The astute reader may object to the title of this section. It seems inappropriate to fashion a title that contains circuit switch concepts with a packet data network! Furthermore, what is this topic doing in a section of network access? This strange marriage of the two technologies is discussed in this section.
In 1995, a few members of the CDPD Forum saw an opportunity to add to the CDPD System Specification through the definition of a new complementary service. The thought was that if CDPD services could be available through the existing cellular voice telephone connections, the requirement for nationwide coverage would be instantly realized.
With the goal of developing a complementary standard to allow mobile devices to access the CDPD network through cellular voice telephone circuit switched connections, the group examined the CDPD system architecture. It became quickly obvious that the layered communications architecture has provided a great flexibility to accomplish this5.29 .
The development team examined the CDPD system architecture on a layer by layer basis. The resultant architecture shown in Figure 5.31 is based on the following considerations.
The physical layer was first examined. There really wasn't much to decide here. Since the intent is to make use of the cellular telephone voice channel, the GMSK modulation scheme cannot be used without changes. Furthermore, since there are already cellular modem devices available, with mass manufactured chip sets, it makes much more sense to rely on that technology. All through the design of CDPD, the philosophy has been to define new technology only when necessary. The development team wisely chose to use the developments of the cellular telephone modem industry.
The data link layer in CDPD is divided into two sub-layers, the Medium Access Control sub-layer and the Logical Link Control sub-layer. In the Circuit Switch CDPD system, the use of individual circuits for each mobile means that there is no sharing of the RF channel in use. As such, there is no need for a Medium Access Control function5.30 .
The Logical Link Control function is responsible for establishment and management of a point-to-point connection between the CS CDPD mobile device and the CS CDPD network. The basic requirements of this layer include:
¥ Reliable delivery of data frames
¥ Sequenced delivery of data frames
¥ Link establishment and disconnection
Much of these requirements are already satisfied by the typical modern day modem equipment. Current modem technologies typically include end to end protocols and procedures to provide error detection and correction, call establishment and disconnection. However, the development team identified additional control parameters that are necessary for efficient operation within the CDPD network. These additional functions are used to ensure transparent operation to the mobile user and efficient use of the RF channel. The requirements were extensive enough to require the establishment of the Circuit Switch CDPD Control Protocol (CSCCP5.31 ) to be used on top of widely available reliable modem protocols. The CSCCP is discussed in greater detail in the next section.
Above the data link layer, there is the Network Layer. The lower sub-layer of the Network Layer is the Subnetwork Dependent Convergence Function. The SNDCF is specifically defined to address mismatch in service requirements of the Network Layer and service characteristics of the Data Link Layer. In the early stages of CS CDPD system design, there were considerations to alter the CDPD SNDCP. However, as the system design progressed, it became obvious that the SNDCP need not be modified. Some minor adjustments in terms of maximum frame size may have provided some efficiency gains, but general concensus was reached that such gains were small and may make implementation of dual mode devices more complex. SNDCP is not changed.
Since the SNDCP is not changed, protocols at the Network Layer and above are also unchanged. This ensures transparency of application operation between CDPD and CS CDPD5.32 .
The purpose of the Circuit Switch CDPD Control Protocol is to provide the services necessary to maintain efficient CDPD mobility management function on circuit switched data modem technology. The goals for this protocol are:
¥ Use of circuit switched connection
¥ Efficient use of circuit switched technology
¥ Continual connection to network
¥ Efficient circuit switched backbone connections
¥ Robust connection
To address these design goals, the specification team developed the following CS CDPD messages:
¥ Connection Request
¥ Connection Response
¥ Reconnection Request
¥ Reconnection Response
¥ SNDCP Data Packet
¥ SNDCP Unitdata Packet
¥ Link Reset
¥ Link Reset Acknowledge
The use of these message to achieve the design goals may best be illustrated through examples of connection events. The events presented include the following:
¥ Initial connection (by the mobile)
¥ CS CDPD M-ES initiated reconnection
¥ CS CDPD MD-IS initiated reconnection
¥ Redirection
¥ Redirection with override
¥ Link Reset
The CS CDPD connection begins with the connection request by the mobile device. Within the CS CDPD specifications, the mobile device is called the CS CDPD Mobile End System or CM-ES. Just as in CDPD, the CM-ES must initiate the connection.
The CM-ES starts by selecting a dial code from the list programmed into the device by the service provider. Using an appropriate dial code, the CM-ES establishes a circuit switched data connection. The peer end point of this circuit is the CS CDPD MD-IS or CMD-IS. The CM-ES then sends a Connection Request message carrying the following parameters:
¥ CM-ES Equipment Identifier
¥ V.42bis data compression parameters
¥ Duration time
¥ Cellular System Identifier (AMPS System ID)
¥ Dial code
The CMD-IS responds with a Connection Response message containing the following parameters:
¥ CMD-IS Identifier
¥ Service Provider Network Identifier (SPNI)
¥ Wide Area Service Identifier (WASI)
¥ V.42bis compression parameter response
¥ Result code
This exchange of messages (see Figure 5.32) allows the CM-ES and the CMD-IS to identify themselves to each other and establish compression negotiation parameters. In addition, the CM-ES informs the CMD-IS of the dial code to use in order to contact the CM-ES.
Once a successful connection has been established, the CMD-IS initiates the exchange of encryption keys. From this point forward the communications process proceeds as in staandard CDPD.
This interchange of messages achieves the first goal of establishing a circuit switched connection.
After the connection has been established, data transfer between the CM-ES and the network proceeds. For most connections, there are periods of inactivity. On a circuit switched connection, these periods are wasteful since the link cannot be shared by other devices. To account for this data traffic characteristic, the CM-ES disconnects after a predetermined idle period and suspends the data link connection.
When the CM-ES has data to send after having disconnected, it must initiate reconnection procedures. This is accomplished by the CM-ES selecting an appropriate dial conde and establishing a circuit connection. However, unlike the initial connection, the CM-ES sends a Reconnection Request message (see Figure 5.33)which contains only the CM-ES Equipment Identifier (EID). This EID allows the CMD-IS to quickly ascertain this to be a reconnection by a previously connected device. There is therefore no need to repeat the exchange of data compression parameters and AMPS system ID. The CMD-IS responds with a Reconnection Response carrying the CMD-IS ID. This allows the CM-ES to quickly confirm that it has reconnected to the same CMD-IS.
Once the reconnection message exchange has been successful, the two peer entities resume the suspended data link connection.
The only other deviation from the CDPD system is the use of an End System Query message to force the exchange of registration data and authentication credentials. This is an added precaution to avoid fraudulent access.
This procedure achieves the second goal of efficient use of the circuit switch technology. There is no need to keep the circuit switched connection active when there is no data to transfer.
After the data link is disconnected due to an extended idle period, it is possible that the network needs to deliver data to the CM-ES. In the usual circuit switch connection scenario, this is not possible. The mobile user must periodically "check-in" for data. However, the CS CDPD designers wanted to offer connection service similar to CDPD. In which case, the network must be able to initiate reconnection to the CM-ES.
One of the optional parameters provided by the CM-ES during intial connection is a dial code. This dial code is to be used when the network wishes to initiate reconnection. Therefore, when the CMD-IS has data to send to the CM-ES, it establishes a circuit switch connection using the earlier supplied dial code. Once connected, the CMD-IS sends a Reconnection Request message (see Figure 5.34)containing the CMD-IS ID. If the CM-ES finds the CMD-IS ID acceptable, it responds with a Reconnection Response containing the CM-ES ID. Once again, the two peer entities resume the suspended data link connection. The End System Query is sent by the CMD-IS to cause registration and authentication.
The procedure achieves the third goal of allowing the CM-ES to be logically continually connected to the network without the need to maintain the circuit switch link.
Due to the mobile nature inherent in CDPD devices, it is possible that a CM-ES relocates to an area where the dial code normally used is not the optimal path through the infrastructure. This can result if the service provider has a local point of presence through a local modem bank (see Figure 5.35), or that the service provider has an alternate CMD-IS at the local system (see Figure 5.36).
In these cases, it may be more efficient for the mobile to use the a different set of dial codes to access the network. This is provided for in the CSCCP through the Redirect result code.
The procedure occurs on the initial connection request. The CM-ES proceeds with an initial connection request but the CMD-IS responds with a Connection Response message containing the optional parameter to indicate a Redirect directive. Along with that directive, a list of alternate dial codes is provided.
The CM-ES, barring other problems, disconnects from the CMD-IS and attempts to re-initiate connection requests with one of the new dial codes. If for some reason the new dial codes are not operational, the CM-ES may retry the connection with the original dial code and issue a Redirect Override indicator. If the CMD-IS cannot accept any connection requests, it may issue a Forced Redirection command.
This procedure allows the service provider to instruct the mobile device to access the network at the most efficient point of presence.
Even though the CS CDPD system has been built on using reliable data transfer mechanisms available from current modem technologies, errors may rise from various internal connection points. To address these errors, the CS CDPD specification included an error recovery mechanism.
The mechanism is achieved in two steps. First, the data transfered is contained in the CSCCP SNDCP Data Packets. These messages contain both a simple checksum and a sequence number. The receiver of each message verifies the checksum. If a checksum failure is detected, the link is reset by the receiver issuing a Link Reset message (see Figure 5.37). This Link Reset message contains the sequence number of the failed packet. The peer entity then responds with a Link Reset Acknowledge packet containing its next expected sequence number. Once the Link Reset message and the Link Reset Acknowledge messages are exchanged, the two entities reset their sequence numbers to 0 and restart the exchange of SNDCP Data packets from the point of the failure.
This procedure corrects the small residual error probability of the link.