 |
|||||||||
 |
home > Research projects > Theory Theory of self-organized, distributed communication and information[projects] [publications] List of projects
Information and coding theory for wireless multi-hop networksContact: Prof. Suhas Diggavi , Emre Telatar , Ruediger Urbanke Relevant publications Network theory for wireless multi-hop networksContact: Prof. Suhas Diggavi , Matthias Grossglauser , Emre Telatar , Patrick Thiran Relevant publications Last updated on: Sept. 05 This work seeks to bring answers to the key question of the scalability of networks coverage, connectivity, and transport capacity. It also intends to find decentralized random access schemes that do not coordinated transmissions between neighboring nodes outside a local area and that achieve a high transport capacity. Finally, it will work to allow networks that are partitioned to communicate.
The project's assets It is specifically interested in applications that exhibit high node mobility and possess some delay tolerance. The central question in this project concerns the architecture and service models that suit highly mobile sensor-actuator networks. The project aims to develop routing schemes for fully mobile networks that take advantage of geographic routing when no explicit positioning information is available. It is also interested about studying what happens when a sensor network registers rare events – natural disasters – resulting in very sparse traffic, but needing to react quickly. What has been achieved so far?In the first phase of this project, it has been demonstrated that node history contains valuable information about a network topology. This led to develop new routing schemes that perform well under high mobility. The current focus in the research community is on epidemic routing algorithms, where a message is flooded throughout the network to eventually be delivered to the intended destination. But it is believed that much more efficient algorithms can be developed. Distributed signal processing and communication in sensor networksContact: Prof. Martin Vetterli Relevant publications Last updated on: Sept. 05
Applications Concrete applications are possible in the area of tomography. The temperature-dependent sound speed, together with the wind, speed can be estimated by measuring the travel time of an acoustic signal between sound sources and receivers, and then solving an inverse problem. Although it has been successfully used in the oceanology for many years, acoustic tomography started to be applied to the atmosphere only few years ago. The project's assets This work deals with one of the biggest challenge in wireless sensor networks: the absence of separation theorem in multi-terminal networks. Also, new modalities for signal acquisitions and reconstruction are interesting because they allow to “see” phenomena unseen before, as well as to get potentially much more resolution than what is expected from a standard set up. In particular, tomographic methods are original in the sensor network arena. What has been achieved so far? The question of sampling, which is related to the spatiotemporal distribution of the physical phenomena has started to be investigated. With encouraging results. For example, the acoustic case has been analyzed in detail. Distributed compression has become a popular topic in the information and communication theory community. Research is being pursued especially on distributed Karhunen-Loeve Transform (dKLT) and on rate-distortion bounds on distributed signals. A precise analysis of the reconstruction error incurred by the dKLT has been provided for different scenarios both of theoretical and practical interest. In the area of tomographic sensing, the scientists have developed an algorithm for reconstructing temperature and wind flow. The estimation procedure takes into account the refraction of the sound rays due to the movement of the medium (flow field). This has not been considered in the existing techniques where only the straight sound ray model is assumed. The algorithm has been tested on synthetic data. Currently, the real model size experiments are conducted. Algorithmic foundations of ad hoc and sensor networksContact: Prof. Roger Wattenhofer , Peter Widmayer Relevant publications Last updated on: July 07
Applications To obtain a complete picture of what is happening in the algorithmic world that relates to multi -hop wireless networks. The project assets This work aims at deeper understanding of the algorithmic fundamentals of ad hoc and sensor networks. It focuses in particular on MAC layer issues and mobile routing. In particular it considers models not yet studied in an algorithmic context, such as the signal to interference plus noise ratio (SINR) model, or the bounded independent neighbourhood model. In addition it is planned to implement and test the most promising practical results on various test beds. What has been achieved so far? The primary focus of this project was on topology control, clustering (dominating sets and variants thereof), geo-routing, unstructured radio net-works, embeddings and virtual coordinates , and getting a deeper algorithmic understanding of interference. Where does the project stand now? The process of understanding old questions in new models is starting For instance, a topology control algorithm in the signal to interference plus noise ratio model has already been developed. SensorScope and applications in environmental monitoringContact: Prof. Marc Parlange , Martin Vetterli Relevant publications Last updated on: July 07 Work in progress During winter 2007, a SensorScope network was deployed on an alpine glacier. The chosen location was the Plaine Morte glacier, in Switzerland. For this campaign, several SensorScope weather stations were deployed during a high pressure event with the objective of gaining understanding of the turbulent subgrid-scale physics near the snow-atmosphere interface. The Plaine Morte glacier is located at 2750 m in the canton of Berne and is one of the largest flat-glaciers in the Alps. It extends for about 10 km2 and is located in the Wildstrubel range. The deployment took place between March 12 and March 16, 2007 and consisted of 13 SensorScope weather stations measuring continuously (every 30 seconds) air temperature and humidity, wind speed and direction, short wave solar radiation, and surface temperature. Besides the environmental data, the weather stations regularly transmit basic status information, such as battery levels, sampling times, solar energy, and consumption. All these measurements are routed through the wireless network to a base-station equipped with a GPRS connection. This GPRS base-station forwards all the collected information to a central server via the GPRS network. During the deployment, all the information was available in real-time through our web interface http://sensorscope.epfl.ch/wiki/index.php/SensorScope Applications Sensors networks are used for environmental monitoring. The can measure large number of data, such as surface temperature, air temperature, humidity. The project's assets Environmental monitoring is considered as one of the prime application fields for sensor networks today. Although several projects already address this issue, dense monitoring of an entire watershed has never been attempted. Dense monitoring means that a number of sensors are located around one point. They will therefore measure larger number of relevant quantities of data, at a lower precision, but on a much denser grid. The fundamental interdisciplinary nature requires a close collaboration between environmental scientists and communication specialists. Finally, the researchers have already a substantial operational experience with SensorScope. |
||||||||
 |
|||||||||
|
|||||||||
| |||||||||
|
© NCCR MICS - mics@epfl.ch |
|||||||||
