Parallel VLSI implementation of a new simplified architecture of Kalman filter

A new approach for implementing a parallel VLSI architecture of a simplified Kalman filter (information filter) with unknown noise statistics is described. At the system level, the proposed approach is applied to a semi-systolic VLSI array to accelerate the matrix operations required in each iteration. In order to maximize the parallelism, a new architecture called the tri-trapezoidal array for implementing the Schur complement algorithm is proposed. The computations steps are decomposed into 2(m+1) parts executable in parallel. This decomposition decreases the computation time by 55%. The overall architecture exhibits a high performance and can be used for real time applications.