On the Undecidability of Mobility Prediction and What to Look at in Mobility to Improve Communication in Mobile Networks

摘要

This work presents an analysis of mobility prediction, concluding that it is an undecidable problem. Even though one cannot always predict even its own future movement actions, it does not mean that there is no use for mobility knowledge. In mobile networks, better knowledge on how and when a node (hereafter referred as a mover) will decide on its next movement actions might lead to near-optimum protocol performance. In such situations, before endeavoring into sophisticated analysis by way of restricted mobility traces gathered just for that purpose, one could start checking on how much we already know (or are able to find out) about mover's actions. Based on that, the next step would be to work on how to use mobility data more appropriately. As we use such data, we can increasingly better understand mobility, making space for adaptive communication protocols. Such methodology does not go against any other analytical studies for capturing mobility properties; on the contrary, it just anticipates other uses for mobility data. Even though it is not feasible yet to consider upgrading existing routing protocols, so that full mobility knowledge is taken into account, one can envision an application routing over an overlay network. There is much hope for such an approach given that mobile networks are going to be more widely available as the Internet-of-Things evolves.