Mobile edge computing (MEC) is envisioned as re- expanded cloud compared to fog computing. As making mobile services computing sank to the edge of network further, performance improvement can be got on time aspect. Therefore, MEC is seemed as a potential technology for delay-sensitive applications. Based on that, a reasonable computation offloading strategy would reduce system consumption for MEC further and release its computational capability. In view of resource shortage, especially bandwidth competition problem among Small Cells and edge computational requirements of HetNet, we focus on distributed uplink offloading for MEC in macro-micro coordination scene. It mainly contains two steps. First, based on Lyapunov to solve offloading decision-making problem for users in each Small Cell. Second, complete offloading update order- making of Small Cells in Macro Cell with proposed deviation update decision algorithm (DUDA). Our strategy makes up for the lack of system stability and uplink analysis in existing research. Numerical results demonstrate the effectiveness of our strategy.