Voice Over Frame Relay: And They Said It Couldn't Be
Done
BY WEI WEI LEE
Frame relay usage has increased dramatically over the past year with the availability
of public global frame relay services, and will continue to grow as new applications are
introduced. Frame relay will also become an access protocol to emerging ATM networks. But
frame relay offers far more than just migration to higher speeds. With the advent of voice
over frame relay technology, you can now significantly reduce the cost of domestic and
international circuits by migrating from expensive leased lines and moving voice traffic
onto frame relay circuits. With a worldwide frame relay data network in place, why not
take advantage of this available and cost-effective bandwidth for voice traffic as well?
FRAME RELAY VERSUS LEASED LINES
Leased lines are the traditional transport medium for voice traffic in today’s global
networks. As a guaranteed connection between two points, leasedlines are extremely
reliable. However, in order to connect multiple points, a leasedline has to be rented from
a provider between every node in the network. While these lines are reliable and provide
very little delay, building a meshed network for redundancy can be expensive
Network managers, trying to find ways to reduce expenses, are turning to shared public
data services such as frame relay. Using statistical multiplexing (the theory that not
everyone uses the circuit at the same time), service providers can interleave many users
on the same frame relay circuit. These users share the bandwidth and the cost of that
circuit. Unlike private leasedlines, frame relay users lose the exclusive right to
transmit only their data over the circuit, however, providers and network managers can and
do arrange Quality of Service (QoS) guarantees.
For years, data users have enjoyed frame relay’s ability to improve performance
(reduced response time) and reduce transmission costs dramatically for a number of
important types of network applications, such as Internet access and file transfers.
Ultimately, frame relay has been most effective for users when:
- The end devices are running an intelligent higher-layer protocol.
- The transmission facilities are virtually error-free such as in modern fiber
infrastructures.
- The application tolerates variable delay.
Unfortunately, voice traffic does not tolerate delay. So, how then, can voice over
frame relay be possible?
IMPLEMENTATION ISSUES
Voice over frame relay is an idea whose time has finally come. With the help of a select
few switch vendors, network managers are beginning to take full advantage of voice over
frame relay due to a number of technological advancements that have developed in the last
year. Three important hurdles that had to be overcome for voice to be a viable application
on a frame relay network include: voice quality (compression and silence suppression);
transmission quality (jitter control); and congestion control (prioritization and
segmentation).
VOICE QUALITY
Unlike data communications, voice communications must be performed in real time. This
means that delay must be kept small enough to remain imperceptible to the user. As
mentioned, bandwidth requirements and transmission delays have previously made realtime
frame voice transmission unattainable. Additionally, speech is burdened with a significant
amount of redundant information that is normal in the natural environment, but unnecessary
(and bandwidth consuming) in a communications network. Analysis of voice signals shows
that only 22 percent of any given sample consists of essential speech components that need
to be transmitted for complete voice clarity. The balance is made up of pauses, background
noise, and repetitive patterns. Today, advances in voice compression algorithms, combined
with state of-the art, high-speed digital signal processor (DSP) technology, make realtime
frame voice systems a practical reality. Technological advances include:
Digital Encoding
Simply put, with digital encoding, the network encodes the incoming analog
waveform at a 64-kbps rate. The network analyzes and processes the actual con tents of the
voice waveform by sampling at a rate of 8,000 samples per second using 8 bits to represent
each sample.
Removal Of Repetitive Speech Patterns
Repetitive patterns are inherent in human speech, and are caused by vibration of the vocal
cords. Nature put these repetitive patterns in our speech so that it can survive passage
through the environment, like a noisy room or a windy day. While traveling through the
environment, perhaps only half of the waveforms will reach the listener’s ear. Yet in
a communications network, all transmitted waveforms are received.
Transmission of these identical patterns represents wasted bandwidth. Recent advances
in compression algorithms, such as Advanced Codebook Excited Linear Prediction (A-CELP)
help to produce very-high-quality voice results.
Voice Frame Formation
Voice frames are formed from the remaining speech data. This process inherently removes
silent periods and pauses, as frames are formed and transmitted only during actual speech.
The frames are variable in length and consist of a 10-byte header and address field.
Some implementations of voice frame technology use much larger frames and complex
header fields. This increases overhead and can result in long transmissions, which are
more susceptible to network delays, which are unacceptable in voice applications.
Therefore, it is important to use variable length frames with short headers — this
minimizes the transmission delay for compressed voice traffic.
Transmission
Human communication is essentially half-duplex by nature. When one person speaks
(transmits), the other listens (receives). Over 50 percent of all con versation consists
of silent periods. With frame voice technology, switches today can exploit these silent
periods by transmitting information (frames) only during actual speech. The remaining,
unused bandwidth can be used for data traffic or other voice calls.
TRANSMISSION QUALITY
The overwhelming cause of poor quality over voice circuits is jitter. Jitter can occur in
a public network when an intermediate switch is busy with another packet. In this case,
the second packet is held at the switch until transmission of the first packet is
complete. Since frame relay packets can vary in length, the amount of delay is variable
and unpredictable. If the jitter exceeds the ability of the receiving device to compensate
through buffering, voice quality will suffer or the signal will be completely obliterated.
To deal with this problem, elastic buffering processes manage a buffer of up to 125
milliseconds on a realtime basis. Continuity of a telephone conversation is ensured, since
jitter can be smoothed. Voice is much less delay-tolerant than data, and can not handle
more than one-half second of delay.
CONGESTION CONTROL
Network managers need to be sure that one application does not overwhelm another. But they
must also provide consistency of performance for all network users. The most effective
method to accomplish this is to provide priorities to regulate access to the network
bandwidth and segmentation to ensure consistency of performance.
Prioritization
The network manager needs to be able to set up priorities not only for each voice/fax
channel but for data applications as well. The highest default priority should be
automatically assigned to any channel configured to support either fax or voice traffic.
The network administrator should also assign individual priority codes to selected data
channels, as well as establish a prioritization scheme within Data Link Congestion
Identifiers (DLCIs). This type of prioritization scheme allows data traffic to be buffered
until all higher priority fax and voice packets have been presented to the network.
Further, a priority management process should allow the switch to handle other
delaysensitive devices such as SNA (Systems Network Architecture) terminals.
Switch vendors are providing switched frame capabilities that incorporate multiprotocol
encapsulation schemes (such as the standard RFC 1490). RFC 1490 supports the encapsulation
of many different protocols. This process provides the optimization of the frame relay
access link by packing the outbound traffic over a single DLCI.
Delay-sensitive fax and voice frames must also be protected against being discarded. A
facility that allows the user to reconfigure the Discard Eligibility (DE) status of
individual channels is necessary. Channels with the DE indicator set indicate that the
information that passes through the marked channels are Discard Eligible. Through this
process, channels can be managed so that critical information can be presented to the
network with reasonable certainty that the presented packets will not be discarded.
Segmentation
Segmentation is used to break up larger blocks of data into smaller, less delay-creating
frames when delay sensitive traffic is present. This ensures an even flow of voice frames
into the network, minimizing delay. For example, a 4000-byte frame can take up to half a
second to transmit over a 64-kbps line. That is too much delay to ensure consistent voice
quality.
The segmentation must involve all of the data in the network to retain consistent voice
quality. Delay will still occur if a voice frame is held up in the middle of the network
behind a 4000byte frame. Today’s advanced voice over frame relay switches will
automatically segment and switch frames to ensure consistent quality of service throughout
the network.
NETWORK MANAGEMENT
Frame relay, whether it’s carrying voice or data, needs to be managed. Why? Unlike
it’s predecessor X.25, frame relay has a very small header with limited inherent
management capability. It assumes that the network is clean. This is one of the things
that makes it so efficient. However, it also means that when problems do occur, network
managers need a way to diagnose problems. Additionally, many carriers are not prepared for
voice over frame relay traffic on their public networks. Subsequently, few offer service
level guarantees. Combined, these issues relate why it’s important to choose a voice
over frame relay switch vendor that has strong network management offering.
Distributed management capabilities are the current favorite when determining a network
management solution. Distributed network management systems monitor and control the
network switches that are carrying the voice over frame relay. With extensive diagnostic
tools, such as datascopes and pattern-generators, these systems provide the capability to
remotely diagnose network anomalies without having to have a technician on-site. In
addition, distributed network management systems decrease problem resolution time.
CONCLUSION
Frame relay was not originally designed to carry voice traffic. There are some inherent
qualities of a frame relay network that make it seem intolerable of voice traffic. But
these have been and can be easily overcome in light of the tremendous economic benefit of
voice over frame relay. It is important for network managers to rely on switch
vendors’ extensive experience with voice and data switching. These vendors have
implemented sophisticated routing and segmentation techniques in their software systems to
support a complete high-quality voice over frame relay network.
With the proliferation of cost-effective frame relay services on a global scale,
migrating voice applications to the frame relay network represents a large opportunity for
cost savings. But this is possible only if voice is handled with the same attention to
features, functions, and network management that has been paid to data applications.
This migration will also provide an effective and intelligent switching layer for use
with ATM networks, enabling the network manager to take further advantage of efficiencies
of voice and data integration.
Wei Wei Lee is senior network analyst at NETRIX Corporation, which manufactures and
markets widearea network switching products for enterprise networks. NETRIX products unify
voice, data, and image applications over the diverse and ever changing array of
communication services. Combining patented, switched, compressed voice technology and
advanced networking capabilities, NETRIX delivers networking solutions that improve
network performance and achieve substantial operational savings. For more information,
visit the company’s Web site at www.netrix.com.
|
Internet Telephony Magazine
Click here to read latest issue
CUSTOMER 
Cloud Computing Magazine
Click here to read latest issue
IoT EVOLUTION MAGAZINE
