The Rise of VoIP and the Creation of an IP Location Imperative

Before proceeding with the development of a location services architecture for IP, it is worth turning to the next page of the E911 story. This is because the next part of that story is about the development and mass uptake of telephony services on the Internet and other IP networks. This is commonly referred to as voice over IP (VoIP) and, just as cellular had done previously, it raised new challenges for the E911 services network. The examination of how this challenge was addressed by the industry is worthwhile because it highlights some additional requirements that an effective IP location services architecture should satisfy.

The Internet Services Model

For the many people who use the Internet today, the model by which services are accessed is very familiar. For example, there are e-mail services, web-based information services such as online banking, and instant-messaging type services, just to name a few. A common characteristic of these services is that it is generally expected that they can be accessed regardless of the manner in which the user connects to the Internet.

So, for example, if the user connects over their home network using a DSL connection, they can log onto their net-banking web site, enter their username and password information for authentication, and utilize the service. If they connect to the Internet at some other time using a wireless hotspot in a coffee shop, the expectation is that they will be able to access the net-banking service in exactly the same way. The net-banking service does not know or care how the user is connecting to the Internet.

That is, Internet services are generally made available to users in a fashion that is transparent with respect to the provider that the user is obtaining their network access from. This model applies equally to VoIP services. While there are certainly scenarios in which a broadband access provider may also provide a VoIP telephony service whose use is constrained to the single access that they provide, this is a special case and is in no way an essential characteristic of VoIP. Contrast this with wireline POTS (plain old telephone service) and it can be noted that it has been generally necessary to order the telephone network connection from the telephone service operator. If a third party was involved, it has been at a wholesale level with no visibility to the subscriber. This is because the definition of a "telephony service," in a POTS sense, includes everything from the physical network connectivity through to the call processing and call delivery functions. This can be characterized as being a "full services model" as opposed to an "Internet services model." IP telephony being obtained under an Internet services model primarily describes a call processing function that is accessible over any independently provided IP network and regardless of the physical connectivity type. This aspect of the Internet services model may be termed "access-provider service-provider decoupling" and has significant implications for location services in an Internet context.

Nomadicity and Mobility

One of the first visible impacts of VoIP to emergency service processing is that VoIP supports user nomadicity and mobility. Nomadicity refers to the characteristic wherein users may be accessing the telephony service from one location at a given time, even if they are tethered to a fixed point such as a wired home LAN on a broadband service, and somewhat later connect from a completely different location such as a hotel broadband service in another city. While this is not mobility in the sense of the user moving while concurrently using the voice service, it does mean that there can be no presumed association between the identity of the user and the location they occupy. For example, the phone number can no longer be used to determine a subscriber's home address with a presumption that this is their location when making a call.

Further, there is a category of Internet access technologies that supports full mobility of the user. For example, wide-area broadband wireless networks (such as WiMAX and Wibro) allow a user to connect to the Internet and access services even while traveling in a car at high speed.

From the perspective of emergency services, at least for getting the initial location of the caller and correctly routing the call, there is little difference between a nomadic or a mobile scenario. The main impact is that static, predefined, databases cannot be used for obtaining caller location based on, for example, their phone number. It becomes necessary to dynamically obtain the location information from one call to the next. This, of course, is the same challenge that was raised with the introduction of the cellular network, and the same fundamental architecture lends itself to addressing this issue. That is, it is necessary to provide network infrastructure to obtain the location information applicable at the time of the call for the purposes of routing and to provide a mechanism to deliver that information out of the IP network and into the emergency network for display to the call-taker at the PSAP.

In 2003, the National Emergency Number Association (NENA) in North America established a committee to look at the specifics of how to support emergency services (E911) access from mass market VoIP services. The solution was divided into three categories, each with a prefix of i, and were thus labeled i1, i2, and i3. The first of these dealt with basic 911. That is, it only concerned itself with defining some acceptable methods by which a call could be delivered to a PSAP at all, though not necessarily with any associated location information. The second, i2, looked at how to provide full E911 functionality with a common global routing infrastructure and a mechanism for delivering location that would work with minimum impact to the existing emergency network and its associated PSAP infrastructure. That is, it reused as much of the existing emergency network interface functionality as possible. The third, i3, looked at a time further down the track when an all-IP network would be established such that the caller has an end-to-end IP call in place with the PSAP. Working groups dedicated to both the i2 and i3 architecture definitions were established.

The i2 Architecture

In 2005, NENA published the technical document describing the solution defined by the i2 working group (See reference 1 at the end of this chapter). There is a great deal of detail in this document, and Figure 1.11 shows but a fraction of it. However, the salient points remain visible. In particular, note that the same voice and location information paths exist via the selective router and ALI network elements, respectively, as were seen for the cellular architecture shown in Figure 1.5. Based on the mandate to minimize the impact on the existing emergency network elements, the i2 working group maintained and reused these mechanisms that had been defined for cellular emergency support.

Figure 1.11: The NENA i2 architecture.

The VoIP position center (VPC), shown in Figure 1.11, provides equivalent functionality to the GMLC displayed in Figure 1.5. The call-server queries the VPC to learn how the emergency call should be routed, and the VPC also caches the location information in anticipation of receiving a request for location from the ALI system. This is all largely the same as the manner in which cellular calls are treated and it addresses the mobility element of VoIP telephony by supporting the mechanism for obtaining location dynamically on a call-by-call basis.

A significant difference compared to the cellular architecture comes into play when consideration is given to the determination of location. The significance is driven not so much from the technical aspects as by the organizational aspects of VoIP. This stems from the access-provider service-provider decoupling that was defined earlier. There is no inherent need for the provider of the VoIP service processing the emergency call to have any direct knowledge of the form, nature, or ownership of the Internet access being used by the caller. While a cellular network operator can identify the radio cells of their network that is currently servicing the caller, the VoIP service provider has no such similar knowledge. While a cellular operator can direct signaling to the correct part of the network to invoke location determination, the VoIP service provider is reliant on some form of information being sent from the device from which the call is being made in order to begin the process of location determination. ^[1]

^[1]This statement is based on a presumption that, in an access-provider service-provider decoupled situation, the access provider need have no explicit knowledge about the application protocol (a VoIP protocol in this case) being used on the access that they provide. In the case of cellular networks, a caller may be roaming in another cellular provider's network. In the case of traditional switched-circuit cellular calls, it is actually the visited network that deals with the processing of the emergency calls. While this model could apply to IP telephony, it would actually require Internet access providers to support an entire emergency calling application that is potentially independent of the voice service and protocols that users would generally use for IP telephony. This would be a considerably more onerous obligation on Internet access providers than the general location information service that is described in this text.