Algorithm For Toeplitz Systems Research Articles

Pipeline architecture for block adaptive LS FIR filtering and prediction

Consideration is given to the development of a highly parallel, block-type, order-recursive algorithm for least-squares finite-impulse-response (LS FIR) filter identification and prediction. The computation of the required reflection coefficients is achieved by a technique that does not involve inner vector product operations. Its origin can be traced to the classical Schur algorithm for Toeplitz systems. The algorithm derived is simple and can be implemented on a linear array of O(p) modular processing elements with localized communication requirements, where p is the order of the system, which makes it suitable for VLSI implementation. In a parallel processing environment, the algorithm can be completed in O(p)+O(L) time units, where L is the length of the data block. A novel mode of operation that computes the reflection coefficients as well as the unknown filter's impulse response concurrently and on the same linear processor array is proposed. The algorithm is suitable either for processing a single block of data or for a block adaptive mode of operation, and it can track variations from block to block. The use of the algorithm for parameter adaptation on each time instant is discussed. >

Fast solution of toeplitz systems of equations and computation of Padé approximants

We present two new algorithms, ADT and MDT, for solving order- n Toeplitz systems of linear equations Tz = b in time O( n log 2 n) and space O( n). The fastest algorithms previously known, such as Trench's algorithm, require time Ω(n 2) and require that all principal submatrices of T be nonsingular. Our algorithm ADT requires only that T be nonsingular. Both our algorithms for Toeplitz systems are derived from algorithms for computing entries in the Padé table for a given power series. We prove that entries in the Padé table can be computed by the Extended Euclidean Algorithm. We describe an algorithm EMGCD (Extended Middle Greatest Common Divisor) which is faster than the algorithm HGCD of Aho, Hopcroft and Ullman, although both require time O( n log 2 n), and we generalize EMGCD to produce PRSDC (Polynomial Remainder Sequence Divide and Conquer) which produces any iterate in the PRS, not just the middle term, in time O( n log 2 n). Applying PRSDC to the polynomials U 0( x) = x 2 n+1 and U 1( x) = a 0 + a 1 x + … + a 2 n x 2 n gives algorithm AD (Anti-Diagonal), which computes any ( m, p) entry along the antidiagonal m + p = 2 n of the Padé table for U 1 in time O( n log 2 n). Our other algorithm, MD (Main-Diagonal), computes any diagonal entry ( n, n) in the Padé table for a normal power series, also in time O( n log 2 n). MD is related to Schönhage's fast continued fraction algorithm. A Toeplitz matrix T is naturally associated with U 1, and the ( n, n) Padé approximation to U 1 gives the first column of T −1. We show how a formula due to Trench can be used to compute the solution z of Tz = b in time O( n log n) from the first row and column of T −1. Thus, the Padé table algorithms AD and MD give O( n log 2 n) Toeplitz algorithms ADT and MDT. Trench's formula breaks down in certain degenerate cases, but in such cases a companion formula, the discrete analog of the Christoffel-Darboux formula, is valid and may be used to compute z in time O( n log 2 n) via the fast computation (by algorithm AD) of at most four Padé approximants. We also apply our results to obtain new complexity bounds for the solution of banded Toeplitz systems and for BCH decoding via Berlekamp's algorithm.