Research on localization systems has shifted from focusing mainly on accuracy towards a more cognitive design, accounting for communication constraints, energy limitations, and delay. This leads to a variety of sensor selection optimization problems that are solved using techniques from convex optimization. We provide a novel formulation of the sensor selection problem over an extended time horizon, aiming to minimize the sensing cost of an entire path while guaranteeing a certain position accuracy. We state algorithms for determining lower and upper bounds on the sensing cost and utilize these in a path selection problem for autonomous agents. Simulation results confirm the usefulness of our approach, where we observe a benefit of optimizing over longer time horizons in low to medium noise scenarios compared to a myopic sensor selection scheme.