Orange is rolling out passive optical LANs (POL), supplied by Nokia, for an initial 20 sites in France, including the headquarters at Issy-les-Moulineaux, as it replaces legacy copper infrastructure serving over 5,000 end points, for WiFi as well as hard-wired terminals.
The POL technology is a localized implementation of passive optical network (PON) fiber technology, which is used by many telcos for their external, customer-serving fiber networks, including Orange itself as a provider of fiber-to-the-home (FTTH) services. Such operators have also been considering use of PON for fronthaul and midhaul in emerging disaggregated 5G networks to meet the challenging requirements for both bit-rate and latency.
The main benefits for Orange of this internal PON/POL deployment are higher performance, smaller footprint, and energy saving through avoidance of active electronic components that require power within fiber infrastructure, as opposed to the terminating points.
“Moving to POL for our intra-office connectivity can save significant energy costs and reduce emissions, which is essential to help Orange meet its ambitious environmental goals,” said Philippe Gacougnolle, Orange France’s director of the internal network domain.
PON technology has been around for well over two decades, with work towards deployment beginning in 1995 followed by standardization of two variants by the ITU. It was developed as an alternative to active optical networks (AON), where the light signals are regularly regenerated by being converted back to electrical signals in optical repeaters. AONs involve routers and switches for signal distribution and transmission to end users, each of which requires a dedicated fiber into the network.
PONs – including POL when applied to a local area network – still require electronic components but only at each end of the fiber path, with optical line terminals (OLTs) feeding the signal in, and optical network terminals (ONTs) at the receiving end close to the user service area. In between, the light waves are transmitted over fiber after they leave the OLT and reach a splitter downstream towards the end users, which uses mirrors and glass to refract the light into multiple paths (up to 128), without any electronic assistance.
This also reduces the equipment needed, since PONs have just one router or switch port and a single fiber between that and the passive splitter to serve those multiple subscribers, sharing the capacity of the light wavelength. This approach is also said by some advocates to avoid need for temperature control around the splitters. This last point is not strictly correct, because glass like any material expands with heating and if this exceeds a certain threshold, errors can occur as a result of refraction bending the light differently. So, some control over temperature is still required.
Advocates also trumpet the reduced probability of component failure by having just one OLT feeding the network, but that is rather disingenuous because this in turn increases exposure to a single component. It is therefore imperative that the OLT itself be of resilient design, just like any critical component that is potentially a single point of failure. At least the OLT itself can be accommodated in secure premises and managed centrally, unlike the splitters, which being passive components out in the network cannot readily be monitored directly. However, given careful installation in a location not subject to excessive temperature variations, the splitters should be robust with no moving or electronic parts.
Security is also levelled as a weakness given the vulnerability of fiber optic networks to tampering and eavesdropping since all signals are often concentrated in a single cable. For this reason, strong encryption of data is recommended and indeed cited by Nokia as a key feature of the Orange installation.
“Security is baked in with built-in encryption and central control as all intelligence of the network resides in the optical line terminal and none at the user end points,” Nokia stated. In this way security is presented as a strength rather than weakness.
The principal attractions of optical fiber networks, whether active or passive, are data transmission capacity and scalability, since bit-rates can be stepped up through improvements in the terminating electronics without having to upgrade the fiber itself.
In this regard Nokia is a strong partner because of its heritage in Bell Labs, which was a leader in fiber optic research three decades before it came into the Finnish company’s hands via acquisitions. Nokia demonstrated what it claimed was the first 100Gbps transmission over a single wavelength of a PON network at Bell Labs in June 2022.
Many current PONs transmit at 25Gbps per wavelength, and as before advances in digital signal processing have been required to encode more data over a given waveband. The alternative way of increasing capacity is to encode in parallel over multiple wavelengths, via wave division multiplexing (WDM). Up to 160 wavelengths can be combined this way with Dense WDM (DWDM) over trunk networks, but scope for this is more limited with PON in the absence of active signal regeneration within the network.
For this reason, Nokia correctly underlines the expediency of achieving high speeds over a single wavelength for PON, even if 100Gbps is more than many customers yet need. At the same time though, there is growing interest in wavelength division multiplexing-PON (WDM-PON) for access and backhaul networks. The use of different wavelengths over a physical point-to-multipoint fiber infrastructure is valuable for separating traffic logically, for point-to-multipoint delivery over a single fiber.
The concept of WDM-PON was first experimented with at Bell Labs in 1989 before PON itself emerged commercially in the mid-1990s, since when it has enjoyed spasmodic spurts of interest but generally failed to gain any traction by dint of being overkill and too expensive – the extra capacity was not needed. But for 5G backhaul and fronthaul, the capacity is needed, and the cost is more likely to be justifiable than in residential fixed-line scenarios.
Nokia has pointed out, though, that each case needs to be analyzed on its own merits, and that WDM-PON will often not be the optimum solution for backhauling radio towers. It noted analysis by Orange in France of a representative area, where there were about an equal number of cell sites and central offices (COs). Under this scenario, deploying a PON of any kind would amount to over-provisioning, with point-to-point (P2P) fitting better since the relationship is almost one-to-one, on the assumption that wireless equipment is retained at all those sites.
However, in emerging 5G architectures WDM-PON, especially, could come into its own. In 5G, options for disaggregated RAN architectures are emerging (though only sparsely deployed as yet). These adopt a three-level structure splitting workloads between central units (CUs), distributed units (DUs) and radio units (RUs). This in turn divides the access bearer network into backhaul, midhaul and fronthaul connections between the three elements, and fronthaul requires bit-rate to the 5G base station of 25Gbps.
Furthermore, this requires extremely low latency of about 10 μs, for the midhaul connection between CU and DU especially, which can only readily be met by fiber connections. A centralized RAN architecture can be attractive for scalability and resource flexibility, but would require many high end xHaul fibers to avoid poor latency and inadequate support for Layer 1 network functions.
Use of WDM-PON, where each wavelength carries 25Gbps, could reduce the fibers needed. But as Nokia makes clear, each case should be considered carefully on its merits, because WDM-PON will only be cost-effective when its capacity is required.