Voice over IP with Obstacles – Why there are still problems

Bob Dalton, Managing Director of Intact Integration Services, an independent, international networking IT services and support company focussed on advanced technologies says there are still problems with VoIP

“Voice over IP (VoIP) is the future; Companies with any sense can see the advantages a uniform infrastructure for data and voice communications can offer. Impressive cost savings with respect to equipment and administration support the convergence of IT and telecom. Sales of conventional telephone syste0ms, for example, are declining in favor of IP-based systems. Nevertheless, many companies still have serious reservations about using VoIP. There are many reasons why: they range from difficulties with voice quality in cross-corporate IP telephony to security problems with unencrypted calls. The following will illuminate where the problems lie.

Telephony via the Internet transmits voice data in real time in IP packets based on RTP (Real-time Transport Protocol). This is based on UDP (User Datagram Protocol), which, as a wireless network protocol, is responsible for addressing, and not for the security of the system, to allow for a quick transmission of the data.

TCP (Transmission Control Protocol), which lies on the same level as UDP and is used for the transport protocols for data communication in IP networks, cannot be used for VoIP because TCP requires handshakes to resend defective or lost data. This would take up far too much time for voice communication.

In the same way as for a PSTN (Public Switched Telephone Network) or ISDN (Integrated Services Digital Network), a signalisation protocol is required to establish or terminate an RTP session. SIP (Session Initiation Protocol) from the IETF (Internet Engineering Task Force) is now also becoming increasingly popular as a quasi industrial standard, although a broad range of systems that work with H.225.0 and H.245 protocols from the ITU (International Telecommunication Union)-T (Telecommunication Standardization Sector) H.323 protocol family are still in use. And there are also proprietary protocols like SCCP (Skinny Client Control Protocol) from Cisco or the Skype protocol currently on the market.

Same codes and signalling required

In any IP telephone or terminal device, the analogue voice is transformed into digital data, whereby different, sometimes very complicated processes are used for voice coding. The very frequently used voice coding based on the G.729 standard requires a bandwidth of only 8 kbps per second for a relatively good-quality telephone call. Higher-quality voice communication is delivered by PCM (Pulse Code Modulation) based on the G.711 standard, but this process requires a bandwidth of 64 kbps. However, there are also a number of other codec processes, such as ADPCM (Adaptive Differential PCM) based on the G.726 and G.727 standards. Only IP telephones that use the same coding and decoding process can communicate with one another. The fact that it is also necessary to use the same signalling increases the complexity of the entire system considerably. Most telephones usually support several different concepts – but there is hardly one that can do it all.

Each voice packet carries a lot of overhead. If, for example, 20 milliseconds of voice are to be transmitted as a voice segment in 20 bytes, the entire packet is at least three times as big. The RTP header itself takes up at least 12 bytes, and then there’s a UDP header (User Data¬gram Protocol) with 8 bytes and an IP data header with at least 20 bytes. That’s why it makes sense to compress the header in some applications.

Jitter and delay cause problems

In principle, voice communication within a company – even large companies with many different locations – does not pose any technical problems using an appropriately sized IP-based network as long as the required quality of service (QoS) is met. The QoS requirements primarily pertain to the bandwidth of the virtual connections between IP telephones, the end-to-end delay of the voice signal, the fluctuation of the transmission time (jitter) and the IP packet loss rate.

To ensure that voice data travels through the network at the necessary speed, it is prioritised. This means that the data streams flow according to their priority classification. There are different concepts designed to achieve this: they include differentiated services, processes for queue management with IP packets before data lines and the RSVP (Resource reSerVation Protocol) with the common goal of making voice communication via IP possible.

End-to-end delay is key

An important factor for quality in telephony is the time it takes to send the voice signal from the mouth of the speaker to the ear of the listener. The G.114 ITU-T recommendation specifies 300 milliseconds as the upper limit at which communication can be classified in real time. However, some people find a time lag of 125 milliseconds annoying. That’s why the ITU-T recommends 150 milliseconds as a value that would be acceptable for everyone.

Among other things, this delay is caused by the intermediate storage of the IP packets in the routers, because every router needs time to interpret the IP header and then forward the packet. It really can come to a time delay in calls from the company network or the administration into the public Internet or a network at another company and there are several different annoying effects that can occur. On the Internet, there is no QoS, all the packets are handled in the same way. The only thing that can reduce delay and jitter is greater bandwidth. Another problem for effective communications is the many different standards and the protocols, which are not compatible with one another: in addition to SIP and H.323, the Skinny Client Control Protocol (SCCP) from Cisco and the VoIP protocol from Skype also play a role. The translation of the protocols usually has to be done by dedicated gateways, which has a negative effect on the time factor.

The many different codecs used might also make it necessary to use separate gateways for the conversion. The infrastructure balloons, becomes more expensive and jitter and delay increase. In the end, voice quality does not meet the requirements,, productivity slumps and the users pay the price.

Security requires resources and time

Voice over IP is just as vulnerable to criminals on the Internet as any other IP traffic. If communication travels public paths, encryption is indispensable. VPN (Virtual Private Network) encryption is then used in most cases.

To connect two mobile employees or two networks, one or two gateways or concentrators is required, depending on the application. The coding and decoding of the data contributes to the end-to-end delay. As a rule of thumb: the more complex the encryption, the longer the time delay.

The structure of the VPN network can also play an important role. Voice traffic in a WAN via an IPSec-VPN with star structure usually takes longer than via an MPLS (Multi-Protocol Label Switching) VPN. In MPLS, a virtual connection is already used to find an optimal route to the destination. Then the IP packets are sent on their way, one after another. This guarantees the sequence the IP packets are sent in and they all travel the same physical pathway through the network – which results in fewer jitters.

To be equipped to handle attacks such as eavesdropping on calls or “logging in” to system components, telephone systems have to offer encrypted IP telephony and support SRTP (Secure Real-time Transport Protocol). SRTP provides greater security in transmission based on the RTP protocol. As a result, the RTP packets are encrypted before the voice traffic is sent and cannot be intercepted by third parties. In addition, the sender of RTP packets can be authenticated without a doubt and the packets can be protected against any alterations during transmission.

The bottom line: VoIP is becoming increasingly important for companies and administrations, but the ideal VoIP world is yet to come. Too many different standards and processes on the market mean that many of the systems are not compatible, which has an adverse effect on their popularity. Since there is currently no well-established “IP telephony clearinghouse”, communications from company to company are still largely dependent on bilateral agreements. That’s why it is very advisable for companies that want to set up a VoIP infrastructure to trust in the know-how of experts in this field.