PACKET CORE FUNCTION AND METHOD OF AUTOMATIC PDSN DISCOVERY, MONITORING, AND FAILURE HANDOVER
BACKGROUND OF THE INVENTION
Technical Field of the Invention
This invention relates to telecommunication systems and, more particularly, to a Packet Core Function (PCF) in a packet-switched network and a method of automatic Packet Data Service Node (PDSN) Discovery, Monitoring, and Failure Handover.
Description of Related Art
Code Division Multiple Access (CDMA) 2000 is the third generation system for CDMA networks. It offers packet-switched call routing via Packet Data Service Nodes (PDSNs) which are used as access concentrators and, in the case of Mobile IP, as foreign agents for roaming mobile subscribers.
FIG. 1 is a simplified block diagram of an existing packet data network 10 providing radio access to a plurality of mobile subscribers utilizing Mobile Nodes (MNs) 11-13. In packet data networks such as CDMA 2000, mobile subscribers gain access to the packet core network through Radio Networks (RNs) 14 and 15 over an air interface link 16. An illustration of the protocols utilized, and where each protocol ends, is shown below the block diagram. An Internet Protocol (IP) network 17 links the RNs to a plurality of PDSNs 18-20 using a Radio-Packet (R-P) interface 21. Each RN may include a plurality of Base Stations (BSs) and Base Station Controllers (BSCs) (not shown). Each BSC, in turn, includes a Packet Core Function (PCF) 22 and 23 which handles the switching of data packets that come from the MNs. The
PCF selects one or more PDSNs to terminate its packet data sessions. When a mobile subscriber desires to initiate a packet data session, the PCF selects a PDSN and establishes an R-P connection 21. This enables a Point-to-Point Protocol (PPP) connection 24 to be set up between the subscriber's MN and the selected PDSN. The PDSNs, in turn, utilize one or more sub-networks 25 and 26, and routers
27 and 28 to access the packet-switched network backbone 29. At that point, the MN
is connected to the IP network 17, and the IP protocol 30 may be utilized to access the Internet.
In order for the PCF to discover (select) a PDSN, the PCF must have a database populated with information regarding the PDSNs that it can utilize. Currently, the PCF is manually populated with the PDSN data by the network operator. Thus, when a new PCF is added to the radio network, the operator must provide an initial database of PDSN data. Likewise, if a PDSN is added or removed from the network, the operator must manually update the PDSN data in the PCF to reflect this change. It would be advantageous to have a method of automatically populating a new PCF with PDSN data, and automatically updating a PCF when the
PDSN data changes.
Additionally, if a PDSN fails, there is currently no failover procedure that can save the active data sessions that were being conducted through that PDSN at the time of failure. If the selected PDSN fails, all of the packet data sessions that are active with that PDSN are lost, and any data being transferred is lost. The MN loses the data stream immediately, but the RN may not realize that a failure has occurred for a considerable amount of time. This is because the status of the PDSN is only reported to the PCF when the PCF registers with the PDSN on behalf of the MN (i.e., the R-P session is established). This status is updated in the PCF whenever the PCF re- registers with the PDSN (i.e., the R-P session is maintained). Thus, the PCF will not detect a PDSN failure until the next periodic PCF registration, and this could be as long as 30 minutes. Thus, from the mobile subscriber's perspective, data transfer simply stops, and the network takes no corrective action. The only existing solution to this problem is for the mobile subscriber to cancel the packet data session, and start a new session and have the PCF select a different PDSN.
Additionally, it should be noted that when the PCF selects a new PDSN after a PDSN failure, it would be advantageous for the PCF to select a new PDSN that is connected to the same sub-network that the failed PDSN was accessing. This would help to balance the traffic load on the routers and to minimize the traffic variations on the different sub-networks when the handover to the new PDSN is performed.
However, current procedures do not provide this routing information to the PCF.
In order to overcome the disadvantages of the existing solution, it would be advantageous to have a PCF and method for automatic PDSN discovery, monitoring, and failure handover. The present invention provides such a PCF and method.
SUMMARY OF THE INVENTION
The present invention is a method of automatically populating a PDSN database in the PCF with PDSN status and routing information, automatically updating the PDSN database when the status of a PDSN changes, and handling the failure of a PDSN such that active packet data sessions with mobile subscribers are maintained. Thus, in one aspect the present invention is a method of automatically populating a PCF with status information for a plurality of PDSNs in a packet- switched network. The plurality of PDSNs automatically send PDSN status information to the PCF, and the PCF stores the PDSN status information in a database. The PDSNs may periodically multicast the PDSN status information to all PCFs in the network, or may automatically unicast the PDSN status information to PCFs that request it. The method also automatically updates the database when a new PDSN is added to the network. When the new PDSN is installed, it automatically sends its PDSN status information to the PCF, and the PCF stores the new PDSN status information in the database. Likewise, the method automatically updates the database when an existing PDSN is removed from the network. When the PCF determines that a predefined number of status messages are not received from the existing PDSN within a predefined time period, the PCF automatically removes the existing PDSN from the database.
In another aspect, the present invention is a method of handling a failure of a Packet Data Service Node (PDSN) that is providing a mobile subscriber with an active packet data session in a packet-switched network that includes a plurality of PDSNs. The method comprises the steps of detecting the PDSN failure, selecting a new PDSN to handle the packet data session, and handing over the active session to the new PDSN so that the active packet data session with the mobile subscriber is maintained. The step of detecting the PDSN failure may be performed by having each PDSN periodically multicast its status to the PCFs in the network, or by periodically
transmitting polling messages from each PCF to the PDSN, and receiving in the PCF a response to each polling message. A response is transmitted by the PDSN when the PDSN is operating normally. If a configurable, predefined number of responses to the polling message are not received in the PCF within a predefined time period, the PCF determines that the PDSN has failed. The PCF may select as the new PDSN, the
PDSN with the shortest round-trip time value for a message sent from the PCF to the PDSN, the PDSN with the fewest hops between the PCF and the PDSN, or the PDSN with the lowest level of congestion. Alternatively, the PCF may select as the new PDSN, a PDSN that has a routing table that is the most similar to the routing table of the failed PDSN. If multiple handovers are to be made, the PCF staggers the handovers so that the new PDSN is not overloaded.
In yet another aspect, the present invention is a PCF in a packet-switched network for switching of data packets that come from Mobile Nodes (MNs) and for controlling handover of active data sessions when a PDSN in the packet-switched network fails. The PCF includes a PDSN database that is automatically populated with PDSN data, and is automatically updated as new PDSN data is received from PDSNs in the network. The PCF also includes a PDSN status monitor that detects when the PDSN fails; a PDSN selector connected to the PDSN status monitor that selects at least one new PDSN to handle the active data sessions that were being handled by the failed PDSN, upon being notified by the PDSN status monitor that the
PDSN has failed; and a PDSN handover controller connected to the PDSN selector that controls handover of the active data sessions to the PDSNs that are selected by the PDSN selector.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention will be better understood and its numerous objects and advantages will become more apparent to those skilled in the art by reference to the following drawings, in conjunction with the accompanying specification, in which:
FIG. 1 (Prior Art) is a simplified block diagram of an existing packet data network providing radio access to a plurality of mobile stations;
FIG. 2 is a flow chart illustrating the steps of the present invention when
automatically populating and maintaining a PDSN database in the PCF;
FIG. 3 is a flow chart illustrating the steps of the PDSN fail-over method of the present invention; and
FIG. 4 is a simplified block diagram of one embodiment of a Packet Core Function (PCF) which has been modified to perform the method of the present invention.
DETAILED DESCRIPTION OF EMBODIMENTS
The present invention is a PCF and a method of automatically populating a PDSN database in the PCF with PDSN status and routing information, automatically updating the PDSN database when the status of a PDSN changes, and handling the failure of a PDSN in such a way that active packet data sessions with mobile subscribers are maintained.
FIG. 2 is a flow chart illustrating the steps of the present invention when automatically populating the PDSN database in the PCF, and updating the PDSN database when the status of a PDSN changes. Utilizing a reserved port at step 31 , each PDSN may periodically multicast, multicast on demand, or unicast on demand from the radio network, the following profile information: (A) The PDSN's IP address. (B) The PDSN's routing table indicating all the hosts that are accessible by the PDSN on the backbone side, and the number of IP hops. This indicates to the radio network in general, and the PCF in particular, which paths are accessible from each PDSN.
(C) The PDSN's operator ID indicating the operator to which the PDSN belongs, and who may connect to the PDSN.
(D) The PDSN's capacity in terms of the number of users that may be simultaneously connected, and the available bandwidth.
(E) The PDSN's average load in terms of the number of users currently connected and the percentage of available bandwidth being utilized. These types of multicast messages are transferred at the lower (control plane) levels of the IP protocol stack, and do not affect the transfer of data at the higher (user
plane) levels.
At step 32, it is shown that each PDSN also listens for multicasted or unicasted requests for PDSN data from PCFs in the radio network. At step 33, the data may be multicasted or unicasted to the requesting PCF upon receiving a request. At step 34, a new PCF is added to the network. At 35, as each PCF in the radio network listens for the multicasted PDSN data, the new PCF automatically builds its PDSN database from the received PDSN data.
The PCFs also update their PDSN databases as the network topology changes. Therefore, if a new PDSN is added to the network at 36, and starts to multicast its profile data at 37, the PCF automatically adds the new PDSN to its database at 38. If a PDSN is removed from the network at 39, and a predefined number of periodic or response messages are not received from the removed PDSN, the PCF concludes at 40 that the PDSN has failed. Therefore, at step 41, the PCF removes the PDSN from its database. Thus, when the network topology changes due to the addition or removal of a PDSN, no manual reconfiguration of the PCFs is required.
Each PDSN may be configured independently, and the information about all PDSNs is automatically propagated throughout the network. Additionally, for load balancing or revenue considerations, network operators may desire that a PCF select PDSNs from a particular portion of the network desired by the operator. The operator can isolate certain portions of their CDMA 2000 data networks by not allowing the multicasts to exit those portions of the IP network or enter the portion of the network accessed by the PCF. As a result, the PCF will delete the PDSNs in that portion of the network from its database. This will ensure that the PCF selects PDSNs only from the portion of the network desired by the operator. The PCF of the present invention performs a method of handling the failure of a PDSN in such a way that active packet data sessions with mobile subscribers are maintained, while not adversely affecting the signaling load in the network or overloading any PDSNs. The fail-over method performs a staggered handoff of the active packet data sessions to another PDSN upon detecting a PDSN failure. In this manner, data sessions are not lost, the new PDSN is not overloaded, and the mobile subscriber does not have to restart the packet data session.
The mobile subscriber in the present invention must be a "Mobile IP" subscriber. The CDMA 2000 standard describes two possible subscription types for mobile subscribers: "Simple IP" and "Mobile IP". The Simple IP subscription type does not allow a data session to be saved by handing over the data session to another PDSN if the original PDSN fails. This is because in Simple IP, the PDSN handling a particular data session for a mobile subscriber provides an IP address to the subscriber's MN. If the PDSN then fails, the MN cannot be handed over to another PDSN because the new PDSN would assign a new IP address to the MN.
In Mobile IP, on the other hand, each MN is pre-allocated an IP address that it always uses regardless of the serving PDSN. A Mobile IP protocol layer on top of the IP layer permits a home agent and a foreign agent to tunnel data towards roaming Mobile IP subscribers. Therefore, during a handover, the new PDSN uses the same pre-allocated IP address as was being used by the failed PDSN.
The failure handover method essentially comprises three steps: (1) detecting the PDSN failure; (2) selecting a new PDSN; and (3) handing over the active session to the new PDSN. Various alternatives and options for each step are described below.
FIG. 3 is a flow chart illustrating the steps of the PDSN fail-over method of the present invention. After an active packet data session is established at step 45, the
PCF monitors the status of the PDSN at step 46. This may be performed in several ways, three of which are illustrated in FIG. 3. hi a first alternative, the PCF may regularly "poll" the PDSN at step 47a to obtain its current status. The PCF may poll every known PDSN or a subset of the known PDSNs. This may be accomplished by sending an Internet Control Message Protocol (ICMP) "echo" request to each PDSN as either a multicast message or a plurality of unicast messages. If the PCF fails to receive a predefined number of responses to the polling message from a particular
PDSN within a given time period, the PCF concludes that the non-responsive PDSN has failed. A configurable number of failed responses is used rather than a single failed response since a response packet may occasionally be lost, and the PDSN should not be treated as failed. The failure needs to be detected as quickly as possible, however, so the polling requests maybe sent approximately every 1-10 seconds. This range is for illustrative
purposes only, and the frequency of the polling may be faster or slower than this range. The polling requests and responses are sent between the PCF and the PDSN where bandwidth limitations are not as severe as they are over the air interface. Thus, instead of the prior art method of relying on the mobile station to detect the session failure over the air interface, in the present invention, the PCF detects the PDSN failure over the R-P interface.
In a second alternative, the PDSNs may multicast status messages at step 47b throughout the network. Multicasts are preferred over broadcasts because broadcasts may be limited to a particular sub-network while multicasts are not. The PCF and the PDSN may be separated by several sub-networks, and broadcasts may not be transferred. The status messages may include the PDSN status, level of congestion, and routing table information such as the sub-network to which the PDSN is connected. The number of IP hops may also be reported to each receiving PCF. The status messages may be sent periodically or upon demand from a particular PCF. The periodic status messages may be multicast while the messages sent upon demand may be unicast to the requesting PCF. The PCF may use a defined protocol to request the more detailed information about the status of the PDSN.
In an third alternative shown in step 47c, the PCF may obtain PDSN status information by periodically re-registering with the PDSN over the R-P interface. Since this interface uses IP transport, the extra signaling is trivial, even if registrations are performed as often as the 1-10 seconds utilized for the polling in the first embodiment.
Whichever method is utilized to monitor the PDSN status, the PCF may detect a PDSN failure at step 48. After the failure of the PDSN is detected, the PCF selects one or more new PDSNs at step 49. Since the PCF polls every known PDSN, it can maintain a list of the current status of each PDSN. Therefore, the PCF can easily identify and select a PDSN that is operational and suitable. For example, referring to the network of FIG. 1 , if PDSN- A 18 fails, the PCF may select PDSN-B 19 rather than PDSN-C 20 since PDSN-A and PDSN-B are both connected to the same sub-network 25. Several optional methods of selecting a PDSN are shown in steps 50a-50d. At step 50a, an optional method of selecting a PDSN may measure a round-trip time value
to determine the PDSN with the best response time among the list of known operational PDSNs. Step 50b, the PCF may select a PDSN at step 51 that is more lightly loaded than others, or may follow a round-robin scheme so that the traffic load in the network can be balanced between the PDSNs. This option maybe utilized with the alternative embodiment described above in which the PCF obtains information about the congestion level of each PDSN. hi step 50c, the PCF selects the PDSN with the fewest hops between the PCF and the PDSN. Finally, in step 50d, the PCF may select as the new PDSN, a PDSN that has a routing table that is the most similar to the routing table of the failed PDSN. Alternatively, since the failed PDSN may have been handling a large number of active data sessions, the PCF may select a plurality of
PDSNs in order to distribute the load from the failed PDSN more evenly in the network.
After selecting a PDSN, the PCF hands over of the active session(s) to the new PDSN(s) at step 51. During the handover, the PCF initiates a new R-P connection towards the new PDSN. This initiates a new PPP session over which the MN performs Mobile IP registration with the new PDSN. A PDSN handover process for a single MN that roams into an area controlled by a different PDSN is defined in the CDMA 2000 standard, and may be utilized for individual handovers in the present invention. It should be recognized, however, that in the present invention, the context is already established in the PCF and is not lost.
As noted above, a PDSN may be handling a large number of active packet data sessions when it fails. Thus, in the present invention, it is recognized that the PCF should avoid handing over all of the active sessions at the same time since this could overload the core network and the selected PDSN. In the preferred embodiment, therefore, the handovers are initiated in a staggered manner only when an individual
MN requests a transfer of data. Since all MNs involved in data sessions are not downloading data continuously, (i.e., some connections are idle) the staggered handover process distributes the handovers over a reasonable amount of time and limits the network load resulting from the PPP session re-negotiations. Alternatively, a timer located, for example, in the PCF or BSC may keep track of how long each data session has been idle. If the PDSN fails, the data sessions are
sequentially handed over to the new PDSN beginning with the data session that has been active most recently, and ending with the data session that has been idle the longest.
FIG.4 is a simplified block diagram of one embodiment of a PCF 55 which has been modified to perform the method of the present invention. A PDSN status monitor 56 monitors the status of the PDSNs 57 known to the PCF. In the embodiment illustrated, the PDSN status monitor controls a PDSN status message trigger/requester 58. In the embodiment in which polling messages are sent to the PDSNs, the trigger/requester may send the polling messages. The PDSNs that are operational may periodically multicast their status information to PCFs in the network, or may send polling response messages to the PCF. The status messages are received in a PDSN status message receiver 59. The status information, which may include congestion level information and routing information, is sent to the PDSN status monitor which stores it in a PDSN database 61. The stored information may be time stamped so that it may be updated when its predefined lifetime expires. A multifunction timer 62 in the PDSN status monitor may determine when the lifetime has expired, thus triggering another update. In the embodiment using polling messages, the timer may also determine when a predefined response time period has lapsed for each polled PDSN. If a particular PDSN fails to respond a configurable, predefined number of times within the response time period, the status monitor determines that the PDSN has failed, and instructs the PDSN selector 63 to select one or more PDSNs for handover of the active data sessions that were being handled by the failed PDSN.
When the PDSN selection is made in accordance with the previously described method, a PDSN handover controller 64 performs staggered handovers to the selected
PDSN(s) utilizing a routing function 65. The handovers may be performed whenever data requests are received from MNs 66 that are involved in the active data sessions. Alternatively, the multifunction timer 62 may provide the handover controller with information regarding how long each data session has been idle. In this case, the data sessions may be sequentially handed over to the new PDSN beginning with the data session that has been active most recently, and ending with the data session that has
been idle the longest.
The present invention thus enables the CDMA 2000 network to be dynamically optimized as the topology of the network or configuration of individual nodes changes.
The network infrastructure also becomes more robust. If a link failure causes the PCF to lose access to some of the PDSNs, the situation is automatically detected, and appropriate alternate PDSNs are immediately identifiable.
It is thus believed that the operation and construction of the present invention will be apparent from the foregoing description. While the PCF and method shown and described has been characterized as being preferred, it will be readily apparent that various changes and modifications could be made therein without departing from the scope of the invention as defined in the following claims.