7.4 The Mobility Challenge

Mobility is an important feature in cellular networks and in any wireless network. Hence, it has been a key design element and an integrated part of current cellular network architectures. However, this is not the case with IP networks, and hence mobility can be considered as one of the biggest challenges for IP. This is because IP was initially envisioned as a universal standard for connecting different hosts or computers for data communications. In spite of its great success, application of IP was limited to fixed or stationary hosts . Without support for mobility, the applicability of IP to cellular networks is quite limited and may result in wireless-specific solutions to handle mobility. GPRS is a perfect example on how mobility-unaware IP can be applied to wireless networks. Until recently, there was no initiative to enhance IP to be mobility aware. Mobility in cellular networks is twofold:

1. Idle mobility

2. Handover

In the following sections, these two aspects are discussed to show how they are being handled by IP.

7.4.1 Idle Mobility

Cellular networks are operated by different service providers, and each service provider manages the network by dividing the network into manageable network areas in a hierarchical fashion, all the way down to the cell level. Mobile nodes are identified by location based on which cell the user is presently in. Cellular networks perform location management by continuously tracking the location of mobile nodes with the help information received from the mobile nodes. The location information determines the cell (or a larger network area) where the mobile node is currently located. The location information is broadcast to all the mobile nodes in the network or cell area.

When someone calls the user, the network infrastructure, and the mobile switching center in particular, retrieves the latest location information of the mobile node and delivers the call. Mobile nodes periodically update their location information to the network. These location updates are otherwise referred to as idle mobility because they are performed when the mobile nodes are not engaged in any active call or other services. These updates are mainly timer based or event based, in case the mobile nodes may cross the network management areas or even network borders into an area operated by a different service provider (Figure 7-2).

Figure 7-2. Idle mobility in cellular networks.

The IETF Mobile-IP Group has defined two different mobility mechanisms for IP, one for IPv4 and another one for IPv6. The fundamental principles are similar, but the protocol capabilities and definitions are quite different. The basic principle is to provide a local care of address (CoA) to the mobile node while it is away from the home network. The mobile node may have a permanent or home address assigned in the home network. The local CoA is provided by the visiting or foreign network where the mobile node is currently present. In IPv6, the mobile node can form its own local CoA through stateless address autoconfiguration by listening to the router advertisement messages from the routers serving that subnet. The mobile node performs a registration or a binding to the home agent (HA) to indicate the local CoA. The HA creates a binding cache between the permanent address and the CoA to tunnel all the packets addressed to the permanent address to the CoA. The same binding is also established at the correspondent node (CN) if there are any active sessions with any nodes. Whenever the mobile node's network point of attachment changes, it obtains a new CoA and performs updates on the binding to the HA and CN. Figure 7-3 shows mobility in mobile IPv6.

Figure 7-3. Mobile IPv6.

If the mobile is far away from the home or CN, it may take a more latency to update the binding. To reduce the frequency of these updates, micro or localized mobility management (LMM) mechanisms are proposed. Cellular-IP, Hawaii, Regional Registration, and heirarchical mobile IP (HMIP) are some of the techniques.

7.4.2 Handover

Another aspect of mobility associated with active calls or sessions is called handover or handoff . When the user is engaged in an active conversation, handovers can be triggered due to movement of the mobile node from one cell to another. The network along with the mobile node can monitor radio conditions like signal strength and capacity constraints, which may be the actual reasons for these handover decisions. The handover process results in moving the mobile's current network anchor point to a different one in the target network.

In cellular networks, the mobile, while it is engaged in an active call, provides periodic signal strength information due to neighboring cells to the current radio network. Based on signal strengths and other criteria, the current or the source cell makes the decision to hand over the mobile node to a target cell. It proactively signals the target cell that the handover will be performed so that the target cell can establish the necessary radio channels before the mobile arrives and provide the information back to the source cell. The mobile node is instructed to perform a handover by providing the target cell information. This kind of handover procedure is called a "make before break" mechanism.

In the case of IP networks, without any special handover support, the handover across different network point of attachments (routers) happens in a passive manner. The mobile node loses the network connection at the present router; it later establishes the radio connection at the target cell and obtains the router information and a new CoA at the new router and updates the binding with home agent and CN, assuming there is no micromobility management. This long and cumbersome procedure is not suitable for real-time communications simply because the call break is a long and noticeable break. Cellular-type handovers require that the glitches or break in communications cannot be greater than 150 ms.

Realizing this importance of handover, a fast handover proposal was introduced into the Mobile-IP group in which extensions to base mobile IP protocol were proposed. According to this proposal, the mobile node must make the decision to perform the handover and inform the source router. The source router determines the target router based on information provided by the mobile node and requests to provide a CoA at the target router. When the mobile node loses the network connection at the old router, the packets are tunneled to the target router with the new CoA. The mobile node accesses the target router and is immediately able to configure the new CoA and receive delayed packets. Although this mechanism may work, it may not be practical for the cellular environments. Situations where radio networks can make handover decisions cannot be handled with fast handover. Another mechanism to handle such situations has been proposed. Bidirectional edge tunnel handover (BETH) defines a handover procedure that reduces network-layer signaling in fast handover and performs handover based on triggering from radio link layers .

Another aspect of terrestrial mobility is to allow mechanisms to exchange user subscription, security, accounting, and service information between different networks. The cellular protocols have defined roaming protocols like IS-41 in CDMA and US TDMA networks and GSM MAP for GSM networks to exchange user subscription information related to security and services to allow roaming from one network and another. The IETF has formed an authentication, authorization, and accounting (AAA) working group to address the large-scale deployment of IP-based mobility, security, and accounting mechanisms. AAA developed a Diameter protocol that provides solutions for the roaming functions and intranetwork, internetwork, and interdomain operations.

7.4.3 Access Independent IP Mobility

The previous sections presented high-level mobility aspects of cellular networks. But the details of the mobility mechanisms in each of the cellular networks at the radio level and roaming across different networks are dependent on the protocols used for that specific cellular technology. For example, mobility functions defined for IS-95 CDMA cellular networks are relevant to only IS-95 CDMA terminals. Users can roam only to the networks that support the same IS-95 CDMA technology. In the same way, GSM users can roam only to similar GSM networks. When other noncellular access technologies (e.g., WLAN) are considered, it is even worse since currently there is no common network infrastructure and protocol exchange to support roaming between these dissimilar access networks.

The IP mobility protocol, defined as part of network layer (IPv6) or above (IPv4), can provide an elegant solution to interaccess network roaming. IP-level mobility provides an abstraction to layer 2 access technologies by hiding specific access network protocols. Any mobility functions defined for access-specific technology can be used for micromobility functions that are meaningful only within that specific access network. IP mobility provides macromobility functions to determine the network location of the mobile node on a global scale. From this aspect, IP mobility can enable seamless mobility while roaming and possibly even handovers across different access technologies. Further, it simplifies the mobile terminal by adopting common mobility-level functions above any specific access link protocol stacks.

7.4.4 Dormancy and Paging

Cellular networks support dormancy for mobile nodes that are idle and not engaged in active conversations. Dormant mobile nodes do not perform frequent updates of their network location information at the cell level. Instead, they wake up only when they move across a larger network area. The main benefit of dormancy is to save power on the mobile nodes, since frequent location updates to the network drains the power. The network does not keep track of dormant nodes at the cell level but within a greater network area where the mobile is currently dormant. The mobile nodes remain dormant until there is a need to wake up and update their exact location to the network. When mobile users initiate calls, the mobile node wakes up to perform a cell-level location update, obtain the radio channels for signaling exchange, and then exchange call control signaling to set up the call. For mobile terminated calls, the MSC issues a paging request to all the BSCs in the area where the mobile node was previously registered dormant. The mobile node responds to the page from the BSC and performs wakeup functions; the call is then delivered to the exact location where the mobile is currently located.

Wireless IP networks can be similarly divided into several paging areas. The paging area information can be broadcast with the help of specific radio broadcast capabilities. The mobile node can remain idle within the paging area without needing to perform idle mobility procedures, thus saving power. The mobile node can switch to dormant mode by registering itself as a dormant node to a network element that handles dormancy and paging functions. It needs to wake up only when it crosses the paging area to update its new paging location. Any downstream traffic toward the mobile node triggers a paging request to wake up the mobile node within that paging area.

The benefit of IP-level dormancy and paging is twofold. It offers these power-saving functions to wireless access technologies like WLAN that do not have such capabilities at the layer 2 level. Although all cellular technologies do provide these functions, implementing IP-level dormancy offers transparency between the layer 2 functions and the layer 3 functions. Another incentive to IP-level dormancy and paging is due to its access network independence, as discussed in the previous section.

Paging when combined with mobility management protocols can provide a very desirable solution for dormancy of mobile hosts in IP networks.