DATA CENTRIC SIR PARTICLE FILTER DESIGN USING BUFFER-LEVEL PIPELINING

摘要

In this paper, we introduce use of sample importance resampling (SIR) particle filter for bearings-only tracking problem using buffer-level pipelining. The SIR particle filter is heavily data centric in that it transfer significant amount of data between processing bloeks. We propose to design the overall filter using buffer-level pipelining where design flow can be significantly simplified. In the buffer-level pipelining, overall execution synchronization is done through controlling patterns of data producing and consuming by each processing block. We verify temporal operation of the filter using Verilog and System C.