Technology developments are making fibre more cost effective

The key challenge facing operating companies is the huge investment needed to install fibre in the first place. But, with the emergence of the latest PON architecture, dubbed NG-PON2, the business case for fibre based access has become more attractive. NG-PON2 will enable operators to run several services over one network and promises dedicated wavelength links to meet more demanding service requirements. The Full Service Access Network (FSAN) group is defining NG-PON2. FSAN, comprising leading operators and telecom equipment makers, has added wavelength division multiplexing (wdm) to the time sharing scheme used by existing PON standards, to create the TWDM-PON scheme for NG-PON2 (see fig 1). Standards compliant NG-PON2 equipment is expected to become available by 2014 and to be deployed by operators from 2015. Image courtesy of the OSA FSAN will work through the ITU, which has previously turned its definition work for the 2.5Gbit/s Gigabit PON (GPON) and 10Gbit/s XGPON into standards. In its basic configuration, NG-PON2 will use four wavelengths, offering a capacity of 40Gbit/s. Support for eight (80Gbit/s) and 16 wavelengths (160Gbit/s) is being considered, with some wavelengths available for dedicated point to point services. Each wavelength supports up to 10Gbit/s downstream and 2.5Gbit/s upstream. However, 10Gbit/s symmetrical services (same rate communications between the central office and the end points) will be available for business users. "The idea is to reuse as much as possible of the XGPON protocol in TWDM-PON and to carry that protocol on multiple wavelengths," said Derek Nesset, co chair of FSAN's NGPON task group. Other NG-PON2 specifications include the support of at least 1Gbit/s service per end point device and a reach of 40km. NG-PON2 will also support links of 60 to 100km, but that will require reach extension technologies such as optical amplification. The exact positioning of NG-PON2's wavelengths within the fibre's spectrum is still to be determined. NG-PON2's additional support for point to point links – wdm overlay – will enable mobile backhaul and businesses connectivity services. A key operator requirement in defining NG-PON2 has been to avoid modification of the deployed PON infrastructure. "The goal is to enable operators to deploy something soon and to leverage all that has been done with existing PON protocols," said Nesset. "We didn't want, from a time and a cost point of view, to start with a clean sheet of paper." For example, having to add wavelength selective filters to existing passive infrastructure – optical splitters – to enable TWDM-PON would be costly. Instead, changes are confined to the end equipment: the central office's optical line terminal (OLT) and the end point's optical networking unit (ONU). The OLT will need to support the 4, 8 or 16 wavelengths using lasers and photodetectors, as well as optical multiplexing, while the ONU will require a tunable laser and a tunable filter to set the ONU to the PON's particular wavelength. Operators will likely deploy NG-PON2 in a variety of ways to meet their differing requirements. Those yet to adopt PON technology may view NG-PON2's extended reach as a way of consolidating their network by reducing the number of central offices they manage. Operators with existing PON deployments will use NG-PON2 to boost capacity, while consolidating business and residential services onto one network. US operator Verizon has GPON and by adopting NG-PON2, can avoid the intermediate upgrade stage of XGPON. "The [NG-PON2] technology choice allows us to have a single platform supporting business and residential services," said Vincent O'Byrne, Verizon's director of technology, wireline access. "We can split the four TWDM wavelengths: we could have a 10G/10G service or 10 1G/1G services and, in time, have also residential customers." Even if the technology for NG-PON2 has been chosen, technical challenges remain. One key issue is the wavelength plan: and many options are being considered (see fig 2). GPON and XGPON1 already use separate frequency bands, while operators such as Verizon use a dedicated wavelength at 1550nm for rf video transmission. Image courtesy of the OSA One proposal is for TWDM-PON's wavelengths to replace those of XGPON. Alternatively, unallocated spectrum could be assigned to ensure coexistence with GPON, rf video and XGPON. However, such a scheme will leave little spectrum available for NG-PON2 and some element of spectral flexibility may be required to accommodate the various coexistence scenarios in operator networks (see fig 3). Image courtesy of the OSA Another spectrum related factor is how widely the wavelengths will be spaced – 50GHz, 100GHz or the most relaxed 200GHz spacing are all being considered. The tradeoff here is spectral efficiency against hardware design complexity and cost. Meanwhile, TWDM-PON will require tunable optical components. "The ONUs should have a cost similar to that of a VDSL or a GPON modem, so there is a challenge there for the [tunable] laser manufacturers," said Nesset. Tunable laser technology is widely used in optical transport and high volumes will help the economics, but this is not the case for tunable filters. He says: "There are few [tunable filter] manufacturers in the market." The size and power consumption of TWDM-PON silicon are further challenges. "In the long term, you want to achieve the same density [as GPON]," Nesset concluded. "Just because you've going to four times the capacity, you don't want to increase the size of the OLT in the central office." Passive optical networks A traditional passive optical network (PON) comprises a single fibre whose bandwidth is split into time slots shared between 32 or 64 end points. At one end of the fibre, an optical line terminal (OLT) coordinates connection to the end points. At each end point, an optical networking unit (ONU) receives its data and transmits on its allocated time slot. Such a PON is time division multiplexed. In between the OLT and ONUs are optical splitters which connect the shared fibre to the fibres linking the end points. Sharing a fibre for the bulk of the link's reach, before reaching the splitters, results in a cost effective architecture. A PON typically spans 20km with a downlink capacity, to the end users, of 2.5Gbit/s for GPON or 10Gbit/s for XGPON1. Upstream rates are 1.25Gbit/s and 2.5Gbit/s, respectively. NG-PON2 standardisation FSAN has a long history of working with the ITU to define PON standards, starting from Broadband PON and Gigabit PON (GPON) to the 10Gbit/s GPON extension known as XGPON1 that operators are now planning to deploy. A competing PON scheme – Ethernet PON (EPON) and 10Gbit/s EPON – is standardised by the IEEE. FSAN has submitted a NG-PON2 requirements document to the ITU. "This sets the framework: what is it this system needs to do?" said Nesset. "This includes what client services it needs to support – Gigabit Ethernet and 10G Ethernet, mobile backhaul latency requirements – high level things the system specification will then meet." A detailed requirements document was submitted in June 2012, as was a preliminary specification for the physical layer. These will be followed by documents covering the NG-PON2 protocol and how the management of the PON end points will be implemented.