The rapid growth in video and IP user services is one of the key trends affecting telecommunications. Cable multiple systems operators (MSOs) are aggressively deploying video on demand (VoD) systems while expanding support of high-definition TV, both broadcast and on-demand. IP services, once consisting only of residential high-speed Internet, are being expanded to encompass VoIP and IP and Ethernet services for business customers.
Meanwhile, traditional telcos and CLECs, in the face of declining opportunities for conventional telephone service, are rushing to deliver a full spectrum of video and IP services to residential customers as well as advanced IP services to businesses. Telcos recognize the need to compete fully with cable MSOs in standard broadcast video, VoD, and HDTV and are responding with fiber to the home and node triple play networks supporting video, IP, and telephony services. The incumbent telcos’ drive to supply IP services to business customers has been a key to their acquisition of traditional IXCs, and VoIP services for business and residential customers are on the horizon.
Video and IP services are similar in requiring very high data rates. Even a single stream of HDTV exceeds 15 Mbits/sec, while an IP service may involve the delivery of several Gigabit Ethernet drops to a single business customer. Furthermore, video and IP services require the delivery of subscriber data from an aggregation point that may be over 100 km away. Clearly, video and IP services require regional optical transport networks to deliver video streams and IP data to residential and business customers. However, high capacity is not enough. Video and IP services each have underlying characteristics that impose specific requirements on the optical transport network.
Video services are undergoing a transformation from standard-definition video streams of about 3 Mbits/sec to high-definition streams of 15 Mbits/sec or more. A further transformation is underway, from the simple broadcast model where each video stream served tens of thousands of subscribers to an increasingly on-demand model with unicast streams to individual subscribers. The result is a dramatic increase in the aggregate transport network bandwidth demand that only DWDM systems can meet while avoiding fiber exhaust. Cable MSOs have been replacing their old analog and digital fiber regional transport networks with DWDM for the past several years. As telcos begin to deploy video services more broadly, they will likely find it necessary to deploy DWDM regional transport networks as well.
Bandwidth scale is not the only requirement, however. Video services are somewhat unusual in that their traffic pattern is highly asymmetrical, with large video-stream traffic sent downstream toward the subscriber but only very limited control traffic sent back upstream. A DWDM transport network that can be configured for asymmetrical traffic can offer significant cost economies to the MSO or telco service provider (SP).
Finally, a key evolution in the deployment of VoD has significant implications for the transport network. SPs are recognizing that capital cost and operational expenses can be reduced with different combinations of decentralized and centralized video-server architectures. That requires the transport network to be easily reconfigurable to accommodate the continued deployment of the VoD service. Traditional DWDM systems employ fixed lasers and optical filters to drop wavelengths at desired locations. These systems can only add or reconfigure services via local technicians, involving costly truck rolls and service disruption. A transport network based on reconfigurable optical add/drop multiplexers (ROADMs), on the other hand, can be reconfigured by software from a remote location, without service disruption.
A further evolution provides video services with carrier grade availability. In this case, the SP decides to deploy redundant video-server locations, imposing a less predictable “any-to-any” traffic pattern on the transport network. Certain optical-network architectures provide the flexibility to allow ROADMs to accommodate any traffic pattern at minimum cost.
IP services may take the form of high-speed residential Internet access, VoIP service, or business IP or Ethernet service. Of course, IP and video services are increasingly coming together in the form of IPTV, soon to be the dominant architecture used by telcos deploying video services. Like video services, IP services have significant implications on optical transport network architecture and tend to impose high-bandwidth demands on the transport network.
Residential high-speed Internet access services typically supply 2-3 Mbits/sec of downstream traffic to a subscriber, compared to 35-50 kbits/sec of older dial-up Internet access, an increase of nearly two orders of magnitude. A VoD service using IPTV requires between 2 and 15 Mbits/sec for each stream, with a stream required for each active user. Most significant, IP and Ethernet business services typically deliver between 100 Mbits/sec and 1 Gbit/sec per drop-nearly three orders of magnitude higher than traditional DS-1 at 1.5 Mbits/sec.
In addition, IP-service traffic patterns are highly variable. The deployment of VoIP and IPTV requires easy reconfiguration as the networks evolve, but business IP services impose the clearest requirement for reconfiguration. In this case, an SP must respond quickly to requests from business customers for additional sites, a new wavelength between sites, or more bandwidth. A DWDM network based on ROADMs meets these requirements.
Finally, the highly distributed topology of IP-service networks with routers, gateways, and servers in many locations requires that the transport network accommodate any-to-any traffic patterns efficiently and economically. As in the case with video services, there is a specific type of ROADM-based architecture that meets this need.
There is an optical transport network architecture that provides the scalability, reconfigurability, and traffic-pattern flexibility to meet the requirements of video and IP services at minimum cost. It consists of two elements: ROADMs based on a broadcast-and-select architecture with tunable filters and lasers and nodes interconnected by a switched ring.
A ROADM based on a broadcast-and-select architecture includes simple two-port couplers and tunable filters (see Figure 1). It allows all traffic to be seen by all nodes, and therefore traffic can be dropped at any or all nodes, instantly reconfigurable by software control.
Compared with conventional ROADM designs based on wavelength-selective switches or wavelength blockers, the broadcast-and-select architecture offers significantly lower cost (typically $25,000-$30,000 less per node) and cost economies for asymmetric traffic such as video. Because of its broadcast nature, a broadcast-and-select network provides comprehensive optical performance monitoring from a single point.
Some broadcast-and-select ROADMs offer a very large number of wavelengths to mitigate the lack of wavelength reuse inherent in this type of architecture. Should exceptional traffic growth require wavelength reuse at some point, a wavelength blocker can always be added (see Figure 2).
Complete flexibility to accommodate any-to-any traffic patterns can be provided by interconnecting broadcast-and-select ROADMs into a ring network, in which the transmitted signals at each hub are combined and duplicated (via a 1×2 coupler) in both east- and west-bound directions with another 1×2 coupler at each receiver.
However, interconnection by a simple ring could cause recirculated amplified spontaneous emission noise to produce an inadvertent lasing phenomenon as well as cause potential multipath interference. A novel solution to this problem is the switched-ring architecture (SRA). An SRA has a distributed switching functionality. By moving protection switching into the overall control plane of the network, the SRA significantly simplifies optical-fiber protection for any-to-any traffic patterns on a ring network.
Any-to-any connectivity with plentiful bandwidth in a protected ring network is highly desirable for emerging diverse video and IP traffic. Cable MSOs, traditional telcos, and CLECs are all aggressively competing for subscribers by deploying advanced video and IP services. These services impose significant demands on optical transport networks in terms of bandwidth scalability, reconfigurability, and traffic-pattern flexibility, making the transport network critical to the SP’s competitive position.
A DWDM network, including ROADMs based on a broadcast-and-select architecture deployed in an SRA network, can meet carrier requirements for scalability, reconfigurability, and flexibility at minimum cost. Such a network enables the SP to capture maximum subscriber revenue from advanced video and IP services.
Karl May is president and CEO of OpVista (Irvine, CA, www.opvista.com).