Joint Power Control and Task Offloading in Collaborative Edge–Cloud Computing Networks
Sai Wang, Yi Gong
Abstract
Mobile-edge computing arises as a promising technology to allow mobile devices (MDs) to offload delay-sensitive and computation-intensive tasks to the nearby edge servers. However, overloaded tasks from MDs lead to a large latency due to limited computing and channel resources. To mitigate this situation, a collaborative edge–cloud computing network is considered in this article. Based on power control and task offloading, we formulate a mixed-integer nonlinear programming (MINLP) problem to minimize the weighted sum of the energy consumption and the latency. This NP-hard problem is decomposed into a real variable problem and an integer linear programming problem. By leveraging the proposed extreme-value descent (EVD) method, the optimal power strategy is obtained. The simplex method and branch-and-bound method are adopted to find the optimal offloading strategy. Although these two subproblems are well solved, the solutions may be unsatisfactory for the original problem. To find a high-quality solution, we use an alternating optimization (AO) method to jointly optimize the decomposed problems. Theoretical analysis demonstrates that the EVD method converges at the rate of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$O(1/s)$ </tex-math></inline-formula> and the AO method can converge to a suboptimal solution if not the optimal solution. Simulation results show that the proposed algorithms significantly outperform other benchmark schemes.