Application-aware computation offloading in edge computing networks
Rongping Lin,Xuhui Guo,Shan Luo,Yong Xiao,Bill Moran,Moshe Zukerman
DOI: https://doi.org/10.1016/j.future.2023.04.009
IF: 7.307
2023-04-15
Future Generation Computer Systems
Abstract:Edge computing involves distributive computation resources deployed at the network edge, unlike cloud computing, which has central computation resources in data centers. Edge computing is a complement of cloud computing because edge computing effectively reduces the computing response delay by processing computation tasks and data near terminals. Considering the dramatic increase of terminals connected to networks and data generated by terminals, computation tasks from different applications may require significantly different services with different computation requirements, storage requirements, and response delay requirements. Application-aware computation offloading and resource allocation in edge computation can provide efficient and guaranteed computation services to terminals. In this paper, an application-aware computation offloading and resource allocation problem is investigated in edge computing networks, where computation tasks from different applications have different requirements. A non-convex optimization problem of energy consumption minimization is formulated, where terminals, edge nodes, and a cloud are considered. We convert the original non-convex optimization problem into a lower-bound convex problem and an upper-bound convex problem. Then, an algorithm based on the branch-and-bound method is proposed to force the lower- and upper-bound solutions to approach the optimal solution. Finally, the performance of the algorithm is analyzed where the gap to the optimal solution is provided. Numerical results show that the proposed algorithm can provide guaranteed services for tasks of different application types, with improvements over application-unaware algorithms.
computer science, theory & methods