Search

The paper considers a family of linear time-invariant and spatially invariant (LTSI) systems that are both distributed and localized. The spatial responses of the distributed plant are localized in spatial neighborhoods of each location. The feedback computations are also distributed and the information flow is localized in a spatial neighborhood of each location. The feedback is aimed at controlling spatial distributions of variables in the systems with a relatively low bandwidth in the time direction. Such systems have many important applications including industrial processes, imaging systems, signal and image processing, and others. We describe a new method for designing (tuning) a certain family of low-bandwidth controllers for such plants. We consider LTSI controllers with a fixed structure, which is a PID or a similar low-bandwidth feedback in time and local in spatial coordinates. Two spatial feedback filters, symmetric and with finite spatial response, modify the local PID control signal by mixing in the error and control signals at nearby nodes. These two filters provide loopshaping and regularization of the spatial feedback loop. Like an ordinary PID controller, this controller structure is simple, but provides adequate performance in many practical settings. We cast a variety of specifications on the steady-state spatial response of the controller and its time response as a set of linear inequalities on the design variables, and so can carry out the design of the spatial filters using linear programming. The method handles steady-state limits on actuator signals, error signals, and several constraints related to robustness to plant and controller variation. The method allows handling the effects of boundary conditions and guaranteed closed-loop spatial or time decay. It does appear to work very well for low-bandwidth controllers, and so is applicable in a variety of practical situations. Index Terms--Bandwidth, distributed system control, loopshaping, multidimensional, optimization, PID.

In this paper we examine a technique by which fault tolerance can be embedded into a feedforward network leading to a network tolerant to the loss of a node and its associated weights. The fault tolerance problem for a feedforward network is formulated as a constrained minimax optimization problem. Two different methods are used to solve it. In the first method, the constrained minimax optimization problem is converted to a sequence of unconstrained least-squares optimization problems, whose solutions converge to the solution of the original minimax problem. An efficient gradient-based minimization technique, specially tailored for nonlinear least-squares optimization, is then applied to perform the unconstrained minimization at each step of the sequence. Several modifications are made to the basic algorithm to improve its speed of convergence. In the second method a different approach is used to convert the problem to a single unconstrained minimization problem whose solution very nearly equals that of the original minimax problem. Networks synthesized using these methods, though not always fault tolerant, exhibit an acceptable degree of partial fault tolerance. Index Terms - Fault tolerance, minimax optimization.

RedZone Robotics Inc's low-level distributed controller, called Distributed Control System, is a flexible intelligent system that controls physical tasks. Originally intended for intelligent military vehicles, the system can incorporate multiple layers of intelligence to permit a wide range of applications. Its architecture provides optimization, robustness, economy and versatility for factory automation and other industrial processes. Future versions of Distributed Control System could include video switching and digital signal processing capabilities.

A compact neural network algorithm for partial-response maximum-likelihood (PRML) sequence detection is presented. Compact neural networks are a class of locally connected neural networks suitable for very large scale integration (VLSI) implementation. The hardware complexity for VLSI implementation of the proposed algorithm grows linearly with the level of the deliberately designed symbol interference effects of the partial-response (PR) signaling scheme. Large dedicated memory for storage of likelihood matrices in digital Viterbi-algorithm-based detectors is not needed for the proposed detector. Detailed analysis on network stability for network topology and time constant of an analog neuron is described. This detector algorithm has competitive bit-error rate performance when compared with the digital Viterbi algorithm under the noise condition for many real-world applications. The proposed algorithm is suitable for analog VLSI implementation because of its low time complexity and linear area complexity for the detection of PRML signaling schemes. Index Terms - Communication, distributed parallel processing, maximum-likelihood sequence estimation, neural networks, partial response.