Secure and Resilient Multi-Access Edge Computing Through Coded Distributed Computing Techniques

Weiquan Ni, Shaoliang Zhu, Alia Asheralieva, Carsten Maple, Xuetao Wei

Abstract

Coded distributed computing (CDC) is envisioned to enhance security and resilience of multi-access edge computing (MEC) systems where tasks are offloaded by MEC servers acting as "leaders" to nearby edge devices acting as "agents" for timely execution. Yet, current CDC schemes suffer from suboptimal task execution delays since each agent can only handle one task at a time. This limitation can increase delays of leaders in recovering the final task result if some agents’ computing results are delayed or erroneous. To address this issue, we propose a new CDC-based task offloading method for MEC systems, in which more tasks can be allocated to faithful and resource-abundant agents to increase the percentage of accurate computing results received by leaders. However, the more tasks assigned by a leader to an agent, the more incentives for the agent must be paid by the leader since agents are usually represented by privately owned edge devices. Thus, in this paper, an optimized task allocation scheme with an incentive provisioning mechanism is proposed to reduce leaders’ task execution delays and rewards for agents in CDC-based MEC systems. Numerical results show the effectiveness of our scheme.
BibTeX
@inproceedings{ni2024secure,
  author = {Weiquan Ni and Shaoliang Zhu and Alia Asheralieva and Carsten Maple and Xuetao Wei},
  title = {{Secure and Resilient Multi-Access Edge Computing Through Coded Distributed Computing Techniques}},
  booktitle = {Workshops @ GLOBECOM},
  year = {2024},
  doi = {10.1109/gcwkshp64532.2024.11100762},
}

Loading related content...

Figure