Summary of Federated Learning with Energy Harvesting Devices: An Mdp Framework, by Kai Zhang et al.
Federated Learning With Energy Harvesting Devices: An MDP Framework
by Kai Zhang, Xuanyu Cao
First submitted to arxiv on: 17 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Signal Processing (eess.SP)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A federated learning (FL) system requires edge devices to perform local training and exchange information with a parameter server, leading to substantial energy consumption. To address this challenge, researchers applied an energy harvesting technique to extract ambient energy for powering edge devices. The convergence bound was established for the wireless FL system with energy harvesting devices, showing that convergence is impacted by partial device participation and packet drops, which depend on the energy supply. A joint device scheduling and power control problem was formulated as a Markov decision process (MDP), solving it to derive an optimal transmission policy with a monotone structure. To overcome the curse of dimensionality, a low-complexity algorithm was proposed, which is asymptotically optimal as the number of devices increases. Additionally, a deep reinforcement learning algorithm was developed for unknown channels and harvested energy statistics, leveraging the monotone structure of the optimal policy to improve training performance. Numerical experiments on real-world datasets validated the results and demonstrated the effectiveness of the proposed algorithms. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Federated learning is a way for devices like smartphones or smart home appliances to work together and learn from each other without sharing their data. But this process uses up a lot of energy, which can be a problem because these devices have limited batteries. To solve this problem, researchers came up with an idea to use the ambient energy around us, like light or vibrations, to power these devices. They showed that by using this energy, they could make sure that all the devices were working together effectively and efficiently. The researchers also developed algorithms to make sure that the devices were sharing their information in a way that was fair and efficient. |
Keywords
» Artificial intelligence » Federated learning » Reinforcement learning
