Sensor Selection With Cost Function Using Nondominated-Solution-Based Multiobjective Greedy Method

In this study, a new greedy sensor selection algorithm with a cost constraint is proposed based on a nondominated-solution-based multiobjective greedy (NMG) method, and its performance is investigated by comparing it with a previously proposed method. The cost function is simultaneously considered with the D-optimality criterion, and the sensor set is selected based on the idea of the nondominated solution. Although a multiobjective optimization method for sensor selection with a cost constraint was previously considered based on the linear combination of the objectives, it requires a hyperparameter that determines the balance between the actual objective and the cost of the optimization. On the other hand, the proposed algorithm can obtain the Pareto solution without tuning the balance between the actual objective and the cost. We demonstrate the effectiveness of the proposed algorithm on the three different real datasets that are related to the sea surface temperature field, the flowfield around an airfoil, and the combustion field in a rocket chamber. A binary cost function is virtually imposed for each potential sensor location, and a sensor selection problem with a cost constraint is simulated. The results of the numerical experiments demonstrated that the NMG method could field a Pareto solution, and the objective values of almost all the sensor sets at a certain cost selected by the proposed method are superior to those selected by the previous method.