No abstract

No abstract

We propose a new dynamic bandwidth allocation (DBA) method for long-reach passive optical networks that can reduce the upstream latency. With this method, an optical line terminal allocates bandwidth to long-distance (up to 100 km) optical network units (ONUs) with a transmission request prediction and reduces the latency of the long-distance ONUs. This has no influence on the latency of short-distance ONUs even if they coexist with long-distance ONUs. The results of real machine experiments show that this new DBA method achieves a smaller latency with long-distance ONUs than the conventional DBA technique, and that there is no influence on short-distance ONUs. The latency time and jitter are kept below 1300 µs and 1000 µs, respectively, for both long-and short-distance ONUs. The experimental results also reveal the effect of the proposed method on bandwidth utilization efficiency and fairness. The degradation in bandwidth utilization efficiency when one long-distance ONU is added is small (0.3%), and the fairness index degradation is negligible (less than 0.1%), when the parameters are optimized. These results indicate the validity of the proposed DBA method.

No abstract

In Ethernet Passive Optical Network (EPON) systems, Dynamic Bandwidth Allocation (DBA), which is based on a queue status request communicated via REPORT messages, is an indispensable function for providing fair and high bandwidth utilization. However, REPORT messages do not always contain the same values even if the same REPORT formats are used under the same queue size conditions because the contents of REPORT messages are not concretely specified and some flexibility remains. In this paper, to solve this issue and improve EPON interoperability, we propose the diversity DBA method, which can accommodate various ONUs with different REPORT implementations.The diversity DBA has two functions, namely, distinguishing of REPORT implications and fair bandwidth allocation for ONUs with different REPORT implications. Moreover, we simulate bandwidth allocation based on diversity DBA and report successful results that confirm its validity.