Unsupervised Deep Learning for Binary Offloading in Mobile Edge Computation Network

Mobile edge computation (MEC) is a potential technology to reduce the energy consumption and task execution delay for tackling computation-intensive tasks on mobile device (MD). The resource allocation of MEC an optimization problem, however, existing large amount may hinder its practical application. In this work, we propose multiuser framework based unsupervised deep learning by offloading servers. binary decision are jointly optimized minimize MDs under latency constraint transmit power constraint. This joint problem mixed integer nonconvex which result in gradient vanishing backpropagation. To address this, novel scheme (BCOS), neural network (DNN) with auxiliary designed. By using as teacher network, student can obtain lossless information training phase. As result, sub-optimal solution be acquired learning-based BCOS. Simulation results demonstrate that BCOS effective solve trained low complexity.