Dynamic Resource Management in Optical Grid

In this brief, for optical grid, we propose three dynamic resource management methods. The proposed methods use a dynamic lightpath establishment and release algorithm, and those are implemented in a job manager. Moreover, those methods use three different task transmission controllers so that a resource manager can store a constant number of tasks. The first controller uses a proportional-integral-differential (PID) control to adjust the number of tasks that are transmitted to a computing site. The second controller uses a model predictive control (MPC) based on mixed integer quadratic programming (QP), and the third controller uses an MPC based on QP. Each task transmission controller manages the computing resources, and the dynamic lightpath establishment and release algorithm manages the network resources. We evaluate the performances of the proposed methods with MATLAB and investigate the effectiveness of each method. Numerical examples show that the computing resources and the network resources can be effectively utilized by using our proposed methods. In addition, we show the implementation capabilities of the three methods.