Rough start, but voice market growing

The recent carnage in the next-generation voice space among established North American players as well as startups is yet another reminder that hype alone won't do the job. Nevertheless, shrugging off setbacks and delays in deployment, everyone seems to be optimistic about the future of voice-over-Internet Protocol.

Analysts predict that the VoIP market will be strong and recent announcements from networking companies about deployment of VoIP systems show promise for international expansion. But equipment vendors have expressed confidence in the packetized voice market even as they announce lower earnings.

The business case for VoIP is based upon the key economic benefits of IP networks. IP offers a connectionless network where resources can be distributed and shared. Treating voice as just another data type, the real value proposition is not about voice-over-IP (transport) but about voice-and-IP (enhanced services).

At first, VoIP was touted as a cheaper way to make a phone call. Then proponents promised lower operating, infrastructure and management costs as well as the positioning of revenue-generating enhanced services. But those promises have been largely unfulfilled. Despite the technical, economic and business advantages, as well as numerous compelling benefits, VoIP has not been adopted as widely as was predicted.

The most significant barrier has been the existence of billions of dollars worth of the high-quality, high-reliability legacy public telephone system. The public switched telephone network offers "toll-quality" voice at unparalleled reliability vs. the voice distortion, echoes and dropped calls on VoIP. Also, deployment of the promised revenue-generating enhanced services has been difficult to achieve in the absence of a high-quality and high-reliability packet infrastructure.

Carriers and service providers are sensitive to the cost of emplacement and provisioning, and are wrestling with the question of what to do about the investments they have already made in their existing communications networks. They would like those investments to be fully depreciated before they consider widespread financing of new platforms, so carriers would like to deploy next-generation networks while continuing to use as much of the legacy infrastructure as is practical.

Quality concerns

One of the major concerns about VoIP is voice quality, which is based on users' experiences and expectations. While users will tolerate reduced quality for the convenience of mobile communications, they are used to and expect toll-quality voice in all other situations. Voice is a delay-sensitive application, and to have good quality it must stay within the delay budget; a 150-microsecond one-way delay is considered acceptable by the International Telecommunications Union while delays over 250 microseconds are noticeable and unacceptable. But keeping voice within the budget on an IP network is difficult because IP networks by nature are characterized by bursty traffic and best-effort delivery. Packet loss, coding techniques and switching delays also can impair voice quality. Therefore, managing the latency and prioritizing traffic in a VoIP infrastructure are the keys to delivering toll-quality voice.

Latency is defined as the average travel time it takes for a packet to pass through the network, from source to destination. That figure varies according to the amount of traffic being transmitted and the bandwidth available at that given moment. If the traffic is greater than the bandwidth available, packets will be delayed. Latency is introduced by two primary sources: gateways and the IP network connecting the gateways. Gateways contribute to codec processing latency, packetization/depacketization latency, framing latency on the transmit side and jitter buffer latency on the receive side. Network latency includes media-access latency, routing latency and firewall-proxy server latency.

The delay problem is compounded by the need to remove jitter, which is variable interpacket timing caused by the network a packet traverses. If the network is congested, arrival times for the packets will vary. Because the voice playback speed must be constant, a jitter buffer is used to remove the variation, or jitter, in the flow of packets to the decoder. This causes additional delay.

The delay imposed by the jitter buffer depends on the variation in delay across the network. A very short jitter buffer is sufficient where congestion control is used, whereas uncontrolled jitter will cause packet loss.

Compression delays

Today, most VoIP gateways use voice compression to maximize network bandwidth use. However, each time voice is compressed (or decompressed at the receiving end) delay is introduced, affecting the overall end-to-end delay. Because the compression codecs add significant delay, the delay budget defining the distribution of allowable delay to the various network elements may require adjustments to accommodate a long encoding delay. When packet loss is introduced, these codecs will show different amounts of degradation, depending on the effectiveness of the associated packet loss concealment algorithm. Given today's increased bandwidth economies, compression is not necessary.

See related chart

All migration plans will continue to move at a snail's pace until VoIP is able to deliver toll-quality voice reliably. To ensure the voice quality, sufficient bandwidth will have to be allocated using Resource Reservation Setup Protocol and packets will have to be prioritized so that voice traffic is transmitted ahead of other types of information that are less time-bound. End-to-end quality of service in the VoIP network is key to making VoIP a reality. It is critical to employ a network-wide intelligent QoS solution for delivering toll-quality voice reliably.