Can 5G and Machine Learning Replace the GPS?

Whereas physical obstacles were mostly associated with signal attenuation in telecommunications, their presence in 5G’s millimeter wave systems adds complex, non-linear phenomena, including reflections and scattering. The result is a multipath propagation environment, shaped by the obstacles encountered during transmission, indicating a strong and highly non-linear relationship between a device’s received radiation and its position. In this presentation, new ways to shape these signals will be discussed so as to estimate a mobile device’s position, including the physical intuition behind those accuracy-enhancing manipulations. To untangle the information hidden in the received signal into a mobile device position, different neural networks architectures can be employed, enabling a low-power single anchor positioning system. This positioning system can be further enhanced so as to track users, using short-term historical data and sequence learning approaches. The discussed system sets a new state-of-the-art for non-line-of-sight millimeter wave outdoor positioning accuracy, while having a much higher energy efficiency when compared to low-power GPS implementations, and thus answering the question in the title: yes, they can!


João Gante

João Gante is a PhD Candidate at IST, researching Machine Learning-based algorithms for positioning and tracking with 5G, supervised by Professor Leonel Sousa. João is currently an ML Engineer at nPlan, London, where ML is the key to predict the outcome of large construction projects.INESC/IST