Blockchain and Learning-Based Computation Offloading in Space-Assisted Power IoT

Abstract

Power systems impose stringent security and delay requirements on computation offloading, which cannot be satisfied by existing PIoT. In this paper, we address this challenge by combining blockchain and machine learning. We also introduce a novel framework named space-assisted PIoT (SPIoT), where the low earth orbit (LEO) satellite block creation delay. Specifically, we propose a Blockchain and semi-distRibuted leArning-based seCure and low-latEncy computation offloading algorithm (BRACE) to minimize the total queuing delay under the long-term security constraint. First, the device-side task offloading is decoupled from the server-side computational resource allocation with Lyapunov optimization. Second, the task offloading problem is solved by the proposed federated deep actor-critic-based task offloading algorithm. Finally, the resource allocation problem is solved by smooth approximation and Lagrange optimization. Simulation results verify that BRACE achieves superior queuing delay and security performances.