Realistic mobility trajectory prediction using LSTM
The goal of this work is to predict the trajectory of a mobile user using recurrent neural networks (eg. LSTM). Having this information, one can then use this to predict, for example, in 5G, to predict when to make handovers.
The tasks involved are:-
Creating realistic datasets (eg. varying user speeds) from the SLAW mobility model and extracting useful features.
Using an optimized LSTM architecture to predict the trajectory and to verify how far in the future the trajectory can be predicted.
Check how transferrable this architecture is to other mobility models.
Lee, K., Hong, S., Kim, S.J., Rhee, I. and Chong, S., 2009, April. Slaw: A new mobility model for human walks. In IEEE INFOCOM 2009 (pp. 855-863). IEEE.
Gebrie, H., Farooq, H. and Imran, A., 2019, May. What machine learning predictor performs best for mobility prediction in cellular networks?. In 2019 IEEE International Conference on Communications Workshops (ICC Workshops) (pp. 1-6). IEEE.
Knowledge of machine learning algorithms (especially recurrent neural networks)