PDF of this tutorialToday’s time-division multiplex (TDM) PSTN was designed to accommodate a single application: uncompressed voice. The network was built to deliver 99.9994 percent up time, low-latency and jitter, highly scalable call routing through the SS7 infrastructure, and value-added voice services such as voice messaging and caller ID.
The PSTN provides voice access services in 64–kbps increments known as DS–0s. The DS–0s are aggregated by digital access cross-connect systems (DACSs) for presentation via DS–1 (1.5–Mbps) or DS–3 (45–Mbps) interfaces to Class 5 switches. The Class 5 switches communicate with access and toll tandem switches to route calls across the telephone network to destination Class 5 switches. The DS–0 voice calls are then split back out by DACSs to their original 64–kbps state.
The multiplexing and demultiplexing of voice traffic works well in a fairly static, voice-oriented environment. Provisioning of the DACS, toll and access tandem switches, and trunks is cumbersome but acceptable, as a result of fairly consistent voice traffic patterns.
During the late 1980s and early 1990s, however, the rapid growth of distributed processing and the Internet changed everything. Dial Internet access devoured PSTN circuits with unexpected call duration times and resulted in diminished quality of service (QoS), as voice users experienced “fast busies.” The avalanche of Internet users forced carriers to invest in additional PSTN capacity, a costly exercise—given that large circuit switches cost between $1 and $3 million and DACSs between $100,000 and $250,000 for traffic that was providing little if any incremental revenue.
In addition, the bursty characteristics and bandwidth demands (DS–1 to DS–3) of data transmission were not well suited to the circuit-switched nature of the PSTN and required few of its value-added voice features. Packet-switched networks could accommodate the rapid growth and handle the bursty characteristics of data applications more efficiently than the PSTN’s fixed 64–kbps architecture. Carriers quickly concentrated their efforts on data-centric architectures and constructed or expanded frame relay, ATM, and IP networks, most of which rely on ATM as the core technology, in the underlying backbone of the network. The specialization of voice and data or packet networks has resulted in carriers supporting a large number of overlay networks (see Figure 1).
Figure 1. PSTN Architecture
British Telecommunications PLC, for example, has no fewer than 17 overlay networks, each of which requires separate network management and back-office systems and support, and each of which has varied technical standards and protocols.
The explosive growth of data traffic also forced carriers to expand their data network infrastructures constantly. The growth has been so rapid that companies such as MCIWorldCom experience annual data backbone growth rates of roughly 800 percent, while voice traffic, by comparison, grows at only 4 percent worldwide. However, it is critical to note that voice service revenues continue to provide service providers with 80 percent of their revenues. Accordingly, service providers, although building infrastructures for future applications, must continue to support high-quality voice services that drive their revenues.
