On restoring data coherence in a GALS system for medical imaging

Abstract

Globally Asynchronous, Locally Synchronous (GALS) systems are widely used in medical applications. Typically, they exhibit high number of node-to-node communication channels that must transmit high data volume at high data rates and maintain data time coherence. The purpose of this paper is to present two structures for restoring data coherence in node-to-node communication, which may be lost due to different (unknown) delays introduced by the various communication channels. Both solutions allow differences among the channel delays larger than one clock cycle. Comparison of their most relevant characteristics is provided. Based on this comparison, one of the structures has been implemented to guarantee synchronism in a medical imaging system, namely in the Data Acquisition Electronics (DAE) of a PEM (Positron Emission Mammography) system. The DAE is a multi-board, multiple-FPGA platform (9 Xilin™ xc2v4000-4bf957 FPGA). The effectiveness of the solution has been validated in a hardware prototype. Test results are presented.