The ever-increasing number of end users accessing the Internet and taking advantage of today's data-transport capabilities has ignited an explosive demand for bandwidth. The growing popularity of such applications has also resulted in longer wait times. Many types of next-generation access systems are helping to improve this situation by pushing optical fiber closer to the home or business premises.
Fiber-to-the-home/business (FTTH/B) network architectures offer major bandwidth improvements compared with alternative optical access network architectures that rely on copper or coaxial cable for the final distribution segment to the customer premises. In addition, small businesses, home offices, and residential users in new development areas can, and should, begin reaping the benefits of FTTH/B, with a new optical access technology currently being standardized within the ITU-T (G.983), generally referred to as ATM PON (Asynchronous Transfer Mode-based passive optical network).
Today's increasing demand for bandwidth has created the need for FTTH/B. While asymmetric digital subscriber line (ADSL) technology can provide 1 to 8 Mbits/sec of downstream capacity directly to a customer premises, this is currently the limit for most existing copper loops. A hybrid fiber/coaxial-cable (HFC) network, meanwhile, can provide 30- to 40-Mbit/sec total capacity downstream using a single 6-MHz analog channel spectrum (typically shared by 100 to 250 homes), but HFC has extremely limited upstream capacity.
Based on today's user needs, these data rates should be sufficient until about 2004. By then, increased multimedia and video services will drive the need for a new access network architecture such as fiber-to-the-cabinet (FTTCab) or FTTH/B.
FTTCab can support very high-speed DSL (VDSL) access, which will increase the data rate to about 13 Mbits/sec at 4500 ft from the customer premises. Moving the optoelectronics to within 3000 ft of the customer premises could increase the bandwidth to up to 26 Mbits/sec, a data rate that should last until about 2010 at today's rapid growth rate. FTTCab is simply an interim step to FTTH/B, the eventual solution for the growing bandwidth needs. As a result, telephone companies are considering running fiber all the way to the home or business as soon as possible.
Immediately deploying ATM PON may not always be the answer, however. For example, ATM PON may not be practical in major metropolitan and commercial areas where excess copper and/or Synchronous Optical Network (SONET) rings are readily available. Yet, an FTTH/B architecture based on ATM PON does make sense in lower-density serving areas, such as outlying residential/suburban areas, where dynamic and unpredictable demand for voice/data services from small offices/home offices (SOHOs) and small independent or shared tenant business premises (such as strip malls) present significant challenges to local-exchange carrier network planners.
ATM PON allows both service providers and end users to gain the benefits of FTTH/B. It provides the access piece for the all-fiber network.
ATM PON is a point-to-multipoint, cell-based, optical-access architecture that facilitates broadband communications between an optical line terminal (OLT) at the central office and multiple remote optical network units (ONUs) over a purely passive optical-distribution network with a reach of up to 20 km.
The first operational prestandard ATM PON system was developed by Fujitsu Ltd. in cooperation with Nippon Telegraph and Telephone (NTT) of Japan. This system began commercial operation in September 1997 with leased-line business services in Japan. Currently, there are hundreds of systems deployed and actively providing service. Fujitsu is still the sole provider of fully functional OLT equipment to NTT.
ATM PON was first proposed as a standardized FTTH solution in the early 1990s by the Full-Service Access Network (FSAN) initiative, which comprised 14 telephone companies from around the world. FSAN is an effort to set common requirements for full-service optical-access networks among all operators globally. The requirements generated within this group were forwarded to the ITU-T Study Group 15 as recommendation G.983.1, which was formally approved last October.In an ATM PON system, a maximum of 64 ONUs can share the capacity of a single fiber using ATM transport and passive optical splitter/combiner technology (see Fig. 1). Full-duplex (simultaneous upstream and downstream) transmission via a single fiber facility can also be achieved using independent wavelengths (1310/1550 nm) for each direction. In FTTH/B network architectures, the functions of an ONU and a network termination unit are often integrated into a single unit referred to as an "optical network termination" (ONT).
The OLT broadcasts a downstream signal consisting of a continuous time-division multiplexed stream of fixed-length (53-byte) time slots carrying physical-layer operations, administration, and maintenance (PLOAM) cells, ATM cells, and idle cells for cell-rate decoupling. The nominal bit rate of the downstream signal can be either 155.52 or 622.08 Mbits/sec. The maximum downstream cell-transfer capacity on a single fiber (excluding a fixed reserved capacity for PLOAM cells) is 149.97 or 599.86 Mbits/sec, depending on the nominal line rate chosen for the downstream facility.
A time-division multiple-access (TDMA) technique is used for upstream communications using 56-byte time slots with burst-mode synchronization performed by the OLT receiver. Upstream media-access coordination and control (MAC) is provided by the OLT, which issues explicit messages via downstream PLOAM cells that grant access to each ONU for transmission within specified upstream time slots.
When an ONU is granted access to an upstream time slot, it transmits a 3-byte header followed by a 53-byte cell at a nominal line rate of 155.52 Mbits/sec. Due to the overhead bytes, the upstream cell-transfer capacity is limited to 147.2 Mbits/sec and is shared among the ONUs based on their allotted upstream bandwidth. Some of this upstream capacity is needed by the OLT for physical-layer overhead and MAC control (the actual amount of upstream overhead bandwidth required is implementation-dependent).
A framing structure is applied to both the downstream and upstream signals to support the TDMA operation. The downstream frame format for the 155.52-Mbit/sec line-rate option consists of 56 consecutive 53-byte time slots containing two evenly spaced PLOAM cells 28 time slots apart (see Fig. 2, which also illustrates the 622-Mbit/sec frame format). The upstream frame format consists of 53 consecutively numbered 56-byte time slots. The two downstream PLOAM cells collectively contain 53 upstream grants (each associated with a specific time slot in the next upstream frame) along with additional physical-layer OAM information broadcast to the ONUs.
In addition to upstream bandwidth allocation via fixed-length time slots, a "divided-slot" mechanism is defined to allow a single upstream time slot to be subdivided into multiple consecutive mini-slots (see Fig. 3). The OLT can allocate one or more divided slots to a group of ONUs, each of which can transmit one or more mini-slots in a prescribed sequence. Each mini-slot consists of a 3-byte header followed by a fixed-length payload (configurable to any integer number of bytes that can be encapsulated within the available divided-slot payload). Configuration of the divided-slot time slot and its associated mini-slots is accomplished via PLOAM messages exchanged between the OLT and associated ONUs.
It has been agreed that among other potential applications, mini-slots will be used by ONUs to submit on-demand bandwidth requests to the MAC arbitration controller within the OLT. However, the format and semantics for these on-demand bandwidth requests have not been defined. Furthermore, it is unclear as to whether an on-demand MAC protocol will ever be subject to standardization.
The OLT establishes an upstream physical-layer OAM channel with each individual ONU by issuing specific grants requesting the transmission of a PLOAM cell. The rate of this upstream PLOAM channel is directly controlled by the OLT; but an upstream PLOAM cell will be requested at least every 100 msec. Upstream/downstream PLOAM cell flows facilitate direct OAM message exchange between the OLT and each individual ONU for physical-layer management and control operations.
Due to the physical convergence of the upstream burst transmissions from each ONU via one or more passive optical splitter/combiner elements, the timing of each ONU transmission must be precisely synchronized with delay compensation to account for unequal distances between the OLT and each individual ONU. To accomplish transmission-delay equalization among all ONUs, the OLT must perform a ranging procedure (to measure the logical reach distance to and from each ONU) and assign a specific equalization delay adjustment to each ONU. When granted access (under normal operation), an ONU synchronizes its upstream transmission based on timing recovered from the downstream signal timing and applies its assigned equalization delay adjustment.
The virtual-path identifier (VPI) field of each ATM cell is used as the multiplexing identifier for associating unique cell flows between the OLT and each specific ONU. That is, an ONU only processes downstream ATM cells containing VPI values that have been explicitly assigned to them (ignoring all other ATM cells broadcast on the shared downstream channel). Point-to-point virtual-path (VP) connections can be established between the OLT and any ONU by assigning a unique VPI value to a single ONU.
A point-to-multipoint VP connection (with the OLT serving as the root node and one or more ONUs as the leaf nodes) can be established by assigning the same VPI value to all associated leaf-node ONUs. For an ONU/ONT to support a standard ATM UNI as a subscriber interface, it must perform VPI translation to maintain local significance of the VPI space.
Due to the multicast nature of the PON facility, downstream cells are broadcast to all ONUs. An optional "churning" function can be individually enabled for point-to-point VP connections to provide additional security against potential eavesdropping. This is a byte-oriented encoding scheme based on a private key (churn key) exchanged between a given ONU and the OLT. The churn key is generated by the ONU and provided to the OLT on request. As an added security measure, the OLT requests the churn key be updated with a new value on a periodic basis (with at least one update per second).
The physical characteristics of the optical splitter/combiner technology induce severe attenuation on the signal transmission from the upstream transmitter of one ONU to the receiver of another. Therefore, there is no need to churn upstream transmissions and it is not likely that one ONU could intercept a private churn key exchange between another ONU and the OLT. If churning is not considered sufficient security for a given service, a more suitable encryption mechanism must be employed at a higher layer.
The most obvious benefit that ATM PON offers to end users is a vast increase in the amount of bandwidth delivered to a home or business premises. It opens the door to more bandwidth-intensive applications such as video-on-demand (VOD) and will eliminate any access network-induced wait time experienced when surfing the Web.
Initial G.983 systems will likely only support the symmetrical line-rate configuration (155.52 Mbits/sec upstream and downstream) with statically provisioned upstream bandwidth allocation and limited split ratios. Future releases will add support for an asymmetric line-rate configuration (622.08 Mbits/sec downstream/155.52 Mbits/sec upstream), MAC enhancements such as supported dynamic session-oriented bandwidth reservation, and arbitrated on-demand bandwidth allocation and support for up to 64 ONUs per PON facility.
Using VOD as an example, the benefit of these enhancements can be better understood. With the increased downstream bandwidth capacity, several subscribers would each be able to order a unique movie requiring, as an example, 6 Mbits/sec of bandwidth. A bandwidth reservation would be granted to one or more ONUs for the duration of the movie, thus reducing the total bandwidth-contention pool remaining for on-demand access by all ONUs on the same PON facility. At the completion of a movie session, the bandwidth reserved for this session would then be returned to the pool of total bandwidth available for on-demand access.
Although ATM PON will enable higher bandwidths to be delivered directly to the customer premises without an increase in cost to the subscribers, it probably will not lower their overall cost, either. In some cases, such as for SOHOs in residential areas, ATM PON may provide the opportunity for a lower-tariffed service. Even though the cost for access may not be reduced, the cost per bit transported will improve significantly with the introduction of ATM PON.
While cable modem service can provide more bandwidth for end users' money compared with Integrated Services Digital Network and ADSL services, it does not offer the secure and deterministic environment that ATM PON provides. FTTH/B provides a dedicated optical connection that is secure all the way to the central office. This type of security is particularly important to SOHO users with a home-office connection to the corporate local area network.
If ATM PON and coaxial cable (used for cable modem service) are both broadcast media, why is ATM PON more secure? The passive optical splitter/combiner technology used by ATM PON is highly directional, so attenuation is relatively low upstream (from every ONU to the OLT) but very high from one ONU to another. Coaxial cable uses a multi-tap (RF splitter/combiner) element to provide service to an individual customer premises. This element is not as highly directional as an optical splitter/combiner and therefore creates a much less secure environment.
The maximum length of the coaxial drop cable in an HFC system is a few hundred feet, resulting in distance constraints on the assignment of individual drops to one coaxial segment (RF splitter/combiner group). In comparison, aside from an overall optical-loss budget, ATM PON has no constraints on the length of a drop segment or the placement of the splitter/combiner elements (only the maximum fiber distance any single ONU may be located from the OLT). This freedom allows a splitter/combiner group to easily be re-engineered to meet any specific demands in terms of absolute reserved bandwidth or privacy requirements for a single home or business premises.
With the OLT at the central office and an ONT in the subscriber's home or business premises, there are no active electronics (such as digital loop carrier or FTTCab electronics) in the outside plant-just optical fiber to the splitter and then to the home. Because the optical splitter is passive, it has no power or electronics associated with it, greatly increasing the life and reliability of the loop plant and thus reducing the chances of a problem occurring to a subscriber's connection or service.
ATM PON also offers subscribers quicker service delivery and no delay in the implementation of new services. With ATM PON, new subscribers can be easily added to an existing PON facility by interconnection at an existing splitter or installing a second-tier splitter for increased drop capacity.
In new development areas, where the ground is already being excavated and laying fiber is relatively simple, the cost of laying fiber is economically justifiable. Here, access-network providers can deploy fiber as an overlay to copper, so subscribers can still receive traditional lifeline "plain old telephone service."
In existing serving areas where the loop infrastructure is already in place, the need for more bandwidth will eventually require rehabilitation of the existing copper loop plant. At that time, telephone companies can consider FTTH/B as a viable, higher-speed alternative to an FTTCab deployment, particularly in areas served by an aerial plant or where dark fiber or conduit already exists deep within the buried plant.
Advancements in ATM PON technology will allow increased split ratios, permit higher downstream capacity, and offer reserved and on-demand dynamic bandwidth allocation. These capabilities will enable end users to request additional bandwidth as needed.
Ray Hogg is principal product planner at Fujitsu Network Communications Inc.