About Centar
Centar is a provider of intellectual property (IP) for use in FPGA and ASIC-based embedded applications. Its focus is on implementation of scalable high performance signal processing algorithms, primarily for wireless applications.
Tools
Because there are always a large number of (parallel) circuit architectures that can be obtained from an algorithm specification, Centar has developed an automated CAD tool, Symbolic Parallel Algorithm Development Environment (SPADE), to make the best choices.
SPADE is the only such tool in existence that can find latency optimal circuits. It does this by looking at all possible schedules, reindexings, localizations and allocations. It has been used to design fast Fourier transform and linear algebra circuits.
Fast Fourier Transform (FFT)
Centar has developed a radically different approach to parallel FFT implementation based on a new matrix formulation of the discreet Fourier transform (DFT), which decomposes it into structured sets of multiplication-free 4-point DFTs.
The result is a programmable circuit that provides higher performance, more functionality, lower power, and better design simplicity than conventional pipelined FFTs. For example, it is capable of computing non-power-of-two transform lengths, dynamically reducing computation time in proportion to data and any FFT circuit can peform any size transform given adequate memory resources.
A 16-bit 1024-point streaming FFT circuit with an 86.9 signal-to-quantization-noise ratio has been demonstrated on an FPGA that achieves a complex sample rate of 600MHz at 6.2µJ per transform.
Linear Algebra
Centar has designed a special circuit capable of executing a variety of very different linear algebraic operations yet still provide scalable high performance. Such flexibility and speed is important in a variety of embedded applications, as for example in advanced wireless communications where it is seen in estimation, multiple-input-multiple-output antenna protocols, multi-user detection, equalization and adaptive antennas.