S. Daijavad

Efficient optimization with integrated gradient approximations

A flexible and effective algorithm is proposed for efficient optimization with integrated gradient approximations. It combines the techniques of perturbations, the Broyden update, and the special iterations of Powell. Perturbations are used to provide an initial approximation as well as regular corrections. The approximate gradient is updated using C.G. Broydens formula (1965) in conjunction with the special iterations of M.J.D. Powell (1970). A modification to the Broyden update is introduced to exploit possible sparsity of the Jacobian. Utilizing this algorithm, powerful gradient-based nonlinear optimization tools for circuit CAD can be used without the effort of calculating exact derivatives Applications of practical significance are demonstrated. The examples include robust small-signal FET modeling using the l/sub 1/ techniques and simultaneous processing of multiple circuits, worst-case design of a microwave amplifier, and minimax optimization of a five-channel manifold multiplexer. Computational efficiency is greatly improved over estimating derivatives entirely by perturbations. >

Exact Simulation and Sensitivity Analysis of Multiplexing Networks

This paper presents a novel approach to the simulation and sensitivity analysis of multiplexing networks. All computations are performed efficiently utilizing the concept of forward and reverse analysis which is elegant and effective in cascaded circuit analysis. Formulas are derived for such responses as input or output reflection coefficient, common port and channel output port return losses, insertion loss, gain slope, and group delay. Exact sensitivities w.r.t. all variables of interest, including frequency, are evaluated. The fundamental assumption is that the transmission matrices for the individual components of the network and their sensitivities w.r.t. possible variables inside them are available. An explicit algorithm is provided describing the details of the computational aspects of our theory. The formulas are applied to the optimal design of practical contiguous or noncontiguous band multiplexer consisting of multicavity filters distributed along a waveguide manifold. An example of optimizing a practical 12 channel, 12-GHz contiguous band multiplexer without dummy channels, which is the state-of-the-art structure used as the output multiplexer in satellite transponders, is presented.

Large Scale Minimax Optimization of Microwave Multiplexers

A new technique for contiguous-band multiplexer design involving an arbitrarily large number of channels and design parameters is described. The technique, based on network decomposition and sensitivity measures of practical microwave multiplexer structures, employs a sequence of relatively small optimization problems which correspond to growing a multiplexer by adding one or more channels at a time to the structure. The approach is illustrated by a 16-channel, 12 GHz multiplexer design involving 240 nonlinear design variables.

Optimal Design of Multi-Cavity Filters and Contiguous-Band Multiplexers

A general contiguous-band multiplexer optimization procedure exploiting exact network sensitivities is described. The structure under consideration consists of synchronously and asynchronously tuned multi-cavity filters distributed along a waveguide manifold. All design parameters, e.g., waveguide spacings (section lengths), input-output and filter coupling parameters, can be directly optimized using a powerful gradient-based minimax algorithm. Nonideal effects such as junction susceptances, dissipation and dispersive effects are readily taken into account.

Case Study in Passive Circuit CAD

This tutorial paper reviews some fundamental principles in applying modern optimization techniques to passive microwave circuit design. Nonlinear minimax optimization is stressed, the exploitation of exact partial derivatives of circuit models is used and justified and a suitable case study is presented. The case study is the design of coupled-cavity, narrow-band filters, the results for which are obtained by a powerful interactive optimization package.

Microwave Device Modelling Using Efficient l/sub 1/ Optimization: A Novel Approach

A powerful modelling technique which exploits the unique properties of the l/sub 1/ norm is presented. Self-consistent models for passive and active devices are achieved by an approach that automatically checks the validity of model parameters obtained from optimization. Practical use of an efficient l/sub 1/ algorithm in complicated problems for which gradient evaluation may not be feasible, is discussed, Examples in modelling of multi-coupled cavity filters and GaAs FETs are presented.

Simple derivation of a general sensitivity formula for lossless two-ports

A simple, yet comprehensive proof of an important sensitivity formula for lossless two-ports stated by Orchard, Temes, and Cataltepe is presented. Our derivation invokes the principle of conservation of energy and the lossless property of the network under consideration and employs the Cauchy-Riemann equations of complex differentiation. Hence, it bears clear physical interpretation and mathematical elegance.