Cloud edge-end collaborative resource allocation method for stereoisomerism electric power internet of things

Abstract

The invention discloses a cloud-side-end collaborative resource allocation method for a three-dimensional heterogeneous power internet of things, namely, cloud-side-end collaborative task unloading and computing resource allocation in the space-air-ground integrated three-dimensional heterogeneous power internet of things are jointly optimized. The method comprises the following steps: (1) constructing a system model to establish a three-dimensional heterogeneous power Internet of Things scene formed by a satellite, an unmanned aerial vehicle and a terminal; (2) refining the model; (3) proposing and converting a queuing time delay constraint and joint optimization problem; (4) based on a Lyapunov optimization principle, combining decomposition and solution of an optimization problem, and minimizing an upper bound of drift plus penalty in each time slot; (5) a cloud edge-end collaborative task unloading decision algorithm based on deep reinforcement learning. According to the method, the high-dimensional task unloading problem is solved based on deep reinforcement learning, and the problem of curse of dimensionality under information uncertainty is effectively solved in combination with the complex function approximation capability provided by the neural network and the decision-making capability provided by the act-critic algorithm.

Description

technical field [0001] The invention relates to the technical field of the problem technical field of the cloud-side-terminal collaborative resource allocation method of the power Internet of Things, in particular to a joint optimization of cloud, edge and terminal collaborative task offloading and computing resource allocation in the space-space-ground integrated stereo heterogeneous power Internet of Things Methods. Background technique [0002] With the rapid development of renewable energy sources such as desert solar energy and offshore wind energy, a large number of power Internet of Things (PIoT) terminals have been deployed in remote areas to ensure the safe and stable operation of power infrastructure. The power Internet of Things is the specific manifestation of the Industrial Internet of Things in the smart grid, which has strict requirements on communication and computing capabilities. The current single terrestrial network is difficult to meet these requirement...

AI Technical Summary

Problems solved by technology

[0005] Second, incomplete information and the curse of dimensionality

In addition, due to network heterogeneity and time-varying resource constraints, the number of system states and actions will increase exponentially, causing the curse of dimensionality problem

Traditional learning-based distributed task offloading methods do not consider high-dimensional optimization problems, and thus perform poorly in terms of optimality and convergence

[0006] Third, long-term queuing delay constraints

However, the coupling between long-term queuing delay constraints and short-term optimization decisions further adds to the difficult dimension of solving joint optimization problems

Images