Human Mobility Support for Personalised Data Offloading

WiFi Access Points (APs) can be used to offload data or computation tasks while users are commuting. However, due APs’ limited coverage, offloading performance is heavily impacted by the users’ mobility. This work proposes leverage human mobility inform tasks, taking a based approach leveraging granular datasets from two cities: Porto and Beijing. We define Offloading Regions (ORs) as areas where commuter’s would enable offloading, propose an unsupervised learning methodology extract ORs traces. Then, we characterise analyse according opportunity metrics such type, availability, total time offload, delay. Results show that in 50% of trips, spend more than 48% travel inside extracted proposed methodology. The ability predict next benefit orchestration. predictability, although crucial, proves challenging, expressed poor predictive well-known models ( $\approx $ 37% acc. for best predictor). regularity properties improve up 35%. Finally, look into impact further OR extraction prediction parameters. exploration phase length does not discovery low relevance ORs, both filtering predicting multiple increase predictability. By characterising trade-off between predictability opportunities transit, highlighting need systems adopt hybrid strategies, i.e., mixing opportunistic strategies. conclusions findings on likely generalise varied urban scenarios given high degree similarity results obtained different independently collected datasets.