MPC-based Rate Calculation for Effective Bandwidth in 5G Communication Systems

The effective bandwidth plays an important role in different aspects of nowadays communication systems. The concept is based on calculating a constant transmission data rate for certain incoming traffic distribution in an asymptotic manner. Meanwhile, the latency and reliability of the users are jointly considered and satisfied. The effective bandwidth is therefore widely used to simplify the analysis of a variety of 5 G communication tasks (e.g. resource allocation for ultra-reliable low-latency communication). Nevertheless, the concept is difficult to extend to traffic patterns other than Poisson distribution. Moreover, the performance of effective bandwidth can hardly guarantee the rate to be tight upper-bound of the incoming traffic when the user’s delay requirement is extremely small. In this work, we consider the effective bandwidth concept in a finite time horizon with a given traffic prediction module. The output rate calculation is simplified and updated according to the model predictive control (MPC) concept. We further study the influence of the length of the time horizon and summarize that our proposed algorithm approximately equals the effective bandwidth concept when having an infinite time horizon. Moreover, a practical calculation method for Poisson distributed incoming traffic is provided and compared with the effective bandwidth concept. The simulation results indicate that the MPC-based solution can provide similar rates to effective bandwidth in both stationary and dynamic traffic scenarios. Meanwhile, the introduced method can reduce resource waste and improve the efficiency of the system. For Poisson-distributed incoming traffic, the optimal MPC time horizon can be easily calculated. With the obtained time horizon, a more accurate data rate to the reference value can be calculated than effective bandwidth.