-use NC codes for the final machining on a milling machine tool. As one example for algorithms that tracks the imported curve segment for the NC code creation, an adaptive secant method is demonstrated. After the machining, the part is measured with a coordinate measurement machine. Again with MATLAB program, the measured results (as actual information) forming a point cloud are visualized, quantified, and compared with the nominal CAD-information showing the performance of the entire program system. For reference purposes, a comparison with certain commercial CAD/CAM/CAQ software systems has also been carried out, confirming the suitability of the proposed approach.

A Welch Power Spectrum Implementation Using Simulink and Xilinx Sysgen

Carsten Siggaard, Senior Consultant, Danish Technological Institute, Århus Denmark
Dan Anov, Senior Consultant, Danish Technological Institute, Kolding Denmark

Passive acoustic emission spectroscopy is a technique used in industrial environments where noise and vibrations carry information about the state of the machinery and the product being processed by the machinery. The vibrations can be measured by laser, microphones, or accelerometers. In the Data Interpretation and Model Management System (DIMMS) project, the processing of the data is done in two steps; the first step is a preprocessing from the time domain into the frequency domain using a FFT and the 2-norm squared of the (complex) result. The result is a real-valued power spectrum. The second step is to match the result of the first step with a chemometric model. The second step will not be described in detail here. The calculations can be performed in software; however, the introduction of DSP slices in modern FPGAs makes the calculation of power spectrum using FPGAs feasible. We will present an algorithm implemented in an FPGA using Simulink and Sysgen from The MathWorks and Xilinx, respectively. The result is an implementation that is capable of calculating power spectra on four channels at a sampling rate of 250Ks/s.

MIRtoolbox: An Innovative Environment, on Top of MATLAB, for Music and Audio Analysis

Olivier Lartillot, Researcher, Finnish Centre of Excellence in Interdisciplinary Music Research, University of Jyväskylä, Finland
Petri Toiviainen, Professor, Finnish Centre of Excellence in Interdisciplinary Music Research, University of Jyväskylä, Finland
Tuomas Eerola, Researcher, Finnish Centre of Excellence in Interdisciplinary Music Research, University of Jyväskylä, Finland

MIRtoolboxis a MATLAB toolbox dedicated to the extraction of musical features from audio files, including routines for statistical analysis, segmentation, and clustering. MIRtoolbox features a user-friendly syntax that enables users to easily combine low- and high-level operators into complex flowcharts. The modular design of MIRtoolboxis guided by a philosophy of expertise capitalization: Techniques developed for certain domains of music analysis are turned into general operators that could be used for different analytical purposes. The feature extraction algorithms are decomposed into stages and formalized using a minimal set of elementary mechanisms and integrating different variants proposed by alternative approaches – including new original strategies – that users can select and parameterize.

Particular attention has been paid to the design of a syntax that offers both simplicity of use and transparent adaptability to a multiplicity of possible input types. Each feature extraction method can accept as argument an audio file or preliminary results from intermediary stages of the chain of operations. The same syntax can also be used for analyses of single audio files, batches of files, series of audio segments, multi-channel signals, and so forth. For that purpose, the data and methods of the toolbox are organized in an object-oriented architecture.

Memory management mechanisms allow the analysis of large-scale corpora, through the integration of automated chunk decomposition mechanisms and of distinctive processes for flowchart design and evaluation. A set of meta-functions have been designed that enable the integration of user-defined algorithms with the help of simple templates.

Anomaly Detection Algorithm Test Bench for Mobile Network Management

Pekka Kumpulainen, Research scientist, Tampere University of Technology, Automation Science and Engineering, Tampere, Finland
Kimmo Hätönen, Senior specialist: Security research, Nokia Siemens Networks, Research, Technology and Platforms, Espoo, Finland

Huge amounts of operation data are constantly collected from the performance monitoring and system logs from servers in communication networks. One of the key applications for system operators is to detect anomalies in the data. The anomalies may indicate various types of situations that need attention. They can be results of malfunctions of the hardware or software components. They may also indicate new unseen activity of the customers or possible unauthorized intrusion. There is a large selection of anomaly detection methods to choose from. In this paper we present an application developed for testing anomaly detection algorithms in network management environment. We introduce a test bench implementation and illustrate its properties with two local anomaly detection methods. One is based on self-organizing maps (SOMs) and clustering. Another one is an extension to a basic clustering method, utilizing two-layer clustering. The test bench consists of compiled MATLAB code and is excellent for rapid prototyping of algorithms. It can be easily deployed in any network operator’s environment for testing and demonstration purposes without massive software installations. It helps the network operators in selection of the algorithm and its parameters for the final applications.

Data Assimilation of Large-Scale Models Within the Petroleum Industry

Rolf J. Lorentzen, Senior research scientist, IRIS, Bergen, Norway
Geir Nævdal, Chief scientist, IRIS, Bergen, Norway

The main focus in this paper is data assimilation of large-scale models within the petroleum industry. The models describe flow of hydrocarbons in porous media (reservoirs) and two-phase flow in pipelines. The measurements are usually production data from the wells (flow rates and pressures) but may also be other observable quantities. MATLAB is used to implement the data assimilation routines and is coupled to a simulator solving the model equations for each application.

The methodology used for data assimilation is the ensemble Kalman filter (EnKF). The EnKF is a Monte Carlo approach and was originally developed to estimate dynamic variables in large-scale atmospheric and oceanographic models. Lately, it has also been used with success within the petroleum industry. The EnKF is well suited for large-scale problems because only a small number of forward simulations are needed. Also, covariance matrices are not explicitly calculated, and memory problems are thereby avoided. In addition, to be more suitable for large-scale models, the EnKF has shown to have better performance than the extended Kalman filter for highly nonlinear problems.

In our applications we have used MATLAB for managing the total workflow, solving the EnKF equations and communicating with in-house and commercial simulators. In addition it is also used for presentation of most of the results.

Life-Cycle Management of Simulation Models

Jing Chen, M.Sc. Student, Royal Institute of Technology, Stockholm, Sweden
Mats Larsson, Senior Analyst, Systemite AB, Göteborg, Sweden
Jan Söderberg, CTO, Systemite AB, Göteborg, Sweden

Simulation and code generation have become well-established methods in electronic systems development. For development of complex systems in a large organization, such as the development of automotive electronics, there is a need for concurrent engineering as well as support for life-cycle management of the product data, including the simulation models. To get the full value out of behavioral simulation activities, these activities should to be integrated with the entire development process, in the same way that CAD drawings and geometrical simulation models already are. Related product development information such as customer requirements and system design should be integrated with the simulation model for improved traceability, product quality, and development efficiency. This integration has traditionally been performed by manual operations, or at best using file-based configuration management tools. The increasing complexity of the systems and the growth of development organizations, however, require the support of information systems, denoted product data management (PDM) or product life-cycle management (PLM) systems. This kind of support is often based on the native file structure used by the simulation tools, but such solutions limit the effectiveness of the integration. This paper presents an approach where the integration is based on the fine-grained structure of the simulation models – for the case of Simulink, the subsystems and other simulation blocks. A prototype implementation has been developed where the simulation model is managed in the PLM system SystemWeaver. Use of the prototype has demonstrated that the fine-grained integration is a possible and promising approach.

Application of MATLAB to Calculate Seismic Traveltime in a Reservoir Layer for Time-Lapse Seismic Analysis

Khanh Duc Nguyen, Ph.D. Student, University of the Faroe Islands, Torshavn, Faroe Islands

In time-lapse seismic analysis, one of the complementary techniques often being used is to study changes in seismic traveltime between different surveys and to interpret such time shifts, often referred to as pull-up or push-down effects, in terms of production-related effects. By measuring traveltime shifts in time-lapse seismic datasets between top and bottom horizons of a reservoir layer, it is possible to estimate the average velocity change within that reservoir layer. To do this conversion we need to calculate the traveltime in the reservoir layer. One of the methods for doing this is to use the ray-tracing technique. However, commercial ray-tracing software only allows us to calculate the total traveltime of a seismic wave from its source to the reflectors or receivers. This paper presents a ray-tracing technique using MATLAB to calculate seismic traveltime in a reservoir layer. The results have been used to estimate the average velocity change in the Utsira formation of the Sleipner field in the North Sea.

Temperature Estimation in Power Modules via Power Loss Calculation

Florin Lungeanu, Control Engineer, Danfoss Drives A/S, Graasten, Denmark
Marian Lungeanu, Hardware Engineer, Danfoss Drives A/S, Graasten, Denmark

This paper focuses on modeling classical three-phase inverter topologies in steady-state conditions, for the final purpose of estimating the junction temperature of semiconductor devices in power modules. The interim result is the power loss in the semiconductors, which will be eventually passed as the input to the thermal model of the whole mechanical system: semiconductors, power module case, and the heat sink attached to it. The idea of this work is to use simple steady-state equations based on averaging the voltage drops and phase currents instead of complex dynamic simulations, an approach believed to give fast and accurate results in spite of the simplifying hypotheses inherently attached to any steady-state analysis. The authors based their work on static examination of the whole electronic topology for one output voltage period, in order to generalize the method and facilitate the implementation of a comprehensive database relating the junction temperature cycles with power module type, modulation pattern, and the type of the motor loading. The results were compared against measurements and shown to be a good fit, demonstrating that fast and accurate engineering design can be done in that popular MATLAB environment instead of more computing-demanding tools such as SPICE or SABER and that the close cooperation between control and hardware engineers can put together superior performance and simplicity.

Automatic Model Generation for Robotic Applications

Omar Al Assad, Ph.D. Student, General Electric Healthcare, Buc, France
Emmanuel Godoy, Professor, Supelec, Gif sur Yvette, France
Vincent Croulard, Senior Engineer, General Electric Healthcare, Buc, France

Robot motion simulators are very important tools used by engineers, researchers, and students for several purposes such as designing controllers and generating trajectories. These simulators can be implemented by means of numerical computing software such as MATLAB and Simulink. Yet the manual implementation of the mechanical equations involved in these models is the source of a number of hidden errors that are very difficult to correct. This implementation is also very burdensome, especially when the developer is not familiar with robotic formulations. Therefore, a new MATLAB application called RobMod has been developed to offer to a large community of users a simple modeling process free of inadvertent human-made mistakes. This application is based on a MATLAB GUI that assists the developers through successive steps to collect the robot parameters. Then, it generates the associated Simulink model.