May 1, 2004

Reapplying the PSTN to Data Services

By Justin J. Junkus

Last month's column introduced the concept of the three-layer network hierarchy that is used to complete VoIP calls over an internet. This month, we will look further into how parts of the public switched telephone network (PSTN) are being reapplied to the internets used for data services and VoIP.

One of the housekeeping details we need to get out of the way before going further deals with naming conventions. When we discuss internets, if the "i" in internet is lower case, we are referring to any set of connected networks, including those run by a cable operator, ISP or an enterprise. If the "I" is capitalized, we are referring to the public Internet-the same network that is the home of the World Wide Web.

The model for a data network, as you may recall, consists of access, distribution and core segments. The core is the backbone network that interconnects the subnetworks that make up an internet. By way of analogy, think of the core as a set of airline hub cities. Just as a major storm in Chicago can delay flights out of Denver, delays in a core node will affect many other parts of the internet. Consequently, the core needs to move data very fast. Layer 3 routing, packet assembly and disassembly, and almost any operation except switching contribute to delays and therefore are avoided as core functions.

Core technology, however, is not uniform. Data has been moving across networks for almost 25 years now, and there are lots of legacy protocols that must be accommodated. Examples are frame relay, asynchronous transfer mode (ATM) and synchronous optical networking (SONET). The core has evolved with data communications advances, but, like everything else in the communications industry, there has been no universal, "forklift" upgrade. What has occurred, however, is a set of extensions to legacy protocols that allow them to intercommunicate between themselves and new, improved protocols.

SONET evolution

SONET is a good example of ongoing protocol evolution and the reapplication of PSTN resources to the Internet. SONET was introduced in the mid-1980's by the telephone industry as a standard for optical transport between telephone switches in the PSTN. There were several driving forces behind SONET, including the need to mix and match vendors within a network, and to find and extract low rate data from a high-speed stream without having to go through several stages of multiplexing. Operations and maintenance were important considerations, and SONET was built to provide a way to provide efficiently both fault detection and monitoring for end-to-end as well as intermediate spans.

SONET is a product of legacy time division multiplexing (TDM) technology. Consequently, it is built upon a set of uniform "frames" that are repeated at regular intervals. To picture a SONET frame, think of a spreadsheet with nine rows and 90 columns. Within each frame, there is a specified location for a data byte (eight ones or zeros) that is associated with a given set of data or with operations and maintenance. In our spreadsheet, these locations would be the cells. If the SONET frame were a spreadsheet, it would allow the data within the cells to be changed at a rate of 8,000 times per second, and at each refresh cycle, the data within the cells could change. In one second, our spreadsheet cells would have processed 51,840,000 bits of data, which is the basic SONET frame rate called OC-1. With further multiplexing, SONET carries data at multiples of OC-1, up to 768 times faster than this rate.

Today, there is a lot of SONET in core networks, at least in part because telephone companies were the first to use interswitch transport to handle backbone data routing as well as interswitch voice traffic. With packet technology rapidly displacing TDM, however, the current challenge is how to continue using existing SONET transport rather than require a forklift network upgrade. The solution is called "next-generation SONET."

Next-generation SONET

Even before the Internet protocol (IP), SONET had been adapted to work with other protocols. The basics of how this is done go back to SONET's original design. By the time SONET was implemented, the earlier North American Digital Hierarchy (NADH) already was being used to multiplex data as well as voice, both in digital formats, but not yet as implementations of IP. The NADH accommodated rates from DS-0 (64 kbps) to DS-3 (44.736 Mbps), using a multiplexing scheme entirely different from SONET. When the telephone engineers built SONET, they had to figure out a way to work with the NADH. The compatibility solution was to partition the SONET columns into subsets called virtual tributaries, which provide an exact location in a SONET frame for data that is in a DS-1 through DS-4 multiplex. (DS-0 is handled by default-once you know the location of a DS-1, you can find the associated DS-0s.)

Next-generation SONET involves mapping bytes to SONET frames from protocol data units originating as IP, gigabit Ethernet, DVB-ASI, fiber channel and other data technologies, using new protocols, such as generic framing procedure (GFP), virtual concatenation (VCAT) and generalized multiprotocol label switching (GMPLS). Each mapping has a different methodology, but the underlying concept is very similar to the way the NADH is carried in SONET: reorganize data into manageable blocks, assign those blocks to parts of one or more SONET frames and provide a pointer in the SONET overhead to locate packet data when present. With time-sensitive data, such as VoIP, a method of maintaining quality of service must be built into the protocol that does the repackaging and assignment.

Provisioning platforms

These new protocols meet next-generation SONET in network equipment called multiservice provisioning platforms (MSPPs). An MSPP is a single network element that interfaces with several protocols and typically provides SONET add-drop multiplexing along with digital cross-connect functions.

There is much more to next-generation SONET and the associated equipment than can be discussed in one column. What we have covered here, however, is a high-level view of how TDM-based technology from the PSTN is being adapted to the packet-based technology in core networks to accommodate most new broadband applications. In future columns, we will discuss how vendor offerings using this technology are being used today by cable operators.

Justin J. Junkus is president of KnowledgeLink, Inc., and Communications Technology's telephony editor. To discuss this topic further, email him at jjunkus@knowledgelinkinc.com.