A modular distributed arithmetic implementation of inner product including quantization

The authors show that no double-precision inner product computation is needed to implement digital filters free from limit cycles if distributed arithmetic is employed. It is also shown that some inexpensive extra circuitry is needed to implement the appropriate signal quantization and overflow nonlinearity such that no nonlinear oscillation is sustained. The main result presented is the development of a modular implementation for inner products with several types of granular quantization and saturation arithmetic for overflow nonlinearity