Adaptive Control for Real-time Computing System

There may be great degradation in service level management (e.g., an email server) due to dynamical variants of real systems and measured disturbance in the network. An adaptive predictive congestion control is presented for service systems to prevent oscillation and delay induced by highly stochastic behaviors of computing system which can not be got rid of by conventional control. We analyze the controller in detail and prove the error can be reduced to zero in steady state. The effectiveness of the proposed methods is demonstrated by the simulation results.