University of California, Santa Barbara, CA, USA
This talk presents a trustworthy information distribution and retrieval strategy, without centralized mechanisms or control, as an alternative to the conventional centralized search engine approach. Source nodes produce information and metadata, and distribute the metadata to participating nodes chosen at random. Requesting nodes retrieve information, by generating requests and distributing them to participating nodes chosen at random. The nodes that receive the requests compare the metadata in the requests with the metadata that they hold. If such a node finds a match, it returns the URL of the information to the requesting node. Distributing metadata and requests to the square root of the number of participating nodes suffices to achieve a high probability of retrieval. This distributed strategy is robust against unreliable nodes and against malicious attacks. Most importantly, unlike centralized search engines, the distributed strategy prevents censorship of information.
Michael Melliar-Smith is a professor in the Department of Electrical and Computer Engineering at the University of California, Santa Barbara. His research interests span the areas of distributed systems, communication networks and protocols, and fault tolerance. He has served as PI for numerous funded research projects, including projects from DARPA, AFOSR, NSF, UC Micro and UC Discovery. These funded research projects include the DARPA funded Thunder and Lightning project to develop a 30 Gbit/s ATM switch and optical transmission system, and the recent AFOSR MURI Protocol Engineering Research Center project, which involved six universities. He has authored or coauthored more than 240 publications, has more than 10 patents granted or pending , and he has supervised more than 100 graduate student researchers. Prior to UCSB, at GEC Computers in England, Dr. Melliar-Smith was principal designer of the GEC 4080, which won the Queen's Award for Innovation. At the University of Newcastle upon Tyne, he invented the definitions of fault, error and failure, as well as the recovery block method for software fault tolerance. As Senior Computer Scientist and Program Director at Stanford Research Institute, he was involved in the design of the Software-Implemented Fault-Tolerant(SIFT) aircraft flight control computer and was leader of the NASA funded Enhanced Hierarchical Design Methodology (EHDM) project for formal specification and verification.