Hybrex Business Telephone Systems Small to Large Business Telephone Systems

VoIP - The Technology

The ability to transpose voice streams of analog data to digital, and back again, has gradually pushed its way out from the core of the telephone network to where it is today. Whereas codecs and digital links were only present between exchanges in the PSTN network, the development of services like ISDN have seen the A to D and D to A processes migrate right out into the PABX or Key Telephone System (KTS) of today. The main infrastructure of course has been under the control of the Telcos, or infrastructure providers, with users charged the relevant usage fees.

With the recent and rapid development of the Internet, as an interactive and mass data carrying network, voice over internet protocol (VoIP) is poised to enhance if not supplant a fair proportion of the traffic of the Telcos mentioned above. The primary reason for this of course is economic: in that once the investment is made in hardware and access to the internet in its "Broadband" form is established the cost of voice communications to anywhere in the world with similar equipment is basically just data cost.

Digital Streams

In todays digital telephone systems the analog voice channel is digitised, usually at the handset, for switching by the system itself. This offers the immediate benefits of digital re signal quality etc. If the CO interface is ISDN the voice data continues in digital form with relevant signalling, if PSTN the data is converted back to analog prior to this point. If the interface is a VoIP channel the digital data is divided up into packets for transmission over the internet according to the relevant protocol.


The power of the internet today is primarily due to the development of the many protocols in use. Since the internet is such a complex array of devices and services there has to be sets of rules for behaviour and structures of these sets to maintain orderly operation. These rules and structures are called protocols and they have been developed according to the needs of the relevant services involved. Protocols are often refered to as "stacks" - this is due to the onion like layers of lesser protocols that go to make up a major protocol according to the OSI model (refer to a networking text if you want to go further here).

For Voice over IP there are three main protocols that have been developed: H.323, MGCP, and SIP.

H.323 was developed by the ITU (International Telecommunications Union) and is a telephony centric protocol that has components for telephony networks as well as internets and has an extensive stack to cope with all aspects of telephony or video transport.

MGCP (Managed Gateway Control Protocol) is a protocol that uses a central controlling agent (or computer) in the supervision of the communication between two end point devices (media gateways or MG). MGCP currently uses H.323 as a subset in its operations. With MGCP the MGC (Media Gateway Controller) assumes most of the control duties of the connection, and as such can handle more complex connections. Since MGCP exerts a higher degree of control it is considered to generate more reliable networks. With MGCP each entity (device) wishing to use a network must be registered with the MGC, or Call Agent, of that network.

SIP (Session Initiation Protocol) was developed by the IETF (Internet Engineering Task Force) in response to what they considered was the rigidity of H.323. Whereas with H.323 an MGC controls the media gateways throughout a connection, with SIP the media gateways themselves do most of the controlling of the connection with reference only to a SIP server or proxy for the relevant address they wish to connect with initially. H.323 is an entire protocol suite, SIP is a single module designed to interwork well with existing internet applications. For example SIP defines telephone numbers as URL's, so that web pages can contain them. The inference to be drawn from all this is that H.323 (&MGCP) is here now and reliable but with its rigidity has some limits; SIP is considered the up and coming lad, more relevant to the internet, but requires a lot more smarts in the MG interface and as such may suffer from interwork problems until equipment/application standards become more developed and widespread.

Comparison of H.323 and SIP
Item H.323 SIP
Designed by ITU IETF
Compatibility (PSTN) Yes Largely
Compatibility (Internet) No Yes
Architecture Monolithic Modular
Completeness Full protocol stack SIP just handles setup
Parameter Negotiation Yes Yes
Call Signalling Q.931 over TCP SIP over TCP or UDP
Message Format Binary ASCII
Media Transport RTP/RTCP RTP/RTCP
Multiparty Calls Yes Yes
Multimedia Conferences Yes No
Addressing Host or Teleph No. URL
Call Termination Explicit or TCP release Explicit or Timeout
Instant Messaging No Yes
Encryption No Yes
Size of Standards 1400 pages 250 pages
Implementation Large and complex Moderate
Status Widely deployed Up and coming
VoIP Call with MGCP (an example)
VoIP Call with MGCP

VoIP Call Sequence

The sequence below is only approximate as the actual protocol and process is more complex than this.

1. Person A at Company A wishes to call Company B lifts receiver and selects VoIP trunk, MG allocates bandwidth

2. MG (VIU card in Hybrex GX) knows Call Agent (MGC) address and signals request for service to Call Agent

3. Call Agent receives request via internet on designated port.

4. Call Agent acknowledges request from (Company A) and

5. Call Agent instructs MG at Company A to serve dial tone and waits for addressing.

6. Person A enters desired number (the extension number of desired port at Company B) which is transmitted to call agent.

7. Call Agent establishes a connection with Company B whose address ( it knows from its database, and instructs MG there - Setup to port (No.dialled). From

8. System at Company B rings relevant extension and returns an Alert message to the Call Agent which is passed back to Company A that dialled extension is ringing at

9. When extension at Company B is answered a Connect message is sent back to the calling party (Company A ) and a channel is established between the two - first to negotiate codecs/rates etc (using RTCP), then the data flows (using RTP) and the call is in progress. At this point the Call Agent is not part of the data stream but oversees call progress, tearing down the connection when the call is finished, or re-assigning if either party wishes to make further calls.

A point to note - this whole process happens pretty quickly these days.

Quality of Service (QoS)

Although the transport protocol used is RTP (Real Time Transport protocol) which is optimised for real time data such as audio. The data is still packetised and the nature of the internet is such that different packets may take different paths through the net. Different paths mean different transit times with the possibility of arriving to late to be used (RTP is not re-transmitted) or being dropped (disappearing) along the way; particularly if a particular route is congested. Events such as this affect the speech quality of the connection. Better quality of service, or QoS gaurantees are provided by some ISP/vendors, albeit at a price!. But bandwidth rollout continues apace and improvements over the already generally good service available continue.