GPON czy XGS-PON? Jak zaplanować migrację bez przebudowy sieci

How XGS-PON differs from GPON

GPON is defined in ITU-T recommendation G.984. It provides up to 2.5 Gb/s downstream and 1.25 Gb/s upstream, that is asymmetric transmission. It is a mature technology, with an extensive device ecosystem and the lowest unit cost, dominant today in access networks worldwide. XGS-PON is standardised in ITU-T G.9807.1 and delivers a symmetric 10 Gb/s in both directions. The letter S in the name stands for symmetry, which distinguishes it from the earlier XG-PON with its asymmetric 10/2.5 Gb/s profile.

What is key for planning a migration, however, is not throughput but wavelengths. GPON operates at 1490 nm downstream and 1310 nm upstream, while XGS-PON uses 1577 nm downstream and 1270 nm upstream. These four wavelengths are separated far enough that they can coexist on the same fibre without interfering with one another. This is the foundation of the entire migration strategy. You do not have to choose between GPON and XGS-PON, because you can have both technologies at once, on the same passive infrastructure, throughout the transition period.

Coexistence on a single fibre

The separation of wavelengths allows GPON and XGS-PON to coexist in the same optical distribution network, which enables a gradual modernisation. In practice this means that a subscriber served by a GPON ONT terminal at 1490 and 1310 nm and a subscriber served by an XGS-PON ONT at 1577 and 1270 nm can be connected to the same splitter, the same feeder fibre, and the same OLT port.

Coexistence is achieved in two ways. The first is a combo OLT with a dual-mode module. A combo module in an SFP+ form factor combines a GPON and an XGS-PON transmitter and receiver in a single device. It contains a continuous 1577 nm transmitter with an EML laser, a continuous 1490 nm transmitter with a DFB laser, and burst-mode APD-TIA receivers for 1270 and 1310 nm. One module, one port, two technologies at once. The second way is an external coexistence element (CEx), that is a passive WDM filter built with thin-film technology, which combines GPON and XGS-PON signals from separate OLT ports onto a single fibre. It is used where the OLT does not support combo modules, or where the network additionally carries RF video at 1550 nm. The choice between these two approaches determines the cost and the scope of work at the OLT node.

What must be replaced, and what to leave alone

‍
Let us break the access network down into components and check each of them.

The feeder and distribution fibre stays unchanged, because the same infrastructure carries all four wavelengths and there is no need to add fibres or change routes. Splitters also stay unchanged, because passive splitters are broadband elements, and a 1:32 splitter serving GPON today will serve XGS-PON without modification. This is the most common misunderstanding in migration conversations. Connectors, splice closures, and patch panels also stay unchanged, that is the entire passive infrastructure.

What requires replacement is the optical module in the OLT, and this is the element that actually has to be swapped. You replace the GPON module with a GPON and XGS-PON combo module, or add a separate XGS-PON module together with a CEx filter. The line card or the whole OLT is a situation that depends on the circumstances. If the existing OLT supports XGS-PON modules and has adequate backplane capacity, a module swap is enough; if not, the card or the device has to be replaced. This is the largest cost item in the whole migration and the main point to verify at the outset. You replace the subscriber ONT terminal, but only where it is needed, because a subscriber using a 300 Mb/s service does not need an XGS-PON ONT. You replace terminals only for those customers who are moving to higher tiers.

The optical budget, the trap that gets forgotten

Migration to XGS-PON is not neutral for the power balance in the network. Combo modules operate in optical budget classes compliant with ITU-T G.984.5 and G.9807.1. Typical classes are C+ and D, where a class C+ module provides a budget on the order of 35/34 dB over 20 km of single-mode fibre.

In practice this means you have to check two things before deployment. First, what the real optical budget is on the longest and most heavily branched sections of your network. If the network was designed with minimal headroom for class B+ GPON, moving to XGS-PON may require higher-class modules or a reduction of the split ratio on some splitters. Second, whether adding a CEx coexistence element does not degrade the balance beyond the allowable level, because every passive element in the path introduces insertion loss that has to be accounted for in the calculation. An OTDR measurement and an optical budget calculation for every branch of the network is the first step of any migration project, not the last.

Three migration scenarios

The first scenario is swapping modules to combo in the existing OLT. The condition is that the existing OLT supports XGS-PON modules and has sufficient switching capacity. The scope of work covers replacing GPON modules with combo modules in the ports serving the subscribers earmarked for migration, configuring XGS-PON service profiles in the OLT, and replacing ONTs for selected subscribers. The cost is the lowest of the three scenarios, with the main items being combo modules and subscriber terminals. The lead time is short, because the work is limited to the OLT node and visits to subscribers moving to higher tiers, and the risk remains low, provided the module-to-OLT compatibility check was done before purchase.

The second scenario is adding a separate XGS-PON OLT with a coexistence element. The condition is that the existing OLT does not support XGS-PON but is to remain in service. The scope of work covers installing a new XGS-PON OLT, mounting passive CEx filters that combine the signals from both devices onto a shared fibre, and reconfiguring the patch panel at the node, plus replacing ONTs for the migrated subscribers. The cost is medium, because the price of a new OLT and CEx filters is added, but the old hardware keeps running and is not written off. The lead time is longer than in the first scenario due to the installation work at the node, and the risk lies mainly on the optical budget side, because CEx filters add attenuation to the path.

The third scenario is replacing the OLT with a combo unit. The condition is that the existing OLT is nearing the end of its life cycle or lacks the capacity to handle XGS-PON traffic. The scope of work covers replacing the OLT with a device that supports combo modules, migrating the configuration, and re-patching the fibres, with ONT replacement following gradually, as needed. The cost is highest at the moment of deployment but lowest over a few-year horizon, because it eliminates another investment two or three years down the line. The lead time is the longest and requires a careful plan for cutting subscribers over without a service interruption, and the risk is operational in nature, tied to migrating services on a live network.

The subscriber ONT, where the costs really lie

For a network serving several thousand subscribers, the cost of subscriber terminals quickly exceeds the cost of node hardware. It is the item that decides the pace of migration. The good news is that coexistence allows this investment to be spread over time. GPON subscribers stay on their terminals for as long as their service tier does not require otherwise, and you install an XGS-PON ONT only where a customer moves to a tier beyond GPON's capabilities.

The practical strategy used by operators looks like this. First, business customers and top-tier subscribers migrate, where the throughput difference is noticeable and justifies the cost of the terminal. Next, new connections, where installing an XGS-PON ONT right away eliminates a later visit. Last, gradually, the remaining subscriber base, at the pace of natural hardware replacement during failures and contract changes.

Why it is worth planning this now, not in two years

XGS-PON has ahead of it, according to market forecasts, a multi-year period of dominance as the access-network standard, before deployments of 25G-PON and 50G-PON genuinely appear. This means that an investment in XGS-PON is not a transitional technology but a foundation for the coming years. More importantly from a planning perspective, the separation of wavelengths also works towards the next generations. An optical distribution network prepared for GPON and XGS-PON coexistence is at the same time prepared for the next step, without changes to the passive infrastructure. A migration carried out today on a coexistence basis protects the return on earlier GPON investments and at the same time opens the way to the next standards.

A checklist before starting the project