Doubly robust power optimization for network localization

Many future wireless applications and services will require reliable and accurate positional information of the mobile nodes. Efficient power resource utilization can not only increase the network lifetime, but also improve the localization accuracy. The optimal power allocation depends on network parameters, which can only be estimated in practice and hence are subject to uncertainties. In this paper, we present a doubly robust optimization framework for power allocation in location-aware networks. Specifically, the first-layer robust strategy is to minimize the node´s maximum directional position error bound (mDPEB), and the second-layer is to further minimize the node´s largest mDPEB for the worse-case due to the network parameter uncertainties. We also develop an efficient algorithm to enable parallel computations. The simulation results show the remarkable performance improvement by the proposed doubly robust scheme over both non-robust and single-layer robust schemes.