A dynamic data source selection system for smartwatch platform
Ebrahim Nemati, Konstantinos Sideris, Haik Kalantarian, Majid Sarrafzadeh · 2016
A novel data source selection algorithm is proposed for ambulatory activity tracking of elderly people. The algorithm introduces the concept of dynamic switching between the data collection modules (a smartwatch and a smartphone) to improve accuracy and battery life using contextual information. We show that by making offloading decisions as a function of activity, the proposed algorithm improves power consumption and accuracy of the previous work by 7 hours and 5% respectively compared to the baseline.