Intelligent Automation, Inc. — Department of Defense SBIR Phase I: MDA17-004

Description

Military applications require data logging in different formats. More effective data compression techniques compatible with existing infrastructure to reduce data logging path bandwidth and decrease memory and storage are actively sought. To address this need, I-A-I and our partners at the University of California, Davis will work on the design and implementation of a highly parallel multi format suitable compression software suite, accelerated via multi-core and GPU architectures. One particular goal of this project will be the design of a fast general purpose bzip2 like compressor for the GPU, centered around a highly parallel implementation of the Burrows-Wheeler transform. This will enable the user to achieve fast high compression results for different data formats. One advantage of a GPU based algorithm is also the ability to add and upgrade GPU cards at low cost, without making significant changes to existing infrastructure, as per stated goals. Apart from bzip2, I-A-I will work on creating high performance adaptive arithmetic coding and efficient xml interchange implementations, also accelerated on multi-core and GPU architectures. For certain formats like text, XML, and coordinate data, it will be possible to use a combination of the algorithms (e.g. arithmetic coding and bzip2) to produce favorable compression ratios.Approved for Public Release | 17-MDA-9395(24 Oct 17)