A Genetic Algorithm Approach for Deriving Direction Finding Antenna Placement on Platforms

Placing a Direction Finding (DF) array onto an existing aircraft is typically a difficult endeavor due to the limitations placed by existing antennas or structures which mandate “keep out” areas, the additional infrastructure required for potentially dispersed DF antennas, and getting all of the modifications for the DF antennas flight certified. Because of these challenges, along with the basic expense of modifying an aircraft for external antennas, the ability to optimize the antenna lay down for peak DF performance is absolutely essential. This paper will describe the use of a Genetic Algorithm in the application of defining antenna locations on a platform to form an optimized broadband Direction Finding array. For this optimization study, the Correlation Interferometry Direction Finding (CIDF) algorithm, will be used to assess candidate array solutions generated by the genetic algorithm. CIDF Correlation domain statistics such as main correlation beamwidth (proportional to DF accuracy), and correlation sidelobe levels (average relates to array robustness, maximum relates to potential for wild bearings) were used to assess each candidate array over the entire frequency band of interest. This paper will show how that the use of a genetic algorithm, with an optimization function based on CIDF correlation statistics, and a fitness function which adjusts the keep-mate-mutate-discard algorithm specifically for the antenna placement problem, will allow for the definition of an optimized DF array configuration. This paper will derive the critical optimization and fitness functions, and then use examples of a large jet aircraft, a medium size business jet, and a small Unmanned Aerial Vehicle to demonstrate the genetic algorithm capability to solve the DF antenna placement problem. As the reader may not be familiar with the theory of interferometric direction finding, or genetic algorithms, a brief tutorial will be provided in Sections I and III respectively.