Adversarial reasoning and resource allocation: the LG approach

Many existing automated tools purporting to model the intelligent enemy utilize a fixed battle plan for the enemy while using flexible decisions of human players for the friendly side. According to the Naval Studies Board, "It is an open secret and a point of distress ... that too much of the substantive content of such M&S has its origin in anecdote, ..., or a narrow construction tied to stereotypical current practices of 'doctrinally correct behavior.'" Clearly, such runs lack objectivity by being heavily skewed in favor of the friendly forces. Presently, the military branches employ a variety of game-based simulators and synthetic environments, with manual (i.e., user-based) decision-making, for training and other purposes. However, without an ability to automatically generate the best strategies, tactics, and COA, the games serve mostly to display the current situation rather than form a basis for automated decision-making and effective training. We solve the problem of adversarial reasoning as a gaming problem employing Linguistic Geometry (LG), a new type of game theory demonstrating significant increase in size in gaming problems solvable in real and near-real time. It appears to be a viable approach for solving such practical problems as mission planning and battle management. Essentially, LG may be structured into two layers: game construction and game solving. Game construction includes construction of a game called an LG hypergame based on a hierarchy of Abstract Board Games (ABG). Game solving includes resource allocation for constructing an advantageous initial game state and strategy generation to reach a desirable final game state in the course of the game.