DESIGN OF 2-DIMENSIONAL SCALED STATE-SPACE FILTERS WITH FIXED-POINT ARITHMETIC

In this paper a new synthesis method has been proposed for the design of 2D scaled state space filters with fixed point arithmetic. It is ba sed on constructing a non-singular transformation matrix which minimiz es the output noise power and maintains the signal-scaling condition, besides zeroing some of the entries of the state matrices. As a result , simultaneous savings in both additions and multiplications are achie ved. Numerical results are included to illustrate the performance of t he proposed structures and to compare them with other structures.