|
Center for Networking of Infrastructure Sensors
|
|
| Home | About Us | People | Sensor Networking | Active Research Grants | Publications | Testbed | ||
|
ABSTRACTS
Traditional coarse pointing, acquisition, and tracking (CPAT) systems are pre-calibrated to have the center pixel
Recent developments in pointing, acquisition, and tracking have enabled the formation of point-to-point FSO or narrow beam directional wireless networks that are capable of dynamic changes in their topology. Autonomous changes to topology in response to varying available link capacities and load demands of various nodes is called topology control. Topology control consists of computing new topologies to dynamically optimize the network under changing traffic conditions, and then carrying out the reconfiguration process to achieve the target topology. Our current work in this area studies the process of topology reconfiguration by using the packet drops that happen during this process as a cost metric. It is shown that the reconfiguration cost can be minimized when the target topology is reached by implementing the topology reconfiguration as a series of smaller steps (successive approximations). It is also shown that a topology computation algorithm that results in lower overall packet drops can be obtained by including the reconfiguration cost in the objective function along with the typical objective of congestion minimization. Simulations are used to evaluate and compare the performance of topology computation heuristics when the objective function includes reconfiguration cost.
Next generation wireless networks are increasingly complex in terms of their heterogeneity (terminal, edge and backbone nodes; directional and omnidirectional wireless links) and dynamic behavior (node mobility, atmospheric obscuration, fading). Modeling such complex systems is becoming a very challenging and cumbersome mathematical problem. This paper proposes a novel physics-based approach to the modeling, characterization and control of complex wireless networks. Heterogeneous wireless networks are modeled as physical systems where nodes are represented as particles and communication links as attraction forces between them. Forces are defined based on network connectivity and include the effects of link distance, link directivity and atmospheric obscuration. The network energy usage is used as a cost function that is shown to be related to the potential energy of the analogous physical system. We formulate the joint coverage-connectivity optimization problem in backbone-based wireless networks as an energy minimization problem and present a mobility control algorithm that mimics the natural reaction of a physical system to minimize potential energy driven by local forces exerted on network nodes. Our mobility control algorithm is shown to be completely distributed, scalable and self-organized. Initial results show the efficiency of our mobility control approach to autonomously adjust the position of controlled backbone nodes in order to optimize coverage and connectivity in dynamic scenarios.
High definition (HD) quality imagery provides clearer and more detailed information for use in advanced video surveillance systems. The deployment of such surveillance systems where no fixed communications infrastructure exists presents an ideal application for high data rate directional FSO/RF links and networks. Next generation surveillance systems using HD imagery will be able to detect and analyze objects in detail and at large distances. Such flexibly deployable surveillance systems will be very valuable for military and homeland security surveillance. Nevertheless, designing these types of systems is not an easy task. First, HD images require large amounts of bandwidth: compressed high definition television (HDTV 1080i) images require bit rates of approximately 20Mb/s, which rise to above 1Gb/s for uncompressed images at 30 frames/s, and an increase in the number of cameras in one single system can saturate the available bandwidth. Second, advanced surveillance requires significant computational power for real-time object detection, tracking, and discrimination. This paper analyzes these issues and proposes a solution with on demand video compression and real-time object detection algorithms. A system architecture of a HD scalable system with the ability to track and discriminate objects and events within the system’s deployed area will be described. Practical examples of autonomous event detection in wirelessly transmitted HDTV images will be given.
In free space optical (FSO) communication networks, pointing, acquisition, and tracking (PAT) techniques are needed to establish and maintain optical links among the static or mobile nodes in the network. First, this paper describes a precise pointing technique to steer the local directional laser beam of an optical transceiver to a target optical transceiver at a remote transceiver node. The pointing technique utilizes Real-Time Kinematic GPS coordinates, local angular sensors, and a reference baseline, to retrieve accurate navigation information (roll, pitch, yaw) of the mobile or static platform that carries an optical transceiver. Through experiments using gimbal pointing stages, we have demonstrated “dead-reckoning” pointing accuracy in the milliradian range in our outdoor testbed. Second, we provide an application example of the pointing method in a bi-connected ring network, in which the pointing technique is combined with heuristic algorithms for dynamic reconfiguration of ring network topology. The heuristic algorithms achieve near optimal solutions in a short amount of time. Lastly, we present a GPS-based autonomous reconfiguration scenario for mobile nodes, which combines the PAT technique and heuristic algorithms.
|
|
|
PRESENTATIONS |
|
|
© Center for Networking of Infrastructure Sensors. All Rights Reserved
Contact arbha@glue.umd.edu This page was last edited Monday, November 23, 2004 11:30:00 PM |