Inference techniques are applied to computing systems to improve the allocation of resources for fault tolerant performance. Using a general model for such systems, the influence of estimation errors for system parameters on the resulting system fault performance is examined. These results are then applied to the problem of error mode testing-finding the underlying error structure of the system. Simulation is used to illustrate the properties discussed.

Abstract A random Petri net model is presented and analyzed. The model has application to systems involving concurrency.

Graphical techniques are presented for estimating Hough transform performance for detecting straight lines in noisy digital pictures. The methods show explicitly the influence of noise due, for example, to quantizing errors on the accuracy of estimating the underlying sets of collinear points. Exact bounds are obtained for sets of collinear points showing that under quite general conditions only the endpoints are of significance. Theorems governing grid size for the accumulator method are given.

Procedures are presented for reducing page faults by program restructuring and utilization of bin packing techniques. A Markov model is presented and used as a basis for the methods presented. Parameter estimation of appropriate software statistics for the model is given. The techniques described are applicable to completed programs as weli as those under development. Experimental results are included.

The transform method for curve detection is applied to curves against a textured background. The objective is maximum scattering of the transformed texture points while preserving clustering properties for the sought after curve. Properties of the extended transform are examined under a statistical model of "grass-like" texture.

A systematic method of providing software system fault recovery with maximal fault coverage subject to resource constraints of overall recovery cost and additional fault rate is presented. This method is based on a model for software systems which provides a measure of the fault coverage properties of the system in the presence of computer hardware faults. Techniques for system parameter measurements are given. An optimization problem results which is a doubly-constrained 0,1 Knapsack problem. Quantitative results are presented demonstrating the effectiveness of the approach.