Implementation Issues

Abstract

In order to determine if a proposed approach may be useful in real-world applications, it is necessary to understand exactly how the algorithm will perform in the situations it may encounter. A best-case estimate of this performance is given by a theoretical analysis of the approach, assuming ideal conditions. The actual performance of the technique, however, is dependent on how robust the approach is with respect to deviations from ideal conditions. This robustness can be measured experimentally, or predicted by identifying the limitations imposed by real-world implementation. Of course, it is not possible to predict a priori every possible scenario which an approach may be confronted with. Therefore the experimental approach is, at best, incomplete. However, it is possible to predict the performance of the approach based on an understanding of the limitations of the equipment used for implementation and the non-ideal nature of the environment in which the technique is to operate.