In this section we describe the core packet data architecture associated with the CDMA2000 radio interface. This architecture is described in 3GPP2 recommendations and TIA standards such as [IS835] and [TS115]. It allows CDMA2000 cellular wireless service providers to offer bidirectional packet data services using the Internet Protocol. To provide this functionality, CDMA2000 utilizes two access methods: Simple IP and Mobile IP.
In Simple IP, the service provider must assign the user a dynamic IP address. This address stays constant while the user maintains connection with the same IP network within a wireless carrier's domain—that is, until the user does not exit the coverage area of the same Packet Data Serving Node (PDSN). Anew IP address must, however, be obtained when the user moves into a geographical area attached to a different IP network—that is, into the coverage area of another PDSN. Simple IP service does not include any tunneling scheme providing mobility on a network layer described in the beginning of this chapter and supports mobility only within certain geographical boundaries.
Note |
One of the significant advantages of Simple IP lies in the fact that unlike Mobile IP it does not require special software of any kind to be installed in the mobile station. All the MS needs is the CDMA2000 terminal capabilities and a standard PPP stack similar to that used to establish wireline dial-up session, usually bundled with most modern operating systems such as PocketPC2002 and Windows XP. |
The Mobile IP access method is mostly based on [RFC2002] now superseded by [RFC3220], described in Chapter 2. The mobile station is first attached to serving PDSN, supporting FA functionality, and assigned an IP address by its Home Agent (HA). Mobile IP enables a mobile station to maintain its IP address for the duration of a session while moving through CDMA2000 or other systems supporting Mobile IP.
For mobile stations compatible with a TIA/EIA [IS-2000] standard attached to a CDMA2000-1x network, available data rate can vary between the fundamental rate of 9.6 Kbps and any of the following burst rates: [3]
19.2 Kbps
38.4 Kbps
76.8 Kbps
153.6 Kbps
These higher-speed bursts are allocated by the infrastructure based on user need (data backlog in either direction), and resource availability (both airlink bandwidth and infrastructure elements). Bursts are typically allocated to a given mobile for a short duration of time of 1 to 2 seconds. The resource and mobile situation is then reevaluated. Burst allocation is performed independently on the forward and reverse channels.
The architecture of CDMA2000 data system is based on the following components (as shown in Figure 4.3):
A mobile station in a form of a handset, PDA, or PCMCIA card in handheld/portable computer supporting Simple IP or Mobile IP client or both
CDMA2000-1x Radio Access Network (RAN)
Packet Control Function (PCF)
Packet Data Serving Node (PDSN) supporting FA functionality in case of Mobile IP
Home and foreign AAA servers
Home Agent (for the Mobile IP access method)
When the mobile station connects to the CDMA2000 base station, it first establishes a connection to a PDSN. In the case of Mobile IP, the mobile station is then connected to its serving HA by a tunnel between PDSN/FA and the HA established using Mobile IP. The IP address of the mobile station is assigned from the address space of its Home network, either statically provisioned or dynamically allocated by the HA at the beginning of the session. On a high-level Mobile IP authentication and authorization is normally performed by both the PDSN and HA by querying the AAA infrastructure (more on this in Chapter 7). In the case of Simple IP, the address must be assigned to mobile station by the PDSN and cannot be statically provisioned in the MS. The authentication for this access method is based only on PDSN.
The connection between the mobile station and its serving PDSN requires a second layer of connectivity to be established for successful IP communication. This connectivity is provided by Point-to-Point Protocol (PPP) as defined by [RFC1661] and supporting IPCP, LCP, PAP, and Challenge Handshake Authentication Protocol (CHAP). [4] PPP is initiated by the mobile station during connection negotiation and is terminated by the PDSN. Between the CDMA2000 radio network and PDSN, PPP traffic is encapsulated into the Radio-Packet (R-P) interface. CDMA2000 protocol stack examples for both the Simple IP and Mobile IP cases are shown in Figure 4.4.
The PCF shown in this figure is the element of the CDMA2000 Radio Access Network responsible for R-P interface setup and processing. It is often implemented as a component of CDMA2000 MSC. One exception is CDMA2000-1xEV DO architecture, which does not rely on MSC. The PCF can be implemented there as a part of 1xEV Radio Network Controller (RNC—or BSC depending on the vendor). Stand-alone PCF implementation is also possible. Once link layer connections are established, the PCF simply relays PPP frames between the mobile device and the PDSN. Another important function of PCF is providing micro-mobility support, which is accomplished by allowing the MS to change the PCF while keeping the mobile anchored on the same PDSN and buffering the user data while a dormant radio link is being re-connected. The significance of the latter feature is explained later in the chapter.
The major role of PDSN in CDMA2000 architecture is to terminate PPP sessions originated from the mobile station and provide FA functionality, in case Mobile IP service is requested, or to deliver IP packets to the appropriate next hop when Simple IP is used. The PDSN is also charged with authenticating the users and authorizing them for requested services. Finally the PDSN is responsible for establishing, maintaining, and terminating the PPP-based link layer connection to the mobile station. Optionally, PDSN must support secure reverse tunneling to the Home Agent (this option is described in detail in Chapter 7).
For basic Internet service using the Simple IP access method, the PDSN assigns a dynamic IP address to the mobile, terminates the user's PPP link, and forwards packets directly toward the Internet via the default gateway router on the service provider backbone IP network. The normal PPP timers are enforced, and the packets from the mobile may be checked to ensure the mobile is using the source IP address assigned by the PDSN. (Among other filtering rules and policies, the PDSN may implement in Simple IP mode.)
For Mobile IP access methods, the PDSN establishes the Mobile IP protocol connectivity to the mobile station's home network represented by the HA, which is responsible for IP address assignment. The PDSN must support an AAA client functionality to aid in partial authentication of the mobile by local AAA server. Per [IS835], the PDSN is also required to support Van Jacobson TCP/IP header compression and three PPP compression algorithms: Stac LZS [RFC1974], MPPC [RFC2118], and Deflate [RFC2394]—the latter mostly used by Linux- and UNIX-based mobile stations.
The R-P interface connecting PCF and PDSN—also defined by TIA/EIA as A10/A11—is an open interface based on the GRE Tunneling Protocol and is used to connect radio network and PDSN. The R-P interface protocol is actually similar to the Mobile IP where the PCF acts as the FA and the PDSN acts as the HA (the R-P interface uses GRE tunnels for the traffic plane and Mobile IP-like RRQ/RRP messages for signaling). There are a few reasons for the introduction of R-P interface or in other words "splitting" PCF and PDSN functionalities. By supporting the R-P interface, IP-based mobile devices can cross MSC boundaries without impacting the continuity of user sessions. In other words, if the user moves to another MSC coverage area, the user session is not disconnected and the user is not forced to reconnect via the new MSC and obtain a new IP address. This is accomplished by performing PCF transfers while keeping mobile devices anchored to the same PDSN. This does, however, require that all serving PCFs have network connections to the same pool of PDSNs (see the case study in Chapter 7). Another purpose of splitting PDSN from PCF is to allow service providers to select PDSNs from third-party vendors, other than those proving the bulk of their infrastructure including MSCs and PCFs. R-P therefore enables wireless carriers to introduce multivendor PDSN solutions into their network. Not surprisingly, the carrier community was the most vocal during the R-P standardization process.
The CDMA2000 mobile station can authenticate with the service provider's HLR for wireless access and authenticate with the PDSN and HA, using the Simple IP or Mobile IP access methods, for data network access. The mobile stations are required to support a standard PPP networking protocol and be capable of supporting CHAP-based authentication during PPP authentication phase for Simple IP service. For Mobile IP service, the mobile device must also support the Mobile IP client as described by the [IS-835]. In this mode, the mobile station communicates with its Home Agent via serving PDSN in the visited network. If the mobile supports one or more of the optional PPP compression algorithm options such as MPPC or Stac LZS, then PPP compression during the connection phase with the PDSN can be negotiated, thus optimizing radio network resources usage and enhancing the user experience via a higher effective data rate.
The mobile device is expected to support airlink "dormancy" (as defined by TIA [IS-707A1]), which allows either the mobile or the MSC to time out the active airlink connection after a period of inactivity and to release the air interface and serving base station resources. If either the mobile station or the associated PCF have packets to send while dormant, the connection is reactivated and the transmission continues. Dormant mobile stations are defined as stations that do not have an active link layer connection to the serving PCF. All mobile stations—active and dormant—registered using Mobile IP access method have an entry in the PDSN visitor list and a binding with the corresponding HA. [5]
The PDSN serving the users on the foreign network serves as the default router for all registered mobile users, active and dormant, and maintains host routes to them. For Mobile IP mode the PDSN/FA keeps track of the time remaining of the registration lifetime for each mobile station in its routing tables and the MS is responsible for renewing its lifetime with the HA. If the mobile does not re-register before the expiry of the registration lifetime, the PDSN will close the link with the PCF for this mobile and terminate the mobile's session (and the HA will do likewise if the mobile has not re-registered via some other PDSN). Once the mobile station's registration lifetime has expired, the PDSN/FA will stop routing packets to it. To receive and send packets, dormant stations must therefore transition to the active state. Given that any registered mobile stations at any moment can be in active or dormant sub-states, the PDSN generally does not require an indication of the state of PPP links to mobile stations except for the current dormancy timer value for that particular link. Traffic may arrive on the dormant link at any time, forcing the associated mobile station to transition to active state. For active, traffic-carrying PPP links, the PDSN terminates the PPP session with the mobile station and relays the encapsulated IP traffic to the mobile from the HA or from the mobile to the HA via reverse tunneling. A separate tunnel exists for each unique HA for all registered users.
There are two basic types of mobile station configurations—relay model and network model. In relay model mobile stations, the CDMA2000 Mobile Terminal is connected to another portable data terminal device such as a laptop, handheld computing device, or some other embedded data terminal. The relay model phone does not terminate any of the protocol layers except for the CDMA2000 physical layer (radio interface) and RLP layers. The attached data terminal device must terminate all other higher-layer protocols (PPP, IP, TCP/UDP, etc.).
Network model mobile stations, in addition to the radio interface, terminate all necessary protocols and do not require any additional data terminal devices. The mobile phone itself provides all user input and display capabilities—as well as a user applications—to make use of the packet data network. Examples of this kind of phone include the "smart phone" or "micro-browser" phone. These devices normally include some embedded Web browsing or information service application, as well as a display screen for viewing the information retrieved from the Internet server. Such kind of terminals may also offer the ability to connect a laptop to a data network via a PPP connection terminated at the terminal itself. In this configuration, the phone can support embedded applications such as a micro-browser and also allow general-purpose use by the external data endpoint.
CDMA2000 packet data architecture defines as many as three levels of mobility for the mobile station, as depicted in Figure 4.5. One level is represented at the physical layer by BTS-to-BTS soft or semisoft handoff, while the mobile station is anchored at the same PCF. This is accomplished by the CDMA2000 radio access and is invisible to both PCF and PDSN.
The second mobility level is represented by the R-P interface on the link layer, which allows for a transparent handoff from PCF to PCF while keeping the session at the same PDSN. In this case, two options described previously come into play: dormant and active. In active state, when the user crosses PCF boundary, the handoff is transparent for the mobile station. The MS participates in a semisoft handoff to the new BSC (or MSC, depending on the vendor), while the link layer data session remains anchored to the original PCF for the duration of the call and the mobile is in the active state. In other words, when the mobile station is in the active state, change of serving PCF will not occur.
When a mobile crosses a PCF coverage boundary while dormant, the mobile will trigger reactivation at a new BSC (MSC) to establish a new PCF connection. That results in a PCF but not necessarily a PDSN change if both the current and previous PCFs were attached to the same PDSN. The new PCF attempts to assign the mobile to its current serving PDSN. If the new PCF has connectivity to that PDSN, the PPP session previously established between the mobile station and the PDSN will be totally unaffected.
The third level of mobility, the network layer, is the inter-PDSN handoff, based on the use of Mobile IP protocol. Let's assume that the mobile station has registered with the HA and PDSN (the MS has been authenticated by each of them) to establish the Mobile IP tunnel over which traffic is delivered. Whenever the mobile roams to a location that is served by a PCF connected to a different PDSN, the mobile receives an indication that it must reregister with this new PDSN. This reregistration updates the mobility binding tables at the HA, so that all subsequent traffic is routed to the new PDSN for this mobile. In this case the mobile's PPP link is impacted by this change while the IP layer stays intact, and the mobility remains invisible to the mobile station's correspondents.
Note that the last type of handoff is not available in Simple IP mode; Simple IP provides only partial mobility, via the other two levels, to the mobile station. One of the functions of the R-P interface is to bring Simple IP service closer in functionality to Mobile IP service, along with addressing other problems. For example, it addresses the situations where the mobile station changes its point of attachment to the network so frequently that basic Mobile IP tunnel establishment introduces significant network overhead in terms of the increased signaling messages. Another often-cited problem is the latency of establishing each new tunnel, which introduces delays or gaps during which user data is unavailable. This delay is inherent in the round-trip incurred by Mobile IP as the registration request is sent to the HA and the response is sent back to the PDSN.
CDMA2000, just like the majority of other cellular systems, supports the concept of home and visited networks. A CDMA2000 subscriber has an account established with one wireless carrier, which provides the user with wireless voice and data services. This same wireless carrier may provide a home network for the mobile subscriber. The home network holds user profile and authentication information. When the user roams into the territory of a different wireless carrier—that is, a visited network—that carrier must obtain the authentication information and the service profile for this particular user from its home network. The service profile indicates what radio resources the user is authorized to use, such as a maximum bandwidth or access priority. In CDMA2000 the user profiles are stored in a Home Location Register (HLR) located in the home network and are temporarily retrieved into a Visitor Location Register (VLR) located in the serving network. The HLR and VLR are databases housed on fault-tolerant computing platforms. Similar procedures take place to authenticate the user access to data networks. CDMA2000 packet data architecture, depicted in Figure 4.6, is based on the concept of home and foreign data networks represented by HA and PDSN and home and foreign AAA servers—for example, RADIUS or DIAMETER—as defined by [RFC3141].
Mobile stations requesting data service in CDMA2000 systems will have to be authenticated twice: on the physical layer and on the link layer (or, using other terminology frequently used in the industry, both wireless access and network access authentication will take place). Physical layer (or wireless access and user terminal equipment) authentication is performed by the cellular wireless system's HLR and VLR infrastructure. It is based on an International Mobile Station (IMSI) and is defined in [IS-2000] (the details of this authentication method are outside the scope of this book). CDMA2000 link layer, or packet data network access, authentication of the mobile station is conducted by the infrastructure of AAA servers and clients, the latter being hosted by PDSNs and HAs. It is based on a Network Access Identifier [6] (NAI, defined by IETF [RFC2486])—that is, a user identifier of the form user@homedomain, which allows the visited network to identify the home network AAA server by mapping the "home-domain" label to the home AAA IP address. A challenge from the PDSN also allows for protection from replay-based attacks.
Among other services, NAI allows for distribution of specific Mobile IP security association information to support PDSN/HA authentication during mobile registration, HA assignment, and inter-PDSN handoff. Note that data network AAA authenticates the user, as opposed to physical layer authentication, which only authenticates the mobile. Therefore, users wishing to gain access to public or private data networks are presented with a login and password sequence, familiar to wireline remote data access users, in addition to mobile device authentication taking place at registration stage, which results in the momentary hesitation at phone startup familiar to most mobile phone users.
The CDMA2000 data subsystem provides two user authentication mechanisms when simple IP or Mobile IP access methods are requested, as defined in [IS835] and [RFC3141]. As mentioned, for the Simple IP access mode, authentication is based on CHAP, which is a part of PPP negotiation. In CHAP, the PDSN challenges the mobile station with a random value to which it must respond with a signature based on MD-5 digest of the challenge, a username, and a password. The PDSN passes the challenge/response pair to the home AAA server for user authentication.
For Mobile IP, the PDSN sends a similar challenge within the agent advertisement message to the mobile station. Again, the MS must respond to the challenge with a signature and NAI that is verified by the home network, but this time the response is sent along with the Mobile IP registration request rather than during PPP session establishment. Both of these mechanisms rely on shared secrets associated with the NAI, which are stored in the home network, and both will be supported by the same AAA infrastructure. In both cases the accounting data is collected in the PDSN and transferred to the AAA server. The PDSN collects data usage statistics for each user, combines these with the radio access accounting records sent by the PCF, and forwards them to the local AAA server. Note that accounting information is collected by both the PCF and the PDSN. For roaming users, the AAA server may be configured to forward a copy of all RADIUS accounting records to the home AAA server in addition to keeping a copy at the visited AAA server.
When a handoff between two PDSNs occurs, an Accounting Stop message is sent to the AAA server from the releasing PDSN, and an Accounting Start is sent to the AAA server from the connecting PDSN (more details on this and other accounting mechanisms is provided in Appendix B). The Accounting Stop from the releasing PDSN may arrive some time after the Accounting Start from the new PDSN (the releasing PDSN may not be aware that mobile has moved, but it must wait for a Registration Lifetime or PPP Inactivity timeout to end the session). This means the billing server must accept multiple stop/start sequences from different PDSNs that contain overlap and treat these as a single session, per [IS 835]. When a PPP inactivity timer or an MIP lifetime expires, or the mobile terminates the session, the R-P link is released and an Accounting Stop is sent to the AAA server.
Having so far discussed the details of packet data services in CDMA2000, in the next section, we give the same level of exposure to the UMTS and GPRS systems.
[3]There is also a parallel set of burst rates defined in the CDMA2000 radio link standards based on multiples of 14.4 instead of 9.6. However, these rates are not normally implemented because the resulting top rate of 230.4 Kbps does not have good spatial coverage for most real-world RF environments, making this rate-set family less attractive in commercial networks than the 9.6 family.
[4]Note that CHAP support is not required for Mobile IP and is optional for Simple IP access methods.
[5]Note that dormancy only affects the radio link level connection, while all dormant mobiles retain a PPP link with the PDSN which is not aware of the active/dormant state of the MN.
[6]More details on NAI and other Mobile IP extensions are provided in Appendix A.