Joint Task offload and Resource Allocation for Cognitive Edge Computing Using AI Algorithm

Abstract

Mobile edge computing(MEC)brings, besides various opportunities, challenges for the resource allocation. The heterogeneity of resources in multiple cells further exacerbates this challenge. For efficient resource utilization, in this paper, MEC is combined with cognitive radios (CRs) to improve better adaptation. In such a context, a computing offload and resource allocation mechanism is proposed, which can be formulated as user pairing scheme based on coalition game. Such an algorithm first match applicable neighbors for each secondary user(SU) in terms of pairing utility.then, compete optimal resources to computing offload within the cognitive edge computing, considering multiple optimization objectives that are derived from user needs. To obtain the optimal network welfare, a gradient descent algorithm of machine learning is proposed to acquire the near-optimal solution. The results of multiple runs of our simulation demonstrate that the algorithm is efficient, which can show better performance in terms of the network welfare compared to existing resource allocation algorithms.