There are significant advantages in using the Internet, known as Voice Over Internet Protocol (VOIP), in place of conventional phone switches. The Montreal Selg Network Support Company is a certified reseller and service provider for INTER-TEL Mitel and CISCO IP phone systems. By using SELG for your IP telephony needs one call can support all your IT needs.

Benefits of Voice Over IP Phones:

    Directs Voice traffic over current High Speed Internet pipeline
        Lowers traditional phone line costs
        Improves telecommuting opportunities for your staff
        Improves communications with remote locations or branch operations
    Consolidates all forms of communication to one In-Box
    Allows for maximum utilization of Microsoft Office Products
    Reduces the number of service providers necessary
    100% Scaleable
    More technology advantages at lower costs than traditional phone systems
        Virtual extenssions dialing
        Conferencing, voice and video
        Security monitoring

SELG has created evolutionary solutions for data networks in anticipation of the dynamic trends and business advantages of integrating voice onto data networks. SELG’s strategy is to provide high quality equipment and services with cost saving implementations that enable networks either to be upgraded using the installed base or architected to start simple and grow fast in cost-effective steps while accommodating new technologies or standards.

 Background Up to this point in time, two distinct communications networks have existed, one for voice and the other for data. The Public Switched Telephone Network (PSTN) evolved from an analog base so that today information is transported on both digital and analog signals. On data networks, information is transported on all digital signals.

The PSTN was built to handle voice traffic which can be characterized as continuous traffic of short duration. Therefore, a dedicated physical connection is established between the origination and the destination until the end of the call, i.e., a signal is sent to the network indicating an end to the conversation or transmission of information is complete. This dedicated end-to-end communication path, which uses a lot of equipment resources, cannot be reused for another purpose until the end of the call. Despite the fact conversations have many pauses with periods of silence, the equipment is still tied up. This results in less efficient use of network facilities, which adds expense. Furthermore, the analog nature of the PSTN requires conversion equipment for digital data to be transported — thus adding more cost to the equation.

In a data network, the terminal produces output in digital format so there is no transformation required for transmission. In contrast with voice information, data transmission tends to be bursty rather than continuous. Data protocols only send when data is present. This allows signals to be statistically multiplexed over the network facilities making more efficient use of the resources.


 The Appeal for Voice and Data integration over the Data Network   Today, market, technology and regulatory factors are converging to make the integration of voice over data networks economically attractive. The emergence of voice compression technology and standards for providing voice over Frame Relay, ATM and IP networks make it possible to provide quality voice services over a data network. A key market inducement for service integration is the need for businesses to reduce costs in order to compete in today’s global marketplace. Finally, the emergence of new carriers combined with the regulated cost structure of the global telecommunications network makes economic bypass possible.

SELG’s Networking Hardware Division identified these trends early and has incorporated voice technologies into its key data communications products so that businesses of all sizes can take advantage of the potential cost savings with a high level of quality and reliability. Since many SELG customers have existing data networks between remote sites and branches, our strategy is to make it easy for our data customers to migrate to an integrated network from the installed base, i.e., by adding voice capability to existing family of communications products. Furthermore, because our family of routers share common code, SELG’s solution assures interoperability amongst all of our products, thus protecting our customers’ investment. This same strategy applies to first time buyers since they can feel secure that SELG will protect their investment the same way.
 Market Trend Emergence of voice compression technology and standards for providing voice over Frame Relay, ATM and IP networks

Need for businesses to reduce costs in order to compete in today’s global marketplace

Emergence of new carriers combined with the regulated cost structure of the global telecommunications network makes economic bypass possible

 Benefits Cost savings in network management and service
Long-distance cost savings
Reduced equipment investment
Bandwidth efficiency


 Requirements Easy, cost effective upgrades

Network flexibility to accommodate both voice and data

Compression of voice conversation bandwidth must be reduced while maintaining high quality

Silence suppression

Quality of Service (QoS) to assure priority for voice transmission

Signaling for voice traffic over traditional PBXs

Echo control

Voice switching to decide whether to route a call over the internal data network or PSTN

 Economic Benefits   The main benefit for integrating voice over data is cost savings in several areas:

Long-distance cost savings
By integrating voice and data over an IP or FR enterprise network, a company can reduce or eliminate long-distance charges for intracompany calls.

Reduced equipment investment and service costs
Companies usually lease or purchase separate equipment and facilities for voice and data. With an integrated network, the cost of purchasing and maintaining a single network.

Bandwidth efficiency
With new protocols, data and voice can be efficiently merged onto the same network thus saving on the amount of facilities required.

 SELG’s Advantage for Voice and Data Integration SELG Networking is committed to delivering a scaleable family of networking products with the best price performance in the industry. Choosing SELG for voice and data integration provides benefits that can only come from a company that provides total end-to-end solutions: Common code— all routers interoperate

Easy upgrade to integrate voice on installed data products

SELG Global Support and maintenance services

Voice enabled routers for FR and IP networks that use key compression, QoS, voice and traffic management protocols.

Integrated Voice and Data features where the network can be efficiently managed to be used for data transmission activities as well as typical telephony activities.

Open standards for interoperability with equipment to allow cost savings as a network is upgraded or deployed.

Portfolio of telecommunications interfaces in order to connect to the PSTN and PBXs.

Solutions for Enterprise, ISP CPE and VPNs.

Support ITU/IETF standards for voice.

Ability to migration from FR to IP (both on same box)


 The Technology for Voice and Data Integration There has been a proliferation of the new codecs (compression and decompression algorithms) that provide nearly the same quality to voice communication on a data network as experienced with the current telephone network with substantially less bandwidth requirements. The codecs also implement silence suppression and voice activity detection algorithms so that the network only uses the bandwidth when conversation is going on. These approaches now make the voice traffic behave like data traffic in the network.

By turning voice information into packets and sending these short streams of information in bursts, network facilities are not tied up for long periods of time. Thus, network facilities are freed up for use quickly and allows more information to be sent. By “packetizing voice,” networks that are usually used for just sending data like e-mail or file transfers can now include voice conversations. Now, the data network is being efficiently used by collapsing two separate networks into one.

Figure 1. Business with an Integrated Voice/Data Frame Relay Network connecting Branch Offices

 Voice over Frame Relay (VoFR) For the company using a Frame Relay (FR) network and branch office connections, the branch office is geared with a router, like the SELG Access Utility (2212), that is enabled to do voice (See figure 1). With upgraded hardware and software, the branch office now has the flexibility to connect not only servers and workstations BUT ALSO telephones to take advantage of the “toll by-pass” savings with using the FR network to route calls over. An integrated data and voice network is metamorphosed from the existing components of the installed base - no “fork lift” is needed - to add voice capabilities to the network.

With the addition of a Voice enabled Frame Relay Access Device (VFRAD) to the network, possibly in the data center, a company expands the number of concurrent voice/fax calls it can make with T1/E1 connectivity to a Private Branch Exchange (PBX). The VFRAD also has the advantage of centralizing the dial plan information of the network since it provides call rout-ing functions for branch-to-branch calls. The centralized dial plan information minimizes the amount of unique dialing information that must ordinarily be configured in the branch office access routers.


Figure 2. Business with an Integrated Voice/Data IP Network connecting Branch Offices

A branch office must first have connectivity to the traditional telephone network. This is usually an analog (i.e. non-T1/E1) connection because the number of concurrent voice con-nections is typically low—two to four voice/fax conversations at any given time. The standard connection types in this environment are FXS, which allows handset/fax machines to plug directly into the router; FXO & E&M, and ISDN BRI which allow connectivity to public branch exchanges (PBX).

Voice compression must also meet certain requirements for a typical branch office. When voice is digitally encoded in the traditional public switched telephone network (PSTN), Pulse Code Modulation (PCM) is used. This results in 64-kbps data stream per voice/fax call. Since this is greater than the typical Committed Information Rate (CIR) used in many branch office locations, some form of compression is required. Typical compression algorithms based on Code Excited Liner Prediction (CELP) are used. The Frame Relay forum has recommended that G.729 be used a the standard voice compression algorithm for VoFR. G.729 is a CS-CELP Conjugate-Structure Algebraic-Code-Excited Linear-Prediction algorithm that results in a 8- kbs data stream per voice call. Other technologies, such as Silence Suppression and Voice Activity Detection (VAD) can reduce the average bandwidth required during a conversation by removing unnecessary information, such as silence, from being transmitted during a call. These functions—Digital Encoding, Voice Compression, Silence Suppression, etc.— are very CPU intensive and are performed on a Digital Signal Processor (DSP) separate from the router’s CPU. The PSTN interfaces and DSP for compression are packaged together as an adapter that plugs into a feature slot on the access router.

Once voice has been digitally encoded and compressed, it must be put in the appropriate envelope for transmission. In the case of VoFR, this envelope is a Frame Relay packet. How-ever, a standard Frame Relay header would only be capable of carrying a single voice call per Data Link Connection Identifier (DLCI). To address this limitation, the Frame Relay Forum defined the FRF.11 specification. The purpose of the FRF.11 specification is to allow for mul-tiple, concurrent voice calls to flow over a single Frame Relay DLCI. This encapsulation of the voice packet into the FRF.11 encapsulation is also done in the DSP. Then a transmission-ready VoFR packet is delivered to the router’s CPU for transmission over a Frame Relay interface. During an active voice call with G.729 compression, the voice adapter will generate a FRF.11 Frame Relay packet to the router every 15ms.

When a voice packet has been delivered to the router, it must share the Frame Relay interface with the router’s data traffic. However, two key requirements, Delay and Jitter, must be met to insure voice quality. Delay is simply the difference in time between when a person speaks and when the listener hears what is spoken. For typical PSTN quality voice calls, the delay is less than 150ms.

Jitter is the variation in arrival rates of voice packets at the receiving end of the connection. This is important because playback of speech at the receiving end must be continuous. In a perfect world, the voice packets would arrive every 15ms to be decompressed and played out to the receiver. However, in reality, the packets must share the network with other traffic, and their arrival may be delayed. To remove jitter from the playback of voice, a number of packets must be “buffered” before playback can begin. This buffering results in an additional delay in the voice call. The less jitter in the network, the less buffering is required at the receiving side. The frame relay interface can control delay and jitter through two key functions. The first function is data segmentation. Line speeds at the access router are typically in the 64-Kbps range and the routers’ data packets range from 1500 to 17000 bytes in length, depending on MTU sizes of the routers’ interfaces. On a 64-Kbps line, it takes approximately 188ms to transmit a 1500-byte frame. Therefore, a voice packet that is queued behind a 1500-byte data packet will have to wait 188ms before it can be transmitted. The problem is worse when the frame is larger. A smaller MTU size can be set, but it results in many undesired side effects. A better solution is to segment the data only on the Frame Relay link. To solve this problem the Frame Relay Forum developed the FRF.12 specification. FRF.12 defines a Frame Relay level segmentation that allows large data packets to be efficiently fragmented and reassembled at the end points of the Frame Relay network.

Once the data has been segmented, only the problem of priority queuing remains on the access routers’ frame relay interface. Since the data has been fragmented, it is necessary to interleave the voice and data on the Frame Relay interface, which is done by the Bandwidth Reservation System (BRS) in the router. With BRS, you can define separate input queues with associated transmission priorities. For VoFR support, BRS will be enhanced so that the router can separately queue FRF.11 packets (Voice packets) and FRF.12 packets (data) on the Frame Relay interface. BRS will then give priority to the voice packets thus minimizing any delay. Since the voice and data traffic have been integrated on the Frame Relay link attached to the access router, we should now turn our attention to the concentration requirements. There are two different boxes at the concentration site—the VFRAD and the router.

The VFRAD is much like the voice card in the access router, with a couple of main differences. First, the number of concurrent voice/fax calls needing support is much higher in the VFRAD. To reflect this, the connectivity to the PBX is now a digital (T1/E1) connection with 24/30 channels worth of DSP supported voice compression/packetization. The second difference is that call processing functions reside within the box. Since Frame Relay is a point-to-point network, branch-to-branch calling requires that call routing functions are performed in this box. This box also contains the centralized dialing plan information for the network. Central-izing the dialing plan information in the HQ minimizes the amount of unique dialing infor-mation that must be configured in the branch access routers.

The router attaches the headquarter’s site to the Frame Relay Network. This router is the peer of the access box and has the same requirements for support: FRF.12 to segment data and BRS enhancements to be able to prioritize the voice traffic on a DLCI separate from the data. It has the additional requirement of being able to switch the FRF.11 voice packets to and from the interface attached to the VFRAD. This functionality will be implemented across the SELG’s router product family (2210, 2212, 2216).


 Voice Over IP (VoIP)   For the business with an Internet Protocol (IP) network, the advantages are very similar to the FR network and even simpler (See figure 2). In a branch office with an Access Utility (2212) equipped with the voice software and voice FR/IP card, it is ready to cash in on the savings of doing branch-to-branch “toll bypass” calls on an integrated voice/data network. No interme-diate VFRAD is needed. The transition to incorporate voice doesn’t require extra space or more equipment to service and manage let alone NO “fork lifting”. Enhanced software and an interface card for the router are the investment. Additionally, a FR/IP card makes the migra-tion from FR and IP very easy and very economical. Further, the combination FR/IP card allows connectivity to either an IP or FR network which brings added value as a network is transitioned from FR to IP (See figure 3). Lastly, enhancements to router software would still include voice prioritization, security and VPN features.

From a PSTN connectivity perspective, the requirements are the same as the VoFR network. You must be attached to the traditional PSTN devices/network. Convert the voice to PCM, compress and packetize the voice data. The compression algorithms for VoIP can be different, but the current compression recommend by the ITU for low bit rate voice-only connections is G.729. This allows you to use the same voice adapter in the branch routers for both VoFR and VoIP and allows the customer to migrate the access router from VoFR to VoIP with only a software change.