Study and Improvement of the Resource Allocation Algorithm in TWDM-PON Based on the Correlation of Switching Periods

This paper focuses on the wavelength distribution in TWDM-PON. A novel fixed-dynamic wavelength distribution (FDWD) mechanism and a coupling algorithm are presented by using the periodic law of users’ traffic. Combining the historical statistics, the Poisson model and the autoregressive (AR) model are used to predict the bandwidth that the user may request. Based on the forecast data, users are distributed into different communities to transmit on different wavelengths, according to the constraints which include the load balance and minimizing tuning overhead. Compared with the existing work, the P-order AR model are presented to model the request traffic. The coupled packet algorithm is optimized by adding the correlation constraints of adjacent switching periods and the wavelength tuning distance is taken into account. The simulation results show that the FDWD can further reduce the tuning cost and improve the load balancing performance.