With multiple operators announcing field trials and devices becoming available, we wanted to write a paper on 25 Gbps PON to explore the bandwidth, latency, upgrade, and cost impacts. The three authors of this paper represent a component supplier, a system provider, and a major operator. From the component level, we want to understand the cost and technology differences between the 10 Gbps PON technologies, 25 Gbps PON technologies, and the future 50 Gbps/100 Gbps PON standards. From a system level, we will explain the coexistence and upgrade paths from lower PON speeds to 25 Gbps. From an operator perspective, we want to explore the use cases for a 25 Gbps symmetric PON solution. Finally, this paper explains the expected bandwidth and latency possibilities with a 25 Gbps PON system. PON has very significant overhead from forward error correction, framing, and burst overheads that will drop the 25 Gbps line rate. In the upstream direction, tradeoffs must be made between low latency and high bandwidth. This paper will present some simple models to explore those tradeoffs.
Since analyses of both the ITU-T and IEEE standards would be prohibitive for a single document, this paper focuses on the 25GS MSA standard that uses the ITU-T’s XGS framework. (We believe that a significant amount of the analysis and conclusions would be the same for the IEEE’s 25G EPON.) The 25GS MSA defines a downstream rate of 25 Gbps with upstream rates of either 10 Gbps or 25 Gbps. This paper will only consider the 25 Gbps symmetric system where both upstream and downstream are 25 Gbps.
The results in this paper are largely based on modeling of the XGS and 25GS PON systems with some spot checking of the model in XGS mode from lab tests. Since 25GS technology is still in development, large system testing results are not available.