Share Email Print

Proceedings Paper

Performance estimates of radar STAP processing on the IBM/Sony/Toshiba cell processor
Author(s): Luke Cico; Jon Greene
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Advances in commercial off-the-shelf (COTS) embedded computing technologies have yielded impressive gains in computational throughput over the past 5-10 years. Adaptive sensor array systems utilizing real-time teraflop class machines are in wide deployment today. Gains in processor density have generally been achieved by steady improvements in semiconductor optical lithographic processes along with less frequent innovations in processor chip architectures. It is likely that the real-time embedded community is entering an era where processor architectural innovations will be bearing most of the burden for producing processing density gains as lithographic processes approach fundamental physical limitations. More and more 'systems on a chip' are emerging to address these trends. An exciting example of this technology trend is IBM's new Cell Broadband Engine (CBE) architecture which offers massive SIMD compute power on multiple computational units interconnected via a high bandwidth internal fabric. This paper explores the application of a computationally intensive adaptive nulling problem on the CBE architecture.

Paper Details

Date Published: 8 May 2006
PDF: 6 pages
Proc. SPIE 6210, Radar Sensor Technology X, 62100D (8 May 2006); doi: 10.1117/12.666216
Show Author Affiliations
Luke Cico, Mercury Computer Systems, Inc. (United States)
Jon Greene, Mercury Computer Systems, Inc. (United States)

Published in SPIE Proceedings Vol. 6210:
Radar Sensor Technology X
Robert N. Trebits; James L. Kurtz, Editor(s)

© SPIE. Terms of Use
Back to Top