A min-max model predictive control approach to robust power management in ambulatory wireless sensor networks

This paper addresses the problem of transmission power control within a network of resource-constrained wireless sensors that operate within a particular ambient healthcare environment. Sensor data transmitted to a remote base station within the network arrive subject to node location, orientation, and movement. Power is optimally allocated to all channels using a novel resource efficient algorithm. The proposed algorithm is based on a computationally efficient min-max model predictive controller that uses an uncertain linear state-space model of the tracking error that is estimated via local received signal strength feedback. An explicit solution for the power controller is computed offline using a multiparametric quadratic solver. It is shown that the proposed design leads to a robust control law that can be implemented quite readily on a commercial sensor node platform where computational and memory resources are extremely limited. The design is validated using a fully IEEE 802.15.4 compliant testbed using Tmote Sky sensor nodes mounted on fully autonomous MIABOT Pro miniature mobile robots. A repeatable representative selection of scaled ambulatory scenarios is presented that is quite typical of the data that will be generated in this space. The experimental results illustrate that the algorithm performs optimal power assignments, thereby ensuring a balance between energy consumption and a particular outage-based quality of service requirement while robustly compensating for disturbance uncertainties such as channel fading, interference, quantization error, noise, and nonlinear effects.