Design and Implementation of an In?Cache Archival Entropy Coder for the Sonoma Persistent Surveillance System

Senecal, J G

关键词: CAMERAS;    COMPRESSION;    CONFIGURATION;    DESIGN;    EFFICIENCY;    ENTROPY;    IMPLEMENTATION;    LAWRENCE LIVERMORE NATIONAL LABORATORY;    STORAGE;    VELOCITY;   

DOI  :  10.2172/945859
RP-ID  :  LLNL-TR-407448
PID  :  OSTI ID: 945859
Others  :  TRN: US200904%%176

学科分类：社会科学、人文和艺术（综合）

美国|英语

来源: SciTech Connect

【 摘 要 】

The Sonoma sensor, developed at Lawrence Livermore National Laboratory, is an electro-optical (EO) sensor composed of a tiled arrangement of cameras, where each camera sends 20 megabytes (MB) of raw data per frame. Depending on the sensor configuration, each sensor image, in aggregate, can total up to 320 MB. I discuss here the design and implementation of a high-speed entropy coder utilizing the Lead-1 encoding method, intended to quickly and losslessly compress sensor data before it is written to storage. More sophisticated (and slower) compression techniques may be performed off-line. The coder, favoring speed over coding efficiency, is simple in principle and tiny in practice: the amount of allocated memory required by the coder can be as small as 26 bytes. On a 2.7 GHz PowerMac G5 this coder is capable of encoding sensor data at a rate of 163 megabytes per second. The overall compression rate is better than that of BZIP2, GZIP, and a range coder, and the total execution speed is superior.