MC IMPULSE EU FP7 Marie Curie Initial Training Network MC Impulse (Monte Carlo based Innovative Management and Processing for an Unrivalled Leap in Sensor Exploitation) is a new Initial Training Network, focussing on research and training in the field of sensor data processing. Within MC IMPULSE 14 early stage researchers and 2 experienced researchers will be recruited and trained. The MC IMPULSE consortium will build a durable cooperation in research and training in the field of sensor data processing. The research programme will focus on the development of novel methods for real-time signal and data processing, based on data collected form sensors or networks of sensors. MC IMPULSE is expected to generate specific innovations such as: • Advanced methods for detecting and tracking objects by processing information from (a network of) sensors, • New methods and algorithms for information fusion from multiple sensors and their efficient processing for on-line applications. One of the key techniques that will be focussed on is Sequential Monte Carlo or Particle Filtering. MC IMPULSE brings together various European industrial and academic key players in the field of sensor and data processing to develop and execute a multidisciplinary and networked European research training programme. The network consists of a mixture of larger and smaller innovative companies, various universities and one research establishment. The participants in the network are: Thales Nederland B.V., Lancaster University, Rinicom Ltd., Twente University, Saab AB, Xsens Technologies B.V. , Linköping University and Fraunhofer Institute for Communication, Information Processing and Ergonomics (FKIE). MC IMPULSE joins different scientific disciplines such as: electrical engineering, communication theory, estimation theory, signal processing and detection theory, mathematics and physics. There are 14 openings for early stage researchers (ESR, 3 year position) and 2 openings for experienced researchers (ER, 2 year positions). For an ESR the requirement is that the applicant has recently earned a M.Sc. degree or equivalent. For an ER the requirements are that applicant has recently earned a Ph.D. degree. More detailed conditions for employment can be found in the Work Programme 2008 People: ftp://ftp.cordis.europa.eu/pub/fp7/docs/wp/people/m_wp_200801_en.pdf The ESR positions are: Subject: Group tracking & distributed estimation in Wireless Networks and System Host: Lancaster University. Contact: Dr. Mila Mihailova (email@example.com) Description: Algorithmic aspects and theoretical underpinnings of distributed estimation, detection and decision making in wireless networks and systems will be studied, including methods for distributed estimation and inference, in particular Gaussian mixture models combined with particle filtering; and methods for group object tracking related with the structure of the sensor network. Subject: Structured models for non-linear filtering Host: Linköping University Contact: Prof. Fredrik Gustafsson (firstname.lastname@example.org) Description: A particular promising area for the Marginalized Particle Filter (MPF) is in Simultaneous Localization And Mapping (SLAM). State of the art is the fast SLAM algorithm which in itself is a MPF. However, the state vector used in literature is low- dimensional, including horizontal position and heading angle only. Linköping University has recently found that a second layer of marginalization could extend the scope to much more complex models. Subject: Track-Before-Detect based on acoustic sensors Host: Saab AB Contact: Dr. Egils Sviestins (email@example.com) Description: Problems related to acoustic sensors are varying characteristics of sound, its propagation and the background noise. The traditional way of resolving these problems is based on thresholding, resulting in a great loss of information. A much better performance is expected by using raw data in the tracking process. Saab believes Track- Before-Detect based on particle filters has great potential Subject: Integrated information extraction Host: Thales Nederland B.V. Contact: Dr. Yvo Boers (firstname.lastname@example.org) Description: Optimal schemes for appropriately exploiting all the available information are currently not available. In principle particle filters allow for optimal solutions once the modelling has been performed. Therefore, correct modelling and an efficient particle filter implementation will be necessary in order to be able to exploit all the available information. Subject: Localization in Unknown Urban Environment Host: Lancaster University. Contact: Dr. Mila Mihailova (email@example.com) Description: Solving localization problems in an urban environment meets different challenges due to obscuration, non-line-of sight, fading and other noises and uncertainties. Therefore, advanced localization techniques will be investigated using: angle of arrivals, time difference of arrivals, Non-Line-Of-Sight (NLOS) or Received signal strength (RSS) data. Bayesian inference techniques have proven their power in such cases and will be the main framework for localization. Subject: Exploiting prior sensor & environmental knowledge Host: Thales Nederland B.V. Contact: Dr. Yvo Boers (firstname.lastname@example.org) Description: Prior sensor knowledge is oftentimes ignored or used only on an ad hoc basis. This project seeks to use prior sensor knowledge and to exploit external information to increase system performance. Subject: Localization and calibration of ground sensors Host: Saab AB Contact: Dr. Egils Sviestins (email@example.com) Description: Locating multiple ground sensors where satellite navigation is unreliable makes it difficult to ensure consistent information. The project aims at developing methods for locating, aligning and calibrating a large number of sensors so that consistent information is given, which is a prerequisite for reliable tracking. Subject: Hybrid Sensory Networks Host: Rinicom Ltd. Contact: Prof. Garik Markarian (firstname.lastname@example.org) Description: The conventional approach to sensory networks assumes a homogeneous network infrastructure. However, most real-life scenarios are based upon the use of hybrid networks. This creates a number of issues related to the reliability of the received data, remote calibration of the sensors and provision of the same quality of service to wireless and wired sensor nodes. This sub project aims at developing and implementing a prototype hybrid network which could dynamically adapt its performance. Subject: GIS for multiple sensor air-to-ground surveillance and traffic monitoring Host: FKIE Contact: Dr. Wolfgang Koch (email@example.com) Description: Context information made available by modern Geographical Information Systems (GIS) can greatly improve the results of sensor-driven ground surveillance. On the other hand, so-called “tracks” of road moving objects provide approximations to the underlying roads, which can be refined to high-precision road and up-to-date road maps. In other words, valuable context information can be gathered from sensor data fusion. Subject: Sensor and map assistance for search and rescue teams Host: FKIE Contact: Dr. Wolfgang Koch (firstname.lastname@example.org) Description: Modern fire fighters or search and rescue teams are equipped with human- borne communication and sensor systems collecting and distributing information on physiological parameters of the persons in action and surveillance sensors for perceiving their surrounding. For enhancing their efficiency in time critical and dangerous situations, this information has to be fused with mapping information Subject: Tracking in sensor networks Host: Linköping University Contact: Prof. Fredrik Gustafsson (email@example.com) Description: A sensor network usually consists of large number of similar sensors distributed in the terrain and their task is to detect, locate and track objects. In literature, little work exists on relative signal strength measurements. The objective is to start a theoretical study on optimizing the use of relative signal strengths for target detection, localization and tracking. Subject: Energy efficient sensor scheduling for sensor networks of simple sensors. Host: Twente University Contact: Prof. Arun Bagchi (firstname.lastname@example.org) Description: Particle filter based processing, combined with dynamic programming and statistical optimisation techniques will be studied aiming at more efficient and better scheduling algorithms. Important issues are to deal with multiple objects or events and parametric uncertainties in the network. Subject: Sensor management aspects Host: Thales Nederland B.V. Contact: Dr. Yvo Boers (email@example.com) Description: Sensors or sensor systems can often operate in many different modes or their settings may even vary over a continuum of possibilities. Depending on the task(s) different settings could be applied and maybe even changed on-line or in real-time. The objective is to optimize sensor tuning and incorporate this with the array of new processing techniques. Subject: Location and Positioning on WiMAX Networks Host: Rinicom Ltd. Contact: Prof. Garik Markarian (firstname.lastname@example.org) Description: Most current positioning systems is either constrained to outdoor environments or limited to a particular building or campus with installed location infrastructure. The objective is to us WiMAX networks for positioning. Multiple-Input Multiple Output (MIMO) networks, Adaptive Modulation and Coding as well as Cross Layer Optimisation will be considered. The ER positions are: Subject: Multiple target tracking in a ground sensor network Host: Saab AB Contact: Dr. Egils Sviestins (email@example.com) Description: If two sensors indicate the presence of a target, are there two targets or is it one target detected by two sensors? And if there are two targets, how does one prevent mixing them up as they move around? So far such problems have not been sufficiently explored. Subject: Indoor orientation and tracking system using MEMS Host: Xsens Technologies B.V. Contact: Dr. Ir. Henk Luinge (Henk.Luinge@xsens.com) Description: The objective is to develop an indoor position and orientation tracking system for general indoor environments. Due to the high dimensionality, a sub optimal nonlinear tracking solution must be defined. Accurate tracking while keeping track of only a limited number of particles and choosing the particles to reach highest pose observability need to be solved.