Circuit versus Packet

With traditional circuit-switched wireless data, dedicated circuits are allocated to subscribers whether or not they are being used. In theory, this provides higher effective bandwidth to users by dedicating a full channel to each user. The data service is provided via a wireless dial-up model, similar to wireline dial-up remote access. The user dials a phone number associated with the network access server (NAS) used for a specific wireless data service. Once the physical connection—that is, the circuit—is established between the mobile station and the NAS, PPP is used to provide end-to-end link layer service. Terminating the user PPP session can be easily accomplished using simple dial-up techniques based on off-the-shelf modem banks or remote access servers (RAS), comprising an Interworking Function (IWF) functionality, like the 3Com (now its spin-off, CommWorks Corporation) Total Control RAS chassis, with some software upgrades making it suitable for wireless environment. The IWF is normally required to terminate the wireless access protocols (Radio Link Protocol, or RLP) and interwork with the Public Switched Service Telephone Network (PSTN) when needed. In some cases the IWF also can relay PPP to a private network using L2TP (details on this and other options are given in Chapters 6 and 7).

In contrast, wireless packet-switched data technologies are based on a wireless access network support for statistical multiplexing of user sessions over the radio interface. Packet data network resources are only consumed during the transfer of the data and unused during idle periods, resulting in a more efficient system in which any source of traffic can use resources not used by others. Known as statistical multiplexing, this is an important property of all packet data systems. Statistical multiplexing makes packet data systems more efficient than circuit-based systems, which guarantee each user a separate, dedicated channel, not fully utilized with bursty data transmissions patterns. However, it also means that users of shared media networks must contend for the available bandwidth, sometimes resulting in congestion, delays, and lower effective throughput per user.

Access contention for shared resources is a typical problem not only for cellular packet environments but also for Wireless LAN. In cellular systems supporting packet mode access, to make efficient use of resources, the radio access bearers are only temporarily allocated to a specific user. After a period of inactivity, the mobile station enters an idle (for instance, in GPRS) or dormant (CDMA2000) mode of operation. This mode allows the mobile station to be constantly reachable by signaling and data sent to its network layer address using location update procedures and paging, while no dedicated resources are active to allow the MS to send and receive data. When data needs to be received, the MS is paged, "wakes up," and issues a request to set up the radio bearer that would allow data reception. The MS issues the same request when it needs to send data and when no radio bearer is already set up.

Packet data user mobility support is conceptually similar across different wireless data systems. It is based on various tunneling mechanisms such as Mobile IP (adopted by CDMA2000) and GTP (adopted by GSM and UMTS), both of which are analyzed later in the chapter. This common packet data tunneling model is shown in Figure 4.1. Tunnels depicted in this figure by thick dashed (for former, discontinued) and solid lines (for current, active) are dynamically established between the mobile station's temporary point of attachment to the wireless network and a tunnel "anchor point" or home network that also acts as a gateway to the mobile data network from which the user is receiving access service. As mobile stations dynamically change location within the network—for example, traveling through certain geographical area from Mobile Switching Center (MSC) to MSC or being at the MSC boundary—tunnels are dynamically established between the MS home network and the visited wireless access network.

Figure 4.1: Wireless packet data tunneling mechanism.

The majority of early data users familiar with using wireline dial-up remote access services had little or no trouble adapting to the wireless circuit-switched data services. This helped in driving circuit data acceptance rate fairly high, especially considering its low bandwidth and other limitations, which limited the willingness to frequently use the service and to stay online for a long time. All the familiar wireline dial-up access features were also present in wireless circuit data: familiar login-password sequence, ability to access corporate network by simply dialing specific telephone numbers, identical configuration procedures on user client devices such as laptop computers.

A relatively similar pattern of service adoption for wireless packet data technologies such as GPRS has emerged, although we have recorded some longer-than-expected service take-up periods mostly caused by the lack of volume terminal production and pricing experiments by wireless carriers. Wireless packet data technologies do not require dialing a number to reach a specific NAS in a corporate or ISP network. Instead, users enjoy the simplicity of always on or on-demand connectivity to the Internet or their corporate network. However, in most cases, this does require predefined relationships between the corporation and wireless carrier, as opposed to circuit data networking.