Distributionally Robust Optimization for Peak Age of Information Minimization in E-Health IoT.

Zhuang Ling,Fengye Hu,Hongliang Zhang,Zhu Han,H. Vincent Poor
DOI: https://doi.org/10.1109/icc42927.2021.9500955
2021-01-01
Abstract:In this paper, we consider a real-time E-Health Internet of Things (IoT) system with the uncertainty of channel state information (CSI), in which a wearable device collects radio frequency (RF) energy from a Personal Digital Assistant (PDA), and then transmits healthcare data status updates to the corresponding PDA promptly. The Peak Age of Information (PAoI) is considered as a parameter to measure the freshness of information. Our goal is to minimize the average PAoI under non-convex constraints related to an uncertain CSI mismatch model. Only mean and variance information is specified in the distributional ambiguity set. This distributionally robust optimization problem is transformed into a tractable semi-definite programming (SDP) problem using the Conditional Value-at-Risk (CVaR) based method. To solve this NP-hard problem effectively, we decompose the PAoI minimization problem into two subproblems, and propose a low complexity iterative algorithm to derive a suboptimal solution. Simulation results show an average PAoI-energy tradeoff in the considered healthcare IoT, and the CVaR based method can achieve a better performance than a non-robust method.
What problem does this paper attempt to address?