Passives Get Active

From oemagazine August 2001
01 August 2001
By Shri Balachandran

The access network is perhaps one of the most complicated elements of an operator's network. In the 1990s, the telecom landscape was transformed by the passage of the Telecommunications Act (1996) and the explosion of long-haul networks. Simultaneously, the percentage demand for broadband connections grew at a double-digit pace, which outstripped availability. Although the core network bandwidth is in the order of terabits and that of metro is in the order of gigabits, the access loop is still languishing at kilobits.

It is apparent that a disparity exists in the interconnectivity network that brings together subscriber-side demand and the network-side bandwidth supply. This is the challenge for the broadband access network.

DSL, point-to-point Gigabit Ethernet, SONET, satellite access and repeater T1 or T3 (1.54 Mb/s or 45 Mb/s) are some of the current breed of access technologies trying to solve this problem. In this article, we present a fiber-lean, ITU standards-based access solution: broadband passive optical networks (B-PONs).

issues with the access network

The biggest challenge in the access network is managing churn. Networks suffer customer turnover as well as changes in services, bandwidth, and service level agreements (SLAs). An access architecture must be able to rapidly add, delete, or change customers, services, bandwidth, and SLAs in a manner that does not affect users.

The access network is very spread out geographically. It must reach customers from a few hundred feet to tens of kilometers away from the point of presence, or the central office. Because the access network contains around 80% of an operator's network assets, any access network technology must provide low first-install cost and low incremental cost.

Managing the outside plant is expensive. To succeed, an access solution must be operationally simple, minimizing active components in the outside plant. It also must support both legacy and emerging network interfaces.

A majority of the customers have a need for mixed services (both voice and data services). Any access technology that does not support TDM, ATM, frame relay, and Ethernet traffic must offer a separate overlay network to carry the unsupported traffic.

Quick analysis reveals the deep chasm and disparity in the access space. DSL- and T1-based access are distance limited, costly to maintain, provide no support for emerging interfaces and data services (like VPN, VoIP), and are ineffective at handling the churn of customers, bandwidth, and services. Point-to-point Ethernet access is operationally simple but does not handle SLAs; nor does it support legacy interfaces or TDM, ATM, and frame relay traffic properly. It also requires a fiber-rich environment (which is seldom the case). SONET is strong in TDM, but it is operationally complex and expensive, it does not support mixed traffic (ATM, Ethernet, etc.), and it is poor at managing customer churn.

As a result, the largest local exchange carriers around the world teamed up to specify a fiber-based architecture that could meet the copper price points, handle churn, and support mixed traffic and links up to 20 km long while remaining operationally inexpensive. Their effort to develop the full service access network (FSAN) culminated in a new ITU standard known as B-PON. The market for B-PONs is slated to reach $1 billion by 2004.

passive aggressive

PON is a fiber-optic access solution that uses totally passive components in the outside system plant. It incorporates a head-end controller known as the optical line terminal (OLT) and a customer premise equipment or service node known as the optical network terminal (ONT). A PON network connects a single fiber from an OLT to multiple ONTs (see figure 1). The line terminal typically resides at a hub, such as a central office or point of presence, and connects into the metro network. One or more passive branching devices (typically an optical power splitter) in the fiber path provide point-to-multipoint connectivity between the OLT and multiple ONTs.

Figure 1. A typical PON sends a signal from the optical line terminal (OLT) at the central office to peripheral optical network terminals (ONTs) via passive splitters, reducing network cost and complexity.

The ONTs provide the subscriber drop interfaces to the customer. An ONT for fiber-to-the-building would be installed at or near the subscriber's premises in the service provider's equipment closet or with LAN servers; whereas one for fiber-to-the-curb might reside in the design area along with the digital loop carrier or wireless tower. A fiber-to-the-home device may reside near or inside a residence.

The ONT generally queues the data while the OLT schedules the traffic; thus, the cost of the ONT is low, which in turn yields a low incremental installation cost. For most applications, there are no intervening elements between the OLTs and the ONTs. The elimination of active elements between the two reduces the need for costly power, right-of-way space, and ongoing maintenance for electronics equipment. In addition, the point-to-multipoint connectivity of a PON also reduces fiber congestion in the OLT site and permits amortization of that equipment across a larger number of subscribers. The PON can use either a single power splitter or cascaded power splitters as branching devices for OLT-to-ONT communications.

Downstream data cells are broadcast by the OLT and received by all ONTs (see figure 2). The ONTs either decode or discard the cells based on the cell header addressing. A distance ranging protocol measures the optical distance from the ONT to the OLT to set delay compensation for differential distances between the various ONTs and the OLT. Churning key algorithm encryption ensures subscriber security.

Figure 2. Downstream, the signal passes from a single OLT to many ONTs. Because the ONT only queues the data while the OLT schedules the traffic, the ONT remains economical.

Upstream transmission (from the ONT to the OLT) uses a TDMA protocol, and the OLT has a burst mode receiver. Upstream cell transmissions from the ONTs are synchronized via a high-speed scheduling algorithm running within the OLT such that cells from one ONT do not collide with cells from the others. The OLT performs concentration and statistical multiplexing of user data from multiple PONs. The OLT then delivers the combined data via standards-based interfaces, such as OC-3, to the upstream network elements.

As mentioned earlier, B-PON is an ITU standard (ITU-T Rec. G.983). Although standards do not guarantee market acceptance, they provide an excellent starting point for examining the realistic minimum performance requirements that equipment vendors of fiber access systems should meet. The essential features of the current ITU-G.983 standard are:

  • Cell based on ATM standard
  • Symmetric (155.52 Mb/s both ways) and asymmetric operation (622.08 Mb/s downstream, 155.52 Mb/s upstream)
  • Range of up to 20 km with up to a 32-way split (limited by optical loss budget)
  • Single or dual-fiber operation.

Because the architecture was designed from the ground up, the FSAN standard specifies economical 1310-nm laser transmitters to send upstream from the ONT and 1490/1510-nm lasers to transmit downstream from the OLT. The small pen-size splitter typically costs only a few hundred dollars. PONs operate efficiently in a fiber-poor environment by sending upstream and downstream traffic on the same fiber. Thus the first install and incremental installation cost of a PON is very low.

advances in PON

Although FSAN provides a great framework for B-PON, it is just the beginning. The FSAN standard for PON is 155 Mb/s bi-directional, but emerging PON players have introduced symmetric higher-speed PONs operating at OC-12 (622 Mb/s) and OC-24 (1.25 Gb/s) data rates.

Companies are also working with dynamic bandwidth allocation, which exploits the bursty nature of data traffic by dynamically reassigning the unused bandwidth from a low-traffic connection to a connection with higher instantaneous bandwidth needs. For example, recently available PONs technology can adjust to any granularity of the PON bandwidth and dynamically redirect both light paths within 250 µs. However, this dynamic bandwidth management continues to protect provisioned services such as DS1 voice.

Adding extra wavelengths on the existing fiber plant can support customers requiring very high data rate, point-to-point bandwidth. The WDMA capability closely matches a "pay-as-you-grow" service-provider business model because WDMA channel technology is only provided when and where it is needed to support enhanced services to a specific subscriber.

Because TDM-voice transmission requires a constant stream that cannot be blocked by data traffic in tight bandwidth situations, the ability to prioritize information is essential. Terawave's TeraPON system provides traffic policing, on a per-flow basis, to ensure information flows stay within the boundaries of the negotiated traffic contract. The system can support SONET-style sub-50 ms path protection switching. Protection schemes that are supported include equipment protection, facility protection, and their combination.

Clearly, there is no optimal network that addresses all applications and scenarios. B-PON provides the compromise solution that scales in bandwidth, supports churn without service disruption, provides SONET-like circuit protection, and enables both voice and data services over an infrastructure that is economical to install and manage.

The passive nature of the outside system plant makes PONs inexpensive to maintain in the long run. PONs not only competes with but also complements other existing technologies. The architecture fundamentally supports residential, business, service-provider equipment traffic backhaul, lambda services, and video delivery over the same fiber infrastructure from day one.

Service provider fiber deployments to business subscribers are accelerating. The lower costs and added benefits of new FSAN-based fiber access technologies will further accelerate the progression of fiber deeper into the access network. oe

alphabet soup

Learn the optical-networking lingo:

ATM-asynchronous transfer mode
DSL-digital subscriber line
FSAN-full service access network
ITU-International Telecommunications Union
OLT-optical line terminal
ONT-optical network terminal
SONET-synchronous optical networking
TDM-time division multiplexing
TDMA-time division multiple access
VoIP-Voice over Internet Protocol
VPN-virtual private networks
WDMA-wavelength division multiple access

--S. B.

making the case

You don't need philosophers' stone to turn your idle glass into gold. A research study done by NEC eLuminant Technologies Inc. (Chantilly, VA) has shown that in most cases, the payback period for PON deployment can be less than a year.

Over the past decade, billions of dollars have been spent on access fiber buildout, connecting large buildings and access points for new residential neighborhoods. However, a large quantity of fibers are yet to be utilized because they cannot quite reach the end users, especially the 6.5 million small-to-medium businesses that are still underserviced in the era of broadband access. Though small-to-medium businesses represent a large area of revenue growth for service providers, the economics of fiber deployment have hindered the migration of their access network to the PON technology.

eLuminant conducted a study in Herndon, VA, an area where there is a high concentration of businesses of various types. The survey included 179 businesses residing in 52 buildings in seven different locations. Among the 179 businesses, 170 of them could be serviced by a PON system with single-fiber connections from existing CLEC or ILEC fiber networks at a cost of approximately $2.8 million.

Once in place, the network would generate an estimated revenue of $564,000 per month for service providers. The deployment investment would be mainly in building access, equipment, and construction of optical distribution network when necessary and would be paid back in an estimated average of five months, assuming all the businesses in the survey connected to the network.

The results indicate that the burden of initial buildout cost is not as prohibitive as people have feared. Service providers can take advantage of the significant economical benefits of the PON architecture and turn the existing dark fibers into a revenue-generating source.

—Jim Holley, NEC eLuminant Technologies Inc.

Shri Balachandran

Shri Balachandran is senior product line manager with Terawave Communications Inc., Hayward, CA.

Recent News
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?