Evolutionarily Reconfigurable Cloud-Integrated Body Sensor Networks

This paper considers a multi-tier architecture for cloud-integrated body sensor networks (BSNs), called Body-in-the-Cloud (BitC), which is designed for home healthcare with on-body physiological and activity monitoring sensors. This paper formulates an optimization problem to integrate BSNs with a cloud in BitC and approaches the problem with an evolutionary game theoretic algorithm. BitC allows BSNs to adapt their configurations (i.e., sensing intervals) to operational conditions (e.g., data request patterns) with respect to multiple performance objectives such as resource consumption and data yield. BitC theoretically guarantees that each BSN performs an evolutionarily stable configuration strategy, which is an equilibrium solution under given operational conditions. Simulation results verify this theoretical analysis; BSNs seek equilibria to perform adaptive and evolutionarily stable configuration strategies under dynamic changes of operational conditions. BitC outperforms a well-known evolutionary multiobjective optimization algorithm, NSGA-III, in optimality, convergence speed and stability.