Distributed Decoding for Coded Distributed Computing

Article Computer Science, Information Systems

Adaptive and Heterogeneity-Aware Coded Cooperative Computation at the Edge

Yasaman Keshtkarjahromi et al.

Summary: Cooperative computation is a promising approach for localized data processing at the edge, especially for IoT applications. Coded computation, which offloads sub-tasks to other devices for computation with the use of erasure codes, is gaining interest due to its higher reliability and lower communication costs. In this article, a coded cooperative computation framework called C3P is developed, considering the heterogeneous and time-varying nature of edge devices. The results show that C3P achieves almost optimal task completion delay, has high resource utilization efficiency, and significantly improves task completion delay compared to baselines.

IEEE TRANSACTIONS ON MOBILE COMPUTING (2023)

添加到收藏夹

Article Computer Science, Information Systems

Incentive-Based Coded Distributed Computing Management for Latency Reduction in IoT Services-A Game Theoretic Approach

Nakyoung Kim et al.

Summary: This article proposes novel coded distributed computing mechanisms on heterogeneous mobile devices to reduce latency in Internet-of-Things services. Through analyzing a Stackelberg game between task publishers and mobile devices, it achieves significant latency reduction and social welfare improvement compared to benchmark mechanisms. The proposed algorithms effectively balance computing speed, energy consumption, and workload allocation in a strategic decision-making structure.

IEEE INTERNET OF THINGS JOURNAL (2021)

添加到收藏夹

Proceedings Paper Computer Science, Hardware & Architecture

Incentive Mechanism Design for Distributed Coded Machine Learning

Ningning Ding et al.

Summary: This paper studies the optimal incentive mechanism of a distributed machine learning platform for motivating proper workers' participation in coded machine learning. It proposes a method to summarize workers' heterogeneity as a one-dimensional metric, enabling efficient selection of workers under incomplete information.

IEEE CONFERENCE ON COMPUTER COMMUNICATIONS (IEEE INFOCOM 2021) (2021)

添加到收藏夹

Article Computer Science, Hardware & Architecture

Edge Computing in the Dark: Leveraging Contextual-Combinatorial Bandit and Coded Computing

Chien-Sheng Yang et al.

Summary: This paper investigates computation offloading over unknown edge cloud networks, proposing an online coded edge computing policy to maximize cumulative expected reward in an asymptotically-optimal manner. Numerical studies show that this policy significantly outperforms other benchmarks in terms of cumulative reward.

IEEE-ACM TRANSACTIONS ON NETWORKING (2021)

添加到收藏夹

Article Computer Science, Information Systems