[Paper Review] Coded Cooperative Computation for Internet of Things.

Yasaman Keshtkarjahromi, Hülya Seferoğlu|arXiv (Cornell University)|Jan 13, 2018

Ferroelectric and Negative Capacitance Devices18 references3 citations

TL;DR

This paper proposes the Computation Control Protocol (CCP), a coded cooperative computation framework for IoT that dynamically allocates sub-tasks to heterogeneous devices using erasure coding to optimize for delay and energy efficiency. CCP reduces task completion delay significantly, achieves near-theoretical performance, and maintains over 99% resource utilization under time-varying conditions.

ABSTRACT

Cooperative computation is a promising approach for localized data processing for Internet of Things (IoT), where computationally intensive tasks in a device could be divided into sub-tasks, and offloaded to other devices or servers in close proximity. However, exploiting the potential of cooperative computation is challenging mainly due to the heterogeneous nature of IoT devices. Indeed, IoT devices may have different and time-varying computing power and energy resources, and could be mobile. Coded computation, which advocates mixing data in sub-tasks by employing erasure codes and offloading these sub-tasks to other devices for computation, is recently gaining interest, thanks to its higher reliability, smaller delay, and lower communication costs. In this paper, we develop a coded cooperative computation framework, which we name Computation Control Protocol (CCP), by taking into account heterogeneous computing power and energy resources of IoT devices. CCP dynamically allocates sub-tasks to helpers and is adaptive to time-varying resources. We show that (i) CCP improves task completion delay significantly as compared to baselines, (ii) task completion delay of CCP is very close to its theoretical characterization, and (iii) the efficiency of CCP in terms of resource utilization is higher than 99%, which is significant.

Motivation & Objective

• Address the challenge of task offloading in IoT systems with heterogeneous, time-varying computing power and energy resources.
• Improve task completion delay and communication efficiency in localized IoT computation.
• Design a dynamic, adaptive offloading protocol that maintains high resource utilization despite device variability.
• Integrate coded computation techniques with real-time resource adaptation to enhance reliability and reduce latency.

Proposed method

• Employ erasure coding to mix data across sub-tasks, enabling robust and efficient offloading to nearby devices.
• Design a dynamic sub-task allocation mechanism that adapts to real-time variations in device computing power and energy levels.
• Formulate a control protocol that balances load across helpers while minimizing completion delay and communication cost.
• Use theoretical modeling to characterize optimal performance and guide protocol design.
• Integrate energy and computation constraints into the sub-task assignment decision process.
• Implement a feedback-driven adaptation mechanism to respond to changing device availability and performance.

Experimental results

Research questions

• RQ1 How can coded computation be effectively adapted to the heterogeneous and dynamic nature of IoT devices?
• RQ2 To what extent can dynamic sub-task allocation reduce task completion delay in IoT cooperative computation?
• RQ3 How close can a practical protocol come to the theoretical delay lower bound in heterogeneous IoT environments?
• RQ4 What level of resource utilization can be achieved in practice under time-varying device conditions?

Key findings

• CCP reduces task completion delay significantly compared to baseline offloading strategies.
• The actual task completion delay of CCP is very close to its theoretical performance lower bound.
• CCP achieves over 99% efficiency in resource utilization, indicating minimal waste of computing and energy resources.
• The protocol demonstrates strong adaptability to time-varying device capabilities and energy levels.
• Coded computation combined with dynamic control enables reliable and low-latency task execution in IoT environments.
• The integration of erasure coding and real-time adaptation leads to superior performance in both delay and efficiency.