SBIR-STTR Award

Techniques of artificial intelligence will be used to design a robot jammer which can devise an optimum jamming strategy. A model of the universe in which the jammer will operate will be developed. This model will include the various forms of activity that can occur within the adversary system, actions that the jammer can take, and modes of operation defining the purpose of the jamming. The jammer's knowledge of adversary activity is derived by intelligently interpreting the electrical signal environment measured by its sensors. The interactions between the adversary system and the jammer will be modeled, using a game theoretical approach. Once a suitable game has been selected, a computer program will attempt to devise an optimum strategy for playing the game. This program will take advantage of the technology developed in artificial intelligence for playing games of incomplete information. Actions will be directed by a production system which makes decisions based upon the values of key variables which make up a dynamic state vector. This production system will be self-modifiable so that the system can learn from its past experience to improve its performance.

Using the (AI)2, artificial intelligence and artificial instinct, architecture from Phase I coupled with an interactive graphics capability, an expert system will be developed for pre-engagement planning and for monitoring and re-planning during actual operations. The output of the system will be an EW estimate including the generation of time-phased graphics overlays of the eeob with emphasis on EW activities, a verbal summary of the various possible courses of action, recommdnations as to which course of action is easiest to support from an EW standpoint, and possibly load parameters for the (semi) autonomous robot jammes. The AI computations will use knowledge bases like terrain maps (dma dlms), friendly EW capabilities, enemy tactical, communication and EW doctrine, enemy communication equipment and communication net structures. Other ad hoc inputs will include sigint reports and weather forecasts. The software structure will include rules for placing radios and ECM equipment in the terrain based on the expected movement of forces and propagation characteristics, taking into account terrain elevation, ground cover and weather. Values will be assigned to both enemy and friendly assets to establish numerical payoffs and numerical risk factors for all actions. The proposed hardware will consist of a computer capable of executing (AI)2, arithmetic and graphics routines; a high resolution color graphics display; and possible a digital plotter as well as a communication interface to transmit EW estimates, maps and plot overlays and for downloading programming parameters to robot jammers.