Loading...

Energy-Aware Computation Offloading for Wireless Powered Mobile-Edge Computing Systems

Bolourian, Mehdi | 2022

281 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 55597 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Shah Mansouri, Hamed
  7. Abstract:
  8. Mobile edge computing (MEC) is envisioned to address the computation demands of Internet of Things (IoT) devices. However, it is crucial for the MEC to operate in coordination with the cloud tier to achieve a highly scalable IoT system. In addition, IoT devices require regular maintenance to either recharge or replace their batteries which may not always be feasible. Wireless energy transfer (WET) can provide IoT devices a stable source of energy. Nonetheless, proper scheduling of energy harvesting and efficient allocation of computing resources are the key for sustainable operation of these devices. In this thesis, we introduce a three-tier wireless powered mobile edge computing (WPMEC) consisting of cloud, MEC servers, and IoT devices. We first formulate a combinatorial optimization problem that aims to minimize the wireless energy transmission. To tackle the complexity of the problem, we propose two approaches. In the first approach, we leverage bipartite graph matching. The second approach corresponds to an online scheme with policy based reinforcement learning. We design a harvest-then-offload mechanism for IoT devices. We also exploit parallel processing to increase the performance of algorithms. Through numerical experiments, we evaluate the performance of our proposed approaches. Our results show that the proposed mechanisms significantly reduce the required energy for operation of IoT devices comparing to different offloading policies. We further show that the computation time can be reduced by 13\% for a computationally intensive task in comparison to an existing work in the literature. The online approach results in up to an additional 60\% decrease in wireless energy transmission
  9. Keywords:
  10. Bipartite Graph Matching ; Reinforcement Learning ; Mobile Edge Computing ; Wireless power Transfer ; Internet of Things

 Digital Object List