Time-critical problem-solving with cached knowledge: A case study in shortwave radio resource allocation

Successful communication between two places through shortwave radio depends on the right choice (sometimes within a very short period) of both frequency and power. This choice involves calling on mixtures of precise models and heuristics about prevailing geophysical conditions, and knowledge of specific past experiences. The design and the features of a knowledge-based system that is capable of answering these requirements for a given transmission request are described. There are two unusual constituents: a caching structure that can accommodate different knowledge representations and time limits within which answers must be completed, and interpolation between items of knowledge in different cells of a cache. Results are demonstrated for simple but realistic radio propagation exercises, to illustrate the effectiveness of caching and interpolation for time-critical knowledge-based computations.