The maturity of outdoor positioning systems based on satellites encourages indoor positioning research to focus on radio technologies. However, specific infrastructures often have to be deployed in this case. Then, inertial sensors appear to be a good relay to radio systems. A system fusing INS and GNSS could thereby compute a position anywhere. Yet, taking advantage of each sensor requires to know which one is the most reliable in real-time. Therefore, a quantification of the sensors' reliabity is introduced in this paper. This approach aims at running both outdoors and indoors. Moreover, the complexity of algorithms is carefully studied here to fit the user mobility requirements. Experiments are conducted in reproducible situations and results show that taking reliabilities into account benefits the hybridization of INS and GNSS for positioning in both convenient and constrained environments.