Carriers and access bandwidth providers have undergone a somewhat frustrating churn through passive optical network (PON) architectures over the last decade in an attempt to get ahead of the bandwidth-escalation curve. Yet, despite billions in infrastructure investments, they have never quite caught up, as the access bandwidth curve continues to grow steeper. Fortunately, WDM-PON technology provides the scalability that access network operators now understand they need to get ahead of bandwidth demands.
Yet initially, the WDM-PON category was defined by approaches that used proprietary technologies and complex architectures in an attempt to lower initial capital expenditures (capex) per subscriber. These early strategies were at best misguided, as they missed the whole point of WDM-PON: Using an access architecture that is compatible with the remainder of the network enables substantial savings and ensures scalability.
Today, a more mature WDM-PON is emerging that leverages off-the-shelf components and other existing technology, as well as standard WDM and Ethernet protocols. Sometimes referred to as WDM access, sometimes next-generation access (NGA), this newest spin on WDM-PON puts network operators cost-effectively in front of the bandwidth curve. It also offers low total cost of ownership (TCO) and flexibly accommodates ongoing growth in bandwidth-intensive, runtime-sensitive service traffic.
The true cost of PON
PONs are “passive” in that they share bandwidth among users via an unpowered, inactive device and “optical” in that fiber is the connection medium. PONs have experienced widespread deployment since the 1990s as a way of sharing bandwidth among subscribers. Seeking to keep in check the number of feeder fibers required across a large geographical area, vendors introduced a series of PON architectures based on traditional TDM bandwidth sharing. Seduced by the ease of migration and the promise of “lowest cost to upgrade,” carriers slipped into a pattern of network upgrades and overlays — Asynchronous Transfer Mode PON (APON) to Broadband PON (BPON), Gigabit PON (GPON), Ethernet PON (EPON) and 10-Gbps PON (10GPON, 10GEPON), with NG-PON on the horizon.
These TDM-based PONs used dynamic bandwidth allocation (DBA) algorithms to mitigate the distribution of available capacity among users competing for the uplink. If 10 subscribers shared the same 10-Mbps link, each user received 1 Mbps of bandwidth in times of high network utilization. The network also could burst temporarily higher in moments of low utilization. TDM approaches with DBA fit the early access bandwidth landscape.
However, capacity needs have steadily swelled, and the nature of bandwidth consumption has changed significantly since PONs first appeared. Traffic patterns were once heavily asymmetrical in favor of downloads from network to user. Now, there’s more balance between upstream and downstream bandwidth, as user uploads and peer-to-peer transfers have increased. And geographical subscriber density has steadily increased as more homes passed now opt into fiber-to-the-X (FTTx) offerings. With these changes, DBA techniques have demonstrated their weaknesses, and once again TDM-based PONs are showing strain.
The real issue here is that legacy PON architectures never put network operators in front of the bandwidth growth curve (see Figure 1). APONs fit easily into existing carrier ATM networks, but they were unable to accommodate bandwidth growth once Internet surfing became mainstream. BPONs delivered the incremental capacity gain carriers required at the moment, but they were unable to accommodate growth in video and triple-play services. GPONs provided video relief, but soon carriers were faced with high-definition television (HDTV) and quad-play, and suddenly work began on 10GPONs. Today, with no sign of access bandwidth demand slowing, there is work on NG-PON, which in some variants tries to add wavelength channels to existing GPONs.
Figure 1. PON timeline
Churn through TDM-based PON architectures drove not only significant operating expenditures (opex) with each generation, but also, paradoxically, hit the very capex budget that network operators had sought to minimize in the first place. Cost-weary network operators today are more sensitive to the TCO of PON architectures — not just upfront cost. While WDM-PON is more expensive in terms of capex than any one of these PON generations by itself, it is substantially cheaper than any two or three of the protocols and certainly costs less in the long run. That’s because WDM-PON breaks the upgrade cycle and offers an access architecture that scales. It enables carriers to get ahead of the bandwidth demand curve.
Perfecting the WDM-PON
The coupling of protocol-agnostic, bandwidth-increasing multi-wavelength encoding and optical amplification has earned WDM a steadily expanding role in optical networks — from core to metro and, now via the WDM-PON, to access.
Network operators were initially skeptical of WDM-PON and the notion of giving each subscriber a dedicated optical wavelength, as it sounded expensive. In an effort to address this perception, early attempts at lowering capex per subscriber leveraged proprietary, immature lab technologies such as broadband light sources, injection-locked lasers, reflective semiconductor optical amplifiers, etc. These approaches suffered many drawbacks, such as lack of path protection, poor noise floors, and the inability of injection-locked lasers to operate at higher bit rates. While these efforts were admirable technically, additional proprietary approaches were not the answer.
A new WDM-PON approach (see Figure 2) overcomes such scalability limitations and delivers ongoing TCO advantages. It’s based on the belief that the most efficient place to process Ethernet packets is not an access cabinet but an enterprise data center or carrier central office.
Figure 2. WDM next-generation access (WDM NGA)
A “Flexible Remote Node” (FRN) — fully passive and temperature-hardened — is fielded to perform access between existing WDM transport and customer-premises equipment (CPE). Wavelength-stabilized lasers are required only in the temperature-controlled network end points. Bandwidth is divided not by reserving time slots, as with TDM-based PONs, but by both wavelength-based separation and Ethernet Virtual Connection (EVC) assignment.
Optical wavelengths may be reserved not only per user, but also per group of users or even application. Within these wavelengths, EVCs may be assigned to users, applications, and traffic classes. Using Ethernet demarcation techniques, each EVC has guaranteed bandwidth that can be remotely provisioned, monitored, and tested in-service. The FRN aggregates the wavelengths onto a single fiber for transport back to the core. This approach reduces to a single strand the amount of feeder fiber required from core to subscriber edge, and a second fiber may be used for 1+1 path protection.
Scalability is infinite, and particular customer links can be upgraded individually. Because this approach does not depend on proprietary fiber, channel grids, or wavelength spacing, traffic can travel throughout the standard WDM transport network without protocol conversion or aggregation. Access, metro, and core networks are seamlessly unified in an infrastructure of unprecedented power, flexibility, and manageability.
Varied, sustainable cost-efficiencies
In comparing access techniques, it is important to examine the entire network architecture. WDM-PON achieves significant cost-efficiencies through a variety of ways that all add up to a sustainably low TCO:
- Infinite scalability — Bandwidth demands should continue to rise for the foreseeable future. Service providers are angling to deliver 100 Mbps to each residential subscriber and 10 Gbps to each business customer without disruption or bottlenecks. WDM-PON offers infinite scalability with flexible bandwidth upgrade capabilities per user, group of users, or class. WDM-PON ends the PON churn and the recurring capex investment cycle it spawned.
- Consolidating networks — Service providers have typically maintained separate networks for residential, business, and wireless backhaul traffic. WDM-PON’s secure, hardware-based traffic separation and guaranteed bandwidth profiles allow different network types to be collapsed onto the same physical infrastructure, while still operating as distinct infrastructures in the virtual domain. This feature significantly reduces redundant overhead costs.
- Reducing protocol conversions — Access architectures today typically include multiple technologies managed under different tools. From subscriber edge to network core, a packet will undergo multiple protocol conversions and encounter many redundant layers of packet aggregation. Not only is this highly inefficient, it also jeopardizes overall service performance. By eliminating redundant aggregation layers and unnecessary protocol conversions, WDM-PON reduces opex and capex.
- Native Ethernet — Network operators are eager to embrace the simplicity and cost-efficiencies of native Ethernet transport. EPON’s were a first attempt at native Ethernet in the access space. However, EPON’s were still TDM based and lacked the necessary operations, administration, and maintanence (OAM) protocol extensions. WDM-PON brings native Ethernet — underscored by key carrier-class enhancements — to NGA networks. Beginning its life as Ethernet, the packet remains Ethernet end-to-end across the network.
- Investment protection — Network operators can cleanly and gradually migrate to WDM-PON without an immediate need to change out CPE. WDM-PON elegantly and cost-effectively leverages existing infrastructure, requiring at first only the deployment of an FRN somewhere in the network (a CPE site, local exchange, wiring closet, field box, etc.). The operator may choose to selectively install WDM-PON-enabled SFPs into existing CPE for additional functionality, though this is not a hard requirement.
- Fewer local exchanges — While bandwidth consumption continues to skyrocket, revenue per bit is actually falling. Faced with increasing margin pressures, network operators want to reduce the number of local exchanges required to reach a group of subscribers — and in the process positively influence a host of opex categories such as real-estate, electricity, and maintenance costs. WDM-PON supports long fiber spans (100 km or more), enabling network operators to deliver more bandwidth to more users from fewer active sites deeper back in the network.
A broad view of TCO
A comprehensive analysis of TCO — taking into account the end-to-end transport of a packet from core to user — revealed the cost-efficiencies of adopting WDM-PON in access.
WDM-PON was evaluated in relation to two other prevalent access architectures: fiber-based GPON and copper-based VDSL2, both connecting back to the core via WDM transport. In each scenario, TCO of the access and first layer of WDM aggregation infrastructure was calculated over 25 years. A large-scale rollout of broadband access services — a guaranteed bit rate of 80 Mbps to 1 million residential subscribers and from 1 Gbps to 10 Gbps for 10,000 business customers — was assumed.
Even with conservative assumptions, the network using WDM-PON in the access delivered a TCO smaller than either VDSL2 or GPON. The WDM-PON consumed half the energy and 70 percent less opex than VDSL2, and a quarter the energy and 60 percent less opex than GPON.
The TCO analysis revealed a clear advantage for WDM-PON, quantifying network operators’ hard-learned lesson over the last decade — initial capital investment is only one part of the story (and a decreasing one at that, over time). WDM-PON offers network operators the unprecedented opportunity to get in front of the bandwidth curve once and for all, avoid subsequent iterative PON investments, and benefit from ongoing opex savings over the lifetime of the network.
Jim Theodoras is director, technical marketing, ADVA Optical Networking