150 likes | 169 Views
This project focuses on research related to advanced media systems, including ubiquitous capture, access, and interpretation of media content. The project involves faculty from the College of Computing at Georgia Tech and is funded by federal agencies such as NSF, DARPA, and DOE, as well as industry partners including HP, Intel, Microsoft, and IBM.
E N D
Advanced media-oriented systems research: Ubiquitous capture, access, and interpretation • Faculty involved with RI-related projects • Kishore Ramachandran, Mustaque Ahamad, Karsten Schwan, Richard Fujimoto, Ken Mackenzie, Sudha Yalamanchili, Irfan Essa, Jim Rehg, Gregory Abowd, Yannis Smaragdakis, Santosh Pande, Calton Pu, Ling Liu, ... • Federal funding • NSF RI, NSF ITR, DARPA, DOE • State funding • Yamacraw, GT Broadband Institute • Industry funding (equipment and personnel) • HP, Intel, Microsoft, IBM
MediaBroker and DFuse:Powerful support for emerging applications Kishore Ramachandran Jim Rehg, Phil Hutto, Ken Mackenzie, Irfan Essa College of Computing, Georgia Tech Kath Knobe, Jamey Hicks, Raj Kumar HP Labs Students: Rajnish Kumar, Matt Wolenetz, Ilya Bagrak, Martin Modahl, Bikash Agarwalla, Junsuk Shin, Arnab Paul, Sameer Adhikari, Nissim Harel, Hasnain Mandviwala, Yavor Angelov, David Hilley
Skiff Skiff • Distributed system components • sensors, actuators, embedded processors, PDAs, laptops, clusters… camera camera Sensor Fusion Sensors Actuators Unix / Linux / NT cluster Data Aggregators “OCTOPUS” DIAGRAM head / arms / tentacles
Killer App? • Application context • distributed sensors with varying capabilities • control loop involving sensors, actuators • rapid response time at computational perception speeds
Emerging Applications • Distributed Collaboration • Emergency Response • Collaborative search and rescue • Evacuation management • “Aware” Environments Key Requirements • Data distribution and Infrastructure adaptation
Application Characteristics • Physically distributed heterogeneous devices • Interfacing and integrating with the physical environment • Diverse stream types (low to high BW) • Diverse computation, communication and power capabilities (from embedded sensors to clusters) • Stream fusion/transformation, with loadable code • Resource scarcities • Dynamic join/leave of application components Two projects: • MediaBroker • DFuse
A Fusion Channel (a ‘Virtual Sensor’) Inputs (sensors or other fusion channels) Consumers (actuators or other fusion channels) F() . . . . . . Abstractions…abstractions • Goals: • Temporal guarantees, efficient streams, distributed synchronization, fusion support • Simplify application development • Key result: Fusion Channels • built on top of D-Stampede [ICDCS 2002]
The MediaBroker • An architecture for Data Distribution • Type management, Scalability, Stream “sharing” • A sample scenario for distributed collaboration • Physically distributed participants in a classroom setting • Interaction via a variety of devices • System takes care of any needed data transformations and distribution MediaBroker Infrastructure
audio video sonar bio fusion derivation sharing registry Re-publish discovery D-Stampede virtual sensors applications MediaBroker Infrastructure( ) • Fusion Channels and Virtual Sensors • Resource Discovery and Sharing • Sensor-Provided Attributes • Sensor-Derived Attributes • Scalability • Access Control and Resource Scheduling • Security
MediaBroker and Typed Streams • Experimental “type-lattice” registry, discovery, and resolution structure Fusion Channel Performs text-to-speech Fusion Channel Outputs A/V as MPEG, Incorporates text as caption Today’s Lecture:
DFuse [ACM SenSys 2003] • An architecture for Infrastructure Adaptation • A sample scenario for an aware environment • Field trip for a class! • Deployed power-constrained sensors • Dynamic wireless network consisting of the students’ PDAs • In-network stream filtering and aggregation
Sensors Collage Sink (Display) Filter Sample surveillance application task graph: filter and collage are the fusion functions. Cameras • Deployed on iPAQ farm! • Tested 4 cost-functions DFuse Fundamentals • Fusion Module: Deploys task graph on sensor network • Comprehensive API for fusion and migration • Low-overhead • Placement Module: Employs a self-stabilizing algorithm to place fusion points in the network • Energy and application aware cost functions • Localized decisions