A polynomial eigenvalue decomposition of paraher-mitian matrices can be calculated approximately using iterative approaches such as the sequential matrix diagonalisation (SMD) algorithm. In this paper, we present an improved SMD algorithm which, compared to existing approaches, eliminates more off-diagonal energy per step. This leads faster convergence while incurring only a marginal increase in complexity. We motivate approach, prove its convergence, and demonstrate some results that...

This paper describes the performance improvements of a reduced instruction set computer (RISC) microprocessor that has migrated from 2.5 V technology to 1.8 technology. The implements copper interconnects and low V/sub t/ field-effect transistors in speed-critical paths an L/sub eff/ 0.12 /spl mu/m. Global clock latency skew are improved by using wires, early mode timings reducing adding buffers. These enhancements, along with environment 2.0 V, 85/spl deg/C, fast process, produced 480-MHz...

In this paper, we show that the paraunitary (PU) matrices arise from polynomial eigenvalue decomposition (PEVD) of a parahermitian matrix are not unique. particular, arbitrary shifts (delays) polynomials in one row PU yield another admits same PEVD. To keep order such as low possible, propose row-shift correction. Using example an iterative PEVD algorithm with previously proposed truncation matrix, demonstrate considerable shortening can be accomplished when using row-corrected truncation.

In this paper, we present an improved version of the second order sequential best rotation algorithm (SBR2) for polynomial matrix eigenvalue decomposition para-Hermitian matrices. The is entitled multiple shift SBR2 (MS-SBR2) which developed based on original algorithm. It can achieve faster convergence than by means transferring more off-diagonal energy onto diagonal at each iteration. Its proved and also demonstrated a numerical example. Furthermore, simulation results are included to...

This paper extends the analysis of recently introduced row- shift corrected truncation method for paraunitary matrices to those produced by state-of-the-art sequential matrix diagonalisation (SMD) family polynomial eigenvalue decomposition (PEVD) algorithms. The row-shift utilises ambiguity in reduce their order. results presented this compare effect a simple change PEVD can have on performance truncation. In case SMD algorithm benefits new approach are reduced compared what has been seen...

A RISC microprocessor remapped in SOI technology exploits the advantages of to boost processor frequency by 20% 580MHz at 2.0V and 85/spl deg/C fast process. The separation implanted oxygen (SIMOX) process produces partially-depleted devices. Source drain capacitances are reduced an order magnitude, improving gate delay 12%. Reduction body-bias effects on device stacks passgate topologies results additional 15%-25% improvement. Speed gains up 35% achieved some designs. frequency-limiting...

A number of algorithms for the iterative calculation a polynomial matrix eigenvalue decomposition (PEVD) have been introduced. The PEVD is generalisation ordinary EVD and will diagonalise parahermitian via paraunitary operations. This paper addresses savings - both computationally in terms memory use that exploit structure being decomposed, also suggests an implicit trimming approach to efficiently curb order growth usually observed during iterations algorithms. We demonstrate with proposed...

This paper expands on a recent polynomial matrix formulation for minimum variance distortionless response (MVDR) broadband beamformer.Within the framework, this beamformer is straightforward extension from narrowband case, and offers advantages in terms of complexity robustness particularly off-broadside constraints.Here, we focus arbitrary 3-dimensional array configurations no particular structure, where incorporation constraints demonstrated simulations, accurately maintains its look...

In this paper we study the impact of polynomial or broadband subspace decompositions on any subsequent processing, which here uses example a angle arrival estimation technique using recently proposed MUSIC (P-MUSIC) algorithm. The are performed by iterative EVDs, differ in their approximations to diagonalise and spectrally majorise s apce-time covariance matrix.We show that better diagonalisation has significant accuracy defining signal noise subspaces, demonstrated much higher P-MUSIC spectrum.

A number of algorithms capable iteratively calculating a polynomial matrix eigenvalue decomposition (PEVD) have been introduced. The PEVD is generalisation the ordinary EVD and will diagonalise parahermitian via paraunitary operations. Inspired by existence low complexity divide-and-conquer solutions to eigenproblems, this paper addresses approach utilising sequential diagonalisation (SMD) algorithm. We demonstrate that with proposed techniques, encapsulated in novel algorithm titled...

Polynomial parahermitian matrices can accurately and elegantly capture the space-time covariance in broadband array problems.To factorise such matrices, a number of polynomial EVD (PEVD) algorithms have been suggested.At every step, these move various amounts off-diagonal energy onto diagonal, to eventually reach an approximate diagonalisation.In practical experiments, we found that relative performance depends quite significantly on type matrix is be factorised.This paper aims explore this...

A recent class of sequential matrix diagonalisation (SMD) algorithms have been demonstrated to provide a fast converging solution iteratively approximating the polynomial eigenvalue decomposition parahermitian matrix. However, calculation an EVD, and application full unitary every time lag in SMD algorithm results high numerical cost. In this paper, we replace EVD with limited number Givens rotations forming cyclic-by-row Jacobi sweep. Simulations indicate that considerable reduction...

Sequential matrix diagonalisation (SMD) refers to a family of algorithms iteratively approximate polynomial eigenvalue decomposition. Key is transfer as much energy possible from off-diagonal elements the diagonal per iteration, which has led fast converging SMD versions involving judicious shifts within matrix. Through an exhaustive search, this paper determines optimum shift in terms transfer. Though costly implement, scheme yields important benchmark limited search strategies can be...

In recent years, several algorithms for the iterative calculation of a polynomial matrix eigenvalue decomposition (PEVD) have been introduced. The PEVD is generalisation ordinary EVD and uses paraunitary operations to diagonalise parahermitian matrix. This paper addresses potential computational savings that can be applied existing cyclic-by-row approaches PEVD. These are found during search rotation stages, do not significantly impact on algorithm accuracy. We demonstrate with proposed...

The Multiple Shift Maximum Element Sequential Matrix Diagonalisation (MSME-SMD) algorithm is a powerful but costly method for performing approximate polynomial eigenvalue decomposition (PEVD) space-time covariance-type matrices encountered in e.g.broadband array processing.This paper discusses newly developed search that restricts the order growth within MSME-SMD algorithm.In addition to enhanced control of degree paraunitary and parahermitian factors this decomposition, new also...

In this work we present a new method of controlling the order growth polynomial matrices in multiple shift second sequential best rotation (MS-SBR2) algorithm which has been recently proposed by authors for calculating matrix eigenvalue decomposition (PEVD) para-Hermitian matrices. effect, introduces elementary delay strategy keeps all row (column) shifts same direction throughout each iteration, therefore gives us flexibility to control selecting that ensure non-zero coefficients are kept...

State-of-the-art narrowband noise cancellation techniques utilise the generalised eigenvalue decomposition (GEVD) for multi-channel Wiener filtering, which can be applied to independent frequency bins in order achieve broadband processing. Here we investigate extension of GEVD broadband, polynomial matrices, akin strategies that have already been developed by McWhirter et. al on matrix (PEVD). In our approach extend Cholesky method calculating scalar matrices. this paper outline...

Recently a selection of sequential matrix diagonalisation (SMD) algorithms have been introduced which approximate polynomial eigenvalue decomposition parahermitian matrices. These variants differ only in the search methods that are used to bring energy onto zero-lag. Here we analyse terms their computational complexities for different sizes matrices verified through simulated execution times. Another important factor these is ability transfer energy. Simulations show more computationally...

This paper describes the boundary-scan and built-in self-test (BIST) functions of IBM token-ring local area network (LAN) adapter chip. These present a number unique features. First, less that 1% available standard cell circuits were needed to implement these functions. Second, clocking methods used in different logical macros merged into comprehensive sequence for self-test. Finally, asynchronous serial parallel interfaces provided facilitate communication between test system chip's...

Fractals have been proven as potential candidates for satellite flying formations, where its different elements represent a thinned array.The distributed and low power nature of the nodes in this network motivates processing when using such an array beamformer.This paper proposes initial idea, demonstrates that benefits strictly limited local capability independent array's dimension calibration can be bought at expense slightly increased overall cost.