Offloading for Edge Computing in Low Power Wide Area Networks With Energy Harvesting
Hai Lin, Zhihong Chen, Lusheng Wang · IEEE Access · 2019
Edge computing and low power wide area networks (LPWANs) have been recognized as promising technologies in the Internet of Things (IoTs) era to provide massive wireless devices with enhanced computation and low-power, long-distance communication capabilities. The emergence of both technologies is to meet the demand of the rapid development of the IoTs, which motivates us to integrate edge computing into LPWANs to enhance low-power devices' computation capability. In our work, edge computing server co-locates with LPWAN base stations to which the end devices connect directly. Due to limited resources, multiple base stations should cooperate to provide better computation services. This paper works on a simple scenario where two base stations with harvested energy cooperate to tackle computation tasks. Different to previous energy harvesting modeling, we define a new correlated stochastic model for it. The whole system is then modeled as a Markov decision process (MDP), in which new features are defined, such as multiple tasks, multiple objectives, and variable time slots. Afterward, we carry out simulations to analyze the performance, showing that our proposition can utilize the energy efficiently and achieve good performance in terms of task's completion rate and total rewards.