Friday June 9th. Place JBH391 (Math Conference Room) Three Presentations at 10:30, 11:30, and 1 pm. Time: 10:30-11:30 Speaker: Sara Woodworth (swood@cs.ucsb.edu) Title: Membrane Systems: A Natural Approach to Computing Abstract: The natural world has been "computing" since life began. Many of these computing mechanisms in nature (such as Artificial Neural Networks, DNA Computing, and Cellular Computing) have been studied in order to find new computing paradigms. Recently, a new area in natural computing has been introduced called Membrane Computing which looks at the manner in which cellular membranes control the evolution and communication of the objects they contain with the environment. The biological cell consists of proteins, enzymes, molecules, etc. (generically referred to as objects) surrounded by a membrane. This membrane is used to separate these objects from the environment surrounding the cell. The membrane allows controlled exchanges of these objects with the environment along with facilitating the evolution of objects (changing protein 'a' into protein 'b') within the membrane. These natural processes can be viewed as a computation. Membrane computing explores, abstracts, and formalizes this new method of computation inspired by the natural membrane model. A number of membrane system models have been examined and most have been found to be computationally complete (i.e. equivalent is computing power to a Turing Machine) along with offering advantages over our current computing models. Currently our computers are based on a model of computing which use a centralized, deterministic, sequential mode of operation. The natural cellular world however, uses a decentralized, nondeterministic, maximally parallel mode of operation. By exploiting the decentralized, parallel power of membrane systems we are able to solve some problems much faster than with our current silicon computers. This distinction can be seen by looking at NP-complete problems (which take an exponential amount of time to solve using deterministic, sequential systems). Some membrane system models have been found to be able to solve NP-complete problems in polynomial (and usually linear) time by operating in a fully parallel manner and trading space for time. In this talk, the area of membrane computing is introduced and many of the current models are defined and explored in terms of both computability and complexity. Most of these models have been shown to be computationally complete. Next, restrictions to these models are discussed. We will look at restricted systems in terms of the number of membranes, the number of objects, and the size of the rules. We find some features of systems are unnecessary in terms of computing power while others lead to non-universal systems. Time: 11:30 AM - 12:30 PM TITLE Mobile Augmented Reality SPEAKER Jason Wither Department of Computer Science University of California Santa Barbara ABSTRACT Augmented Reality (AR) exists in the space between the real world and virtual reality, containing both real world imagery and computer graphics which augment the user view with more information than would be available otherwise. The following three characteristics are often used to help define an AR system: 1. Combines real and virtual 2. Interactive in real time 3. Registered in 3-D Wearable computers along with sensory input devices and wearable displays allow mobile users to be immersed in augmented reality. Mobile AR has many potential uses, from construction and maintenance to tourism to military applications among many others. Superman x-ray vision could even be possible with mobile AR. There are however many difficulties to overcome for successful mobile AR applications. This talk will describe what many of these challenges are, and summarize work that has been done to begin solving them, focusing on three main areas. We will first look at the hardware requirements of mobile AR, and the limitations currently imposed by hardware. We will next talk about the registration problem of aligning virtual objects with the real world, including several approaches to solving this problem both in a lab environment and outdoors. Lastly we will talk about techniques used in different areas of mobile AR. We will discuss techniques for interaction, outdoor modeling, authoring of AR content, information management and collaboration. We will conclude by discussing in general what problems currently exist with mobile AR, and possible directions to solve them. Speaker: Stephan Karpinski Time: 1-2pm Title: Wireless Traffic: The Failure of CBR Modeling Abstract: This presents reserach that challenges and disproves a fundamental assumption that has been made in almost all research using simulation or experimental deployment to evaluate wireless network performance. This assumption is that the pattern of user and application traffic generation does not affect the performance of lower levels of the network enough to be worth worrying about. Currently, it is standard, for evaluations of new wireless protocols and technologies to use a uniform, constant bit-rate (CBR) traffic model: some arbitrary number of constant bit-rate flows of arbitrary, fixed size and duration are exchanged between randomly chosen nodes in the wireless network. It is clear that this is not a realistic usage pattern for any but the most contrived networking scenarios. The implicit assumption in research that uses such models is that uniform CBR traffic is ``good enough'' to still allow a meaningful performance evaluation. We will first formalize the notion of being ``good enough'' by defining what it means for a simulation model to be sufficiently realistic. We then present a new and statistically rigorous methodology for evaluating whether commonly used traffic models are sufficiently realistic or not using a differential simulation technique. Finally, we demonstrate that the answer is resoundingly negative: important performance metrics at all levels of the protocol stack are drastically skewed when uniform CBR traffic models are compared to real traffic.