Placement support for signal intelligence units

Sammanfattning: The goal of this thesis was to develop an optimization model that automatically finds optimal groupings for signal intelligence units, aiming to maximize surveillance capability in a user-defined target area. Consideration was taken to transportation possibilities, type of terrain, and the requirement of radio communication between the direction finders. Three scenarios were tested, each providing its own topographical challenges. Several different derivative-free optimization methods were implemented and evaluated, including global methods to find approximate groupings using a geometrical model that was developed, and the local method pattern search that was using the already existing model. Particle swarm and a genetic algorithm turned out to be the best global solvers. A grouping found by the global method was later improved by pattern search by evaluating possible groupings nearby. The greatest practical challenge for particle swarm and pattern search was the ability to find feasible placement points given a desired direction and step length. Workarounds were developed, allowing for more dynamic search patterns. For future use, the placement support should be tested on more scenarios with different prerequisites, and the approved terrain types have to be adjusted according to the kind of vehicle carrying the direction finder.