Fixed-point implementation of a robust complex valued divider architecture

In this paper a fixed-point implementation of robust complex valued divider architecture is presented. The architecture uses feedback loops and time multiplexing strategies resulting in a fast and area conservative architecture. The architecture has good numerical properties and the result is accurate to less than one ulp. A combination of low latency and high throughput rate makes the architecture ideal for modern high speed signal processing applications. The complex valued divider architecture was implemented and tested on a Xilinx Virtex-II FPGA, clocked at 100MHz, and can easily be ported to an ASIC. The FPGA implementation is used as a core component in a matrix inversion implementation.