Headline

We will have two invited talks (by postdoc candidates) in this slot:

14-15h: “Tracking and Communication in Wireless Networks Using Differential Forms and Ricci Flows”, by Dr. Rik Sarkar 15-16h: “Joint Power and Rate Control in Linux mac80211 systems”, by Thomas Hühn

Details about the talks and authors are available below.

TITLE: “Tracking and Communication in Wireless Networks Using Differential Forms and Ricci Flows”

ABSTRACT: Small and portable devices are becoming the norm in computing. The first part of this talk is about monitoring mobile objects in a network where movements of devices are detected by nearby sensors or communication nodes. I will describe a protocol called “differential forms” for storing the detection data locally and answering questions about object locations at low cost. For example, the number of objects in any user given region is counted by inspecting just the boundary of the region. The technique preserves user anonymity, and is naturally robust to detection errors, coverage holes, node failures and motion.

The second part of the talk describes an approach to location based processing of sensor data. Simple and elegant algorithms that use sensor locations have been designed for routing and storage. However, the protocols do not work well when monitoring an irregularly shaped region with empty 'holes' where there are no sensors. In such cases, some location based methods fail, others become inefficient. I will discuss a method of assigning “virtual coordinates” to the sensors with a technique called Ricci Flow. The coordinates are designed in a way that the simple location based algorithms can be applied reliably and efficiently.

SPEAKER: Rik Sarkar is a Postdoctoral Research Scientist sponsored by the German Science Foundation, working at TU - Berlin and FU - Berlin. He completed his Ph.D. from Stony Brook University, USA, in 2010, and M.Tech from IIT Bombay, India, in 2005. His research specializations are in protocols for wireless networks and mobile communication.

• advantages comparing to LLS (Abraham etal 2004).

• Greedy routing may fail when there is 'network holes' (space). –> ricci flow.
• ricci: theory, distributed algorithm, covering space.
• compare to NoGeo.
• Geo hash table: distributed data storage.

TITLE: “Joint Power and Rate Control in Linux mac80211 systems”

ABSTRACT: Todays'IEEE 802.11-based wireless networks perform far below the achievable limits when multiple participants share the same frequency spectrum in an uncoordinated manner. Furthermore, many of today’s WiFi access routers are under individual administrative control which creates the need for decentralized resource allocation algorithms to optimally select transmission rates and power levels. On the other hand, the current state of the art in resource allocation algorithms are rather limited. For instance practical Transmit Power Control (TPC) is managed in the IEEE 802.11 PHY layer by individual device drivers and assign one static power level to all transmitted packets. A more sophisticated scheme requires knowledge about wireless neighbors, transmission rates and medium access state per link, which are typically handled at the MAC layer. Such schemes are expected to provide a better network performance by decreasing the level of overall interference and therefore, a higher efficiency of common spectrum usage.

Our measurement results serve as a guideline to design a robust and stable controller. In this talk we take a bottom up approach: we leverage information and control capabilities of the IEEE 802.11 PHY layer, abstract the driver specifics and implement a joint rate and power controller that works on a per link basis. Our algorithm “Minstrel-Blues” uses statistical feedback obtained by sampling different rate and power settings. By jointly controlling rate and power, our controller does significantly decrease the transmit power per communication while maintaining the same link performance in different scenarios. This directly improves spectrum efficiency as using less transmission power decreases the overall interference.

SPEAKER: Thomas Hühn received his diploma of Business Administration and Computer Engineering from the Technical University of Ilmenau in 2003. After founding a registered corporation to provide wireless broadband internet connectivity in rural areas of the federal state Thüringen, he joined Prof. Anja Feldmann’s group as a PhD student in August 2007 . Starting to design and build the research testbed in MAGNETS, which turned into BOWL, he focused on decentralized resource allocation in 802.11 networks for his research. He expects to finish PhD degree within April 2012.