Voice over data networks

Citation

Curwen, P. (2003), "Voice over data networks", info, Vol. 5 No. 3. https://doi.org/10.1108/info.2003.27205cab.001

Voice over data networks

A regular column on the information industries

Voice over data networksPeter Curwen

It is still the case, as it has always been, that if one picks up a handset, whether fixed or mobile, one will probably be connected to your destination via the Public Switched Telephone Network (PSTN). This connection will be instantaneous and, in the case of fixed-wire connections, extremely reliable – and that is what subscribers have come to expect. Even mobile calls tend not to fail unless one enters tunnels or very rural areas, even if there is rather more interference. For these reasons, an inferior quality of connection is generally considered to be unacceptable unless it is far cheaper than the PSTN, and even then not if the call is important.

Voice over Internet Protocol (VoIP) has until recently been perceived as failing the quality/reliability test without sufficient compensation in terms of cost, but the protective barrier this erects around incumbent operators is increasingly under attack. The reason is nothing new in principle, since local area networks (LANs) have been in existence for a couple of decades. However, to provide competition for the PSTN the term "local" must mean something more than a set of computers and peripherals connected together within an office, or even to another office within the locality. At the heart of the original small-scale LANs lay the Ethernet which provided a common protocol enabling communication between LANs. However, both the distances between connectable LANs and the devices that could be connected were limited. What is happening that provides the threat to the PSTN is the increasing ability to connect up LANs irrespective of their location and equipment using TCP/IP networks, thereby facilitating cost-effective and efficient VoIP.

On the face of it, PSTNs have a considerable advantage that stems from incumbency, not to mention their stability – almost nothing short of severe earthquakes causes anything other than very temporary malfunctions – organised numbering systems, published directories and a variety of special services such as directory enquiries operating via call centres. These were costly to set up and, as recent sales of yellow pages have shown, are still regarded as highly valuable. Data networks operating via computers can certainly be connected to a telephone jack but lack the support infrastructure of the PSTN. Furthermore, the companies involved such as Net2Phone are hardly household names even though it is their job to intercept calls as they emerge from a conventional handset, reroute the voice data via the data network and then convert it back into a conventional voice call at the other end.

Obviously, there would be no point in competing with the PSTN if this process did not save money, and the amounts that can be saved on international calls are considerable. In good part this stems from the fact that the PSTN requires that a continuous pathway be maintained between caller and called, whereas the whole point of TCP/IP is to permit bits of data to fill gaps in whatever happens to be the quickest route open at the time, no matter how long it might be in physical terms, trusting that the destination tags attached to the data will allow them to be assembled in the correct order at the other end prior to delivery. Filling gaps is cheap, but naturally this matters little to the majority of households which make mainly local calls where the effect on the monthly phone bill of converting to a data network might well be insufficient to induce a switch from the incumbent – it is anyway well established that a large proportion of households simply refuses even to consider the matter. Hence the issue of Internet telephony has so far been restricted largely to the corporate sector. The traditional response of incumbent operators has been to lease its lines to a company that wishes to set up a private network – effectively, the operator creates a virtual private network even though it uses the PSTN – which offers some protection to an incumbent's revenue stream. However, in a liberalised world, it becomes possible for private organisations literally to create their own networks which completely bypass the PSTN. Alternatively, public organisations such as central or local government, can set up private networks connecting all of their "subsidiaries" and might open such a network to private organisations if it is deemed to be in the public interest.

From the incumbents' perspective there are clearly two broad strategic responses – restructure to compete or watch the business steadily decline. Since incumbency will restrict the pace of decline, the latter does not by any means imply gradual death by a thousand cuts, but it is evident that if, as in the USA, there are multiple incumbents, a vicious price war could ensue which so seriously affects them that not all will survive. Doing nothing may not, therefore, provide anything other than a short-term response, but it should not be forgotten that legacy circuit-switched networks are fully depreciated and hence can operate at very low unit cost levels until such time as maintenance becomes an issue. Not surprisingly, however, given the growth of data traffic and the limited capacity of legacy networks, plus the fact that the threat posed by VoIP largely stems from its different technological approach to providing essentially the same service, incumbents' thoughts have tended to dwell on upgrading their own technology. Historically, this has meant the introduction of asynchronous transfer mode (ATM) whereby multiple calls can be sent along the same digital circuit using data tags to keep them apart. This has some advantages over, say, Ethernet, because ATM is utilising a network from which the glitches have long been removed and is supported by reliable methods of billing whereas, as readers well know, computer-based networks tend to have their off days. Another issue is latency – that is, does the conversation occur without off-putting delays between words – in which respect the PSTN has no peers although companies with Intranets (private data networks) will not notice any difference. It may be added that if packets sent via TCP/IP are reassembled in the wrong order for whatever reason, or some go astray en route, then combined with latency a conversation can become an interesting experience.

Nevertheless, ATM does not have it all its own way. One reason for this is that voice calls are steadily moving away from the fixed-wire networks towards wireless. Wireless devices can take a variety of forms, although currently mobile handsets far outnumber PDAs and smartphones. The point is that all such devices are increasingly being used for activities other than voice calls, for example Internet access and location finding, and TCP/IP is undoubtedly the way forward in this respect. Very recently, wireless LANs (W-LANs otherwise known as Wi-Fi or "hotspots") have begun to develop rapidly and 3G mobile is on the way. This suggests that flexibility is an important characteristic which clearly favours Ethernet over ATM, although it can be argued that the last thing most companies want is to be constantly reconfiguring their systems and hence they tend to be reluctant to commit to VoIP until they see it as providing roughly the same level of stability as ATM.

At the present time, roughly 10 per cent of international voice minutes – its most obvious initial market – use VoIP. The likes of Ovum estimate that this will rise to 40 per cent by 2006, although it remains unclear how this will be divided between the global wholesalers with managed IP networks and the specialist service providers such as ITXC and iBasis which route calls over the Internet. Naturally, incumbents are fighting back, raising issues such as problems with the available pool of numbers – an issue which is about to diminish in importance as IPv4, with its 32-bit IP addressing system, is replaced by IPv6 with its 128-bit addressing capability – and in the longer term preparing to set themselves up as VoIP wholesalers. In addition, some 70 governments such as in Panama either prohibit or strictly regulate the use of VoIP in order to maintain the flow of income from termination charges, although the reality is that their incumbents, who actually collect the hard currency needed by the governments, may themselves be using VoIP. Nevertheless, the bottom line is that as networks, whether fixed or mobile, carry proportionately more data and less voice, with the household sector increasingly being drawn in by the switch to mobile and the pursuit of recreation and entertainment, voice telephony will become uneconomic to provide as a stand-alone service. Packaged up with data services, there will be economies of scope (whereby it is cheaper to provide multiple services together rather than separately) to be gained from a switch to VoIP. Furthermore, if data services bring in sufficient revenue, operators will be able to keep voice telephony prices down to the level associated with a legacy circuit-switched network. With the cost of modern technology in free-fall, new TCP/IP networks will become as cheap to operate as legacy systems once these become more expensive to maintain, although, as noted, they will retain the advantage stemming from the ability to ignore sunk costs for some time yet. As for reliability issues, these are on the wane, with VoIP pioneer Vocaltec claiming that there is no longer a perceptible difference between legacy and TCP/IP networks, a view supported by ITXC which has developed a method of sending out "probing packets" to detect congestion, enabling it to route its traffic around major obstructions.