In this work an investigation is performed in whether the algorithms Simultaneous Perturbation Stochastic Approximation (SPSA) and Virtual Force Algorithm (VFA) are suitable for real-time strategic placement of sensors in a dynamic environment. An evaluation of these algorithms is conducted and compared to Simulated Annealing (SA), which has been used before in similar applications.
For the tests, a computer based model of the sensors and the environment in which they are used, is implemented. The model handles sensors, moving objects, specifications for the area the sensors are supposed to monitor, and all interaction between components within the model.
It was the belief of the authors that SPSA and VFA are suited for this kind of problem, and that they have advantages over SA in complex scenarios. The results shows this to be true although SA seems to perform better when it comes to smaller number of sensors to be placed