Networks Converge

Currently, voice and data travel on separate networks. Convergence will come when a single network has the ability to reliably transport voice, video, and data. Once a single network transports all forms of communications, communications service providers can sell you telephone service, video conferencing, radio and TV channels, Internet access, virtual private networks (VPNs), and firewalls. (See “Business Services”)

What’s in a converged network? There’s no single answer, but all networks require switches, access devices, and transport protocols. With that in mind, I’ll show you a generic model of a network and point out different ways that telecom service providers build these networks. By “build,” I’m referring not just to switches and cables, but to the signaling and protocol layers that organize bits into meaningful transmissions. Next month, I’ll discuss methods for testing these packet-switched networks and their components. The tests range from voltage measurements on individual bits, through bit error rate (BER) tests and protocol analysis, quality of service (QoS) measurements, and performance measurements.

Access devices

The overall network consists of several physical parts. The access network is the on-ramp that connects customers to a central office (CO). In Figure 1, digital subscriber line (DSL) technology connects a customer’s premises—small office, apartment building, or hotel—to a telecom carrier’s CO.

According to Rob Brown, network architect at Sedona Networks (Kanata, ON, Canada), a business may use an integrated access device (IAD) to get access to a communications network. An IAD may contain a few dozen plain-old telephone service (POTS) ports and one or two T1 ports that connect a private branch exchange (PBX) to a packet network. The IAD also may have Ethernet ports so users can connect computers for Internet access.

On the telecom network side, an IAD may support one or two DSL lines. It also may have a T1 port for customers who lease a T1 line from a telecom provider.

A DSL modem (or IAD with a built-in DSL modem) communicates with a DSL access multiplexer (DSLAM) located in a CO of a local exchange carrier, the company that provides local telephone service. DSLAMs, which complete the access network, demodulate upstream DSL signals sent from a subscriber’s premises, and they modulate downstream signals before sending them to subscribers. DSLAMs contain several line cards, and an exchange carrier needs different cards for different DSL services such as asynchronous digital subscriber line (ADSL) and symmetric digital subscriber line (SDSL).

The IAD also segments upstream data into asynchronous transfer mode (ATM) cells before modulating the bits onto a DSL carrier. DSLAMs then multiplex, or concentrate, disparate ATM streams into a larger stream of cells. Conversely, DSLAMs demultiplex downstream transmissions and place them on the proper local loops. A DSLAM may produce Frame Relay frames or Internet Protocol (IP) packets rather than ATM cells, depending on the transport network. The network in Figure 1 uses ATM as its core. Table 1 lists a network’s electrical and optical line speeds; some of these interfaces appear in Figure 1.

Figure 1. The converged network carries voice, video, and data through access devices and aggregates traffic through switches for transport over a core network.

On the customer side, multiservice edge switches may connect to corporate networks that use ATM, Frame Relay, IP, or ISDN as well as connecting to DSLAMs. On the network side, the multiservice edge switch sends and receives ATM cells to and from the core transport network. These multiservice switches form what is sometimes called an aggregation network.

Another set of switches may reside between the aggregation network and the core network. These switches, which physically reside in a CO, form an “IP service network” that adds value to the customer’s service. For example, a service provider may use an IP service switch to host a small company’s firewall so the customer doesn’t have to invest in a firewall’s hardware or maintenance.

A communications service provider may use IP service switches to provide a VPN so customers can get access to their Intranet or corporate network from outside the company. In addition, the provider may give users access to an Internet service provider (ISP) or application service provider (ASP).

Not all telecom networks have separate IP service switches. Some switches combine two or more functions into one chassis. For example, some multiservice switches can add IP services.

Switch manufacturers also combine other functions. A switch may combine an ATM switch with a telecom switch or with a voice gateway. Such a product connects DSLAMs to the public switched telephone network (PSTN). The integrated switches let businesses that already use IP phones make calls over the PSTN, the network that currently carries most phone conversations.

Packets and protocols

DSLAMs, multiservice edge switches, IP service switches, and core-network ATM switches can pass billions of bits each second. An exchange carrier must manage all those bits to ensure data and voice signals arrive at the proper destinations. In the case of audio and video transmissions, bits must also arrive at the right time. A seemingly endless set of protocols brings order to the bits.

Protocols organize data into groups called frames, cells, and packets. Each contains a header and a payload, and sometimes a trailer. The header holds a destination address and may include the length of the frame or packet and a checksum for error correction. Some headers also contain time-stamp information, which tells the receiver how long the data took to travel over the network. That’s important in voice and video transmissions where the timing affects the quality of the delivery.

The protocols are organized around the familiar Opens Systems Interconnection (OSI) seven-layer protocol stack, although not every communications technology requires protocols at every layer. Data protocols such as hypertext transfer protocol (HTTP), file transfer protocol (FTP), and simple mail transfer protocol (SMTP) reside at layer 7, the applications layer. These protocols handle the processing of items such as Web pages, data files, and e-mail, respectively.

When you download a Web page, for example, a Web server sends HTTP packets that get encapsulated into lower layer packets for transport. Specifically, the Web page’s code gets encapsulated into transmission control protocol (TCP) packets at the transport layer, which are in turn encapsulated into IP packets at the network layer.

When the packets reach the core network, they may have gone through several encapsulations. IADs, DSLAMs, and ATM switches perform a process called segmentation and reassembly. Segmentation breaks the packet information (which may have a variable length) into fixed-length cells. Reassembly takes the received cells and builds the packets from the cells’ payloads.

An ATM cell’s header informs switches of the cell’s destination and QoS level. QoS defines the end-to-end delivery limits of cell loss ratio, cell transfer delay, and cell delay variation to the call. ATM network components negotiate with each other to assign higher QoS levels to real-time audio and video than they assign to cells carrying data files, because delays and variations in cell delivery more profoundly affect audio and video than they affect file transfers. The cell’s payload carries headers and payloads from higher-layer protocols. The more encapsulation needed to transport bits, the less efficient the medium.

ATM protocols have several sublayers that set rules for how the protocol creates its cells. According to Oded Agam, vice president of testing solutions at RadCom (Mahwah, NJ), ATM adaptation layer 2 (AAL2) supports the transport of audio signals that have undergone compression or silence suppression, while ATM adaptation layer 5 (AAL5) frames carry data such as Web pages and files.

Figure 2. Codecs compress digitized audio and video where transport-layer protocols manage the payloads and set up calls.

Audio and video require their own sets of higher-layer protocols. Most voice over Internet Protocol (VoIP) systems use a protocol stack called H.323, which resides at the applications layer and contains audio and video compression and coding algorithms, called “codecs.” Figure 2 shows the VoIP protocol stack. A codec will compress audio data into packets, which then get encapsulated at the transport layer.

Real-time transport protocol (RTP) packets contain encoded audio bits, which an IP phone at the other end of the network uses to decode into analog audio signals. The packet headers also include time-stamp information, which allows the receiver to compensate for latencies and jitter in the arrival of the voice-encoded packets.

User datagram protocol (UDP) packets, which encapsulate RTP and real-time transport control protocol (RTCP) packets, reside at layer 4, the transport layer. UDP adds multiplexing and checksum services to the transmissions from layer 7 so a network can pass more than one audio conversation at a time. (Ref. 1).

According to Jeff VanZwol, director of product marketing manager at Oresis Communications (Beaverton, OR), IP packets will eventually run directly on top of the physical layer, such as synchronous optical network (SONET) or dense wavelength division multiplexing (DWDM). Right now, though, many networks use ATM to guarantee the QoS that voice customers require. RadCom’s Agam also points out that voice over ATM using AAL2 is considered a competing technology to the VoIP.

IP lacks the QoS that audio and video transmissions require, so running IP packets encapsulated into ATM frames adds the QoS to the transmission. A new protocol, multi-protocol label switching (MPLS), adds the QoS feature to IP packets that let IP reliably send packets containing audio and video without the need for ATM. (Ref. 2). In addition to adding QoS to IP packets, MPLS adds a “time-to-live” field because if packets containing audio and video take too long to reach their destination, they have no value. The time-to-live feature effectively deletes the packet after the time expires.

Next month, I’ll discuss some of the measurements you need to take in a telecom network. I’ll discuss physical measurements on bit shapes, BER tests, and internetworking tests, including how to configure test setups. The article will discuss how to test IADs and multiservice edge switches. T&MW

References

1. Schulzrinne, H., et al., “RTP: A Transport Protocol for Real-Time Applications,” January 1996. www.cis.ohio-state.edu/htbin/rfc/rfc1889.html.

2. A World of Protocols, RadCom, Mahwah, NJ.August 2000. p. 407. www.protocols.com.

For more information

Agilent Technologies white papers and application notes about network technologies are available at: onenetworks.comms.agilent.com/
WhitePapers.asp.

The ATM Forum Web site contains ATM specifications. www.atmforum.com .

The International Engineering Consortium Web site contains dozens of tutorials on telecom technology. www.iec.org/tutorials.

Poster: “Convergent Networks,” Agilent Technologies. onenetworks.comms.agilent.com/forms/
ConNetworks_poster.asp. 

Poster: “Network Communication Protocols,” Agilent Technologies. onenetworks.comms.agilent.com/ forms/IntAdv_poster.asp. Editor's Note 10/24/03: This page has moved: http://xpi.comms.agilent.com/forms/IntAdv_Poster.asp.

RadCom’s www.protocols.com lists dozens of protocols and their structures. The company also offers a poster, “World of Protocols 1999–2000.”

Rowe, Martin, “Measure VoIP Networks for Jitter and Loss,” Test & Measurement World, December 1999. p. 29.   

Stallings, William, Data & Computer Communications, 6th ed., Prentice Hall, Englewood Cliffs, NJ, 1999.

Telecom Glossary 2000, www.its.bldrdoc.gov/projects/telecomglossary2000.

Martin Rowe has a BSEE from Worcester Polytechnic Institute and an MBA from Bentley College. Before joining T&MW in 1992, he worked for 12 years as a design engineer for manufacturers of semiconductor process equipment and as an applications engineer for manufacturers of measurement and control equipment. E-mail: m.rowe@tmworld.com.