For the past 15+ years the majority of PON (Passive Optical Network) equipment sold worldwide has been based on the GPON (Gigabit-capable PON) standard. This is the story of how GPON was created and why it became universally adopted. Even though this is nearly 20 years old, this story is relevant today because it offers a lesson on the importance of industry standards and the importance of the specific standards bodies that create them.
For well over a decade GPON has been the dominant protocol for Passive Optical Network (PON) equipment globally. Considering it was ratified by the International Telecommunications Union (ITU-T) back in 2004, GPON will soon be surpassed by a next generation PON protocol. The question is: Which protocol will be the next GPON?
There are several contenders. From the ITU/FSAN we have XGS-PON, which has a line rate of 10 Gbps symmetric. The secret decoder ring for “XGS” is as follows: The “X” is read as the Roman numeral X for 10; the “G” stands for Gigabit per second; and the “S” stands for symmetric as in 10 Gbps downstream and 10 Gbps upstream.
There are two 25 Gig PON protocols (one from the IEEE and one from the 25GS-PON MSA). The ITU-T, which created GPON and XGS-PON, declined to establish a 25 Gig PON standard because it was only a 2.5x increase over XGS-PON. The FSAN group believe that each new standard should be 4 to 5 times higher than the previous one due to the high cost for an operator to change from one standard to the next. Accordingly, the ITU-T has two higher speed standards available – the 40 Gbps NGPON2 and the recently-ratified 50 Gbps protocol designated G.9804.
All these higher speed protocols certainly meet (or exceed) the foreseeable market requirements. But they differ significantly on price. Recall from one of my previous blogs the main cost driver for PON equipment is the process of upstream bursting, which is itself very sensitive to the upstream line rate. At present, 10 Gbps upstream is just on the cusp of affordability. Normally, this would mean that even XGS-PON would be considerably higher cost than GPON. But there’s also a fortuitous situation here regarding something called Forward Error Correction (FEC) that makes XGS-PON only modestly more expensive than GPON. Let’s look into why that is.
Briefly, FEC is a mechanism that allows a receiver to detect and then correct bit errors introduced during the transmission of digital signals. To do this the transmitter performs 3 steps prior to sending data: 1) It divides whatever data it needs to send into fixed length blocks of, say, 216 bytes. 2) For each block it calculates a hash code, which is a unique signature that also has a fixed length of, say, 32 bytes. 3) It then appends the hash to the end of each block of data and then sends the combined message of 248 bytes. The receiver knows to split the 248 bytes received into the data block and hash. It does its own hash calculation on the data and compares that with the hash it received. If they match, there were no bit errors. If they don’t the receiver performs an iterative process to figure out which bits are in error and then corrects them. [BTW: The hash is not a run-of-the-mill hash like SHA-1. It’s tailored to make the iterative process easier.]
The GPON standard has FEC as an option, but it was never needed. The downstream and upstream line rates of 2.4 Gbps and 1.2 Gbps were selected, in part, to make FEC generally unnecessary. By the time GPON products were developed and made available to the market, even the low cost/high volume optical transceivers were just good enough that FEC was not necessary.
The situation is the different with XGS-PON in that transceivers produced at high volume are not quite good enough without FEC. For the foreseeable future, FEC is required in both the upstream and the downstream. But there is a serendipitous side effect here. Because FEC works so well, vendors can use lower quality transceivers [i.e., ones w/ inherently higher Bit Error Rate (BER)] and still get all the errors corrected. This effectively increases the manufacturing yield for XGS-PON transceivers. And this, in turn, lowers the cost of XGS-PON equipment. Keep in mind that the most expensive component in an ONT is the optical transceiver. Consequently, the cost of XGS-PON equipment is lower than it would be without FEC.
At present this fortunate situation does not extend to 25-Gig PONs. 25-Gig PON products are still relatively new and manufacturing efficiencies have a longer way to go relative to XGS-PON. Considering the time from GPON being affordable to XGS-PON being mostly affordable, it could be many years before 25-Gig PON is cost effective.
In the meantime, I’m putting my money on XGS-PON. It meets all the foreseeable market requirements, and it is by far the most cost effective.