Scott Feller
Product Manager
GPON & EDC

Gigabit passive optical networking (GPON) is the next big thing for telecom carriers hoping to bring triple-play services to their subscribers. However, GPON chipsets have been late to market, and system designers are still grappling with the challenge of creating optical network termination (ONT) and optical line terminal (OLT) products that maximize performance and deployment flexibility while minimizing costs. The last of the four main documents of the International Telecommunication Union-Telecommunication Standardization Sector (ITU-T) GPON standard was approved in June 2004. While early GPON systems have been built, we will not see widespread availability of merchant silicon and optics until the second half of 2006. Several factors have caused the delay, including the technical challenges of GPON, competing technologies, the life cycle of existing technologies, and the strategic position of the companies with the technology to build GPON products.

While GPON has been in development, the early PON technologies of broadband PON (BPON) and Ethernet PON (EPON) have been deployed with significant market success. In many parts of the world, including Japan, the technology of choice has been EPON. A major acceptance milestone was recently passed in Japan, where the number of new subscribers for PON has exceeded the number of new digital subscriber line (DSL) subscribers. In the United States, Verizon has been aggressively deploying BPON. These deployments clearly show that PON can be effective at providing triple-play services.

With three PON technologies, however, the debate becomes which technology is the most effective for service provider deployments. GPON has the advantage of being developed after EPON and BPON; GPON developers had the advantage of looking at the EPON and BPON standards, picking up their strong points, and filling in the areas that needed improvements. Being the third standard also puts the burden on GPON to improve enough to displace two widely deployed technologies.

Many comparisons of the technologies focus on comparing the standards themselves instead of comparing the popular implementations of the standards. However, it is useful to compare the technologies in terms of their cost of deployment, revenue-generating capability, capability to provide services, and relative ease of management in a network. Since the implementation can differ from the standards, it is more interesting to compare the common implementations of the standards.

EPON is one example of the implementation differences. The EPON Institute of Electrical and Electronics Engineers (IEEE) 802.3ah standard does not define methods for security, video overlay, or quality of service (QoS). However, the system developers of EPON have designed these features into their systems. While GPON has defined methods in the standard, the EPON system developers have had to use proprietary methods to implement these features. This proprietary nature of EPON implementation points to the issue of interoperability. In the field today, there are four major implementations of the EPON standard at the media access control (MAC) level. This impact can be minimal if the same silicon and devices are used throughout the PON. However, this can prevent systems providers from upgrading to new suppliers for cost or technology improvements.

One of the key differences in the technologies is the rate of revenue-producing bits, which is based on several factors, including the data encoding method, bit rate, number of subscribers, and the size and type of data. Figure 1 shows the downstream data rate (from the central office to end users) for the different technologies.

For this comparison, the data being sent is assumed to be Ethernet packets with a payload of 512 bytes. While different mixes of traffic will cause the ratio of overhead to change, this figure shows two major differences in the technologies that also exists for the different traffic patterns.

First, the data rate being implemented is different: GPON systems are 2,488 Mbps, EPON is 1,250 Mbps, and BPON is 622Mbps. As one would expect, the increase in data rate significantly increases the amount of data available to the end users.

The second issue is that GPON and BPON use a different method of encoding than EPON. GPON and BPON encode the data by scrambling the bits by a known polynomial. This coding technique adds 0 bits to the data stream. EPON uses 8 B/10 B encoding, which requires 2 extra bits for every 8 bits of data, resulting in a 20 percent tax on the effective bandwidth required.

It is also interesting to note that the packet type used for this comparison is Ethernet packets. It is quite common for EPON proponents to point out that EPON is Ethernet and has a significant advantage as the access and backhaul networks move toward IP-based infrastructure. GPON has taken into account the importance of Ethernet and will actually transport the Ethernet packets more efficiently than EPON. For an Ethernet packet, there is typically 39 bytes of overhead, such as interpacket gaps, addresses, headers, and frame markers. GPON uses a technique called GPON encapsulated mode (GEM) that allows the Ethernet packets to be encapsulated and strips the overhead that is unnecessary for PON transport. This technique cuts the overhead to 23 bytes for Ethernet packets. The combination of these factors gives a significant advantage to GPON. In the upstream direction (from the user to the central office), there are additional effects to consider in addition to the coding and bit rate. The bit rate difference is not as significant in this case. While the GPON standard can operate at 2,488 Mbps in this direction, the first implementations are using 1,244 Mbps to keep down the cost and time to market of the analog and optical components. The current implementations for the upstream data rate transmissions are 1,244 Mbps for GPON, 1,250Mbps for EPON, and 155 Mbps for BPON. Again, one of major differences is in the coding of the data, with EPON requiring a 20 percent tax on the data rate.

Another area where the standards differ is in the burst mode capability of the devices. In any PON system, several end devices need to transmit to the central office. Each end device is scheduled a time slot when it can use the fiber. This creates complexities for a central office device trying to interpret the multiple streams. To interpret the data blocks, guard times and acquisition sequences must be added to the data transmission. In GPON, guard and acquisition time has been reduced to less than 100 ns, while in EPON, this time is typically in the range of 750 ns.

The extra bandwidth available in GPON can be used several ways. For example, the bandwidth could be used to reduce the cost per user of the system, enabling more subscribers to share a single fiber to reduce the cost per subscriber. This could be the case where a single endpoint is serving multiple subscribers such as a multiple-tenant unit or an apartment complex. Another area for cost savings is with the use of IPTV. Today, most of the EPON and BPON systems use an extra wavelength in the optical module to transmit the video data. GPON has enough bandwidth to eliminate the extra wavelength and reduce the number of frequencies required for the optical modules to one upstream and one downstream. This will reduce the complexity of the optical devices, which are one of the major cost components in the system.

The most appealing part of the extra bandwidth is the potential for service providers to bundle and offer new services. This allows providers to offer premium services such as video on demand, video telephones, and on-line gaming. Naturally, these new services can bring in new revenue. The additional benefit is that service providers are getting less churn from customers due to the availability of more desirable services.

There are many aspects for service providers to consider when selecting a PON technology. The EPON and BPON technologies exist and are being successfully deployed throughout the world. GPON has taken technology leaps to increase the capability of the PON networks. The technology pieces required to build GPON systems now exist. As multiple suppliers release higher-performance and lower-cost versions of these systems, the market will shift to GPON.

Educational content provided by the Cortina Systems