US 20050025116 A1
An method for PDSN (Packet Data Serving Node) initial assignment and re-selection in a wireless communication system is provided. The method is implemented in PCF (Packet Control Function) within a third-generation (3G) CDMA Radio Access Network (RAN). The methodology reduces the number of point-to-point (PPP) re-establishments, when a Mobile Station (MS) roams to a different packet zone/PCF. The method generates a table of PDSN identification numbers cross-referenced to the PDSN Internet protocol (IP) addresses, residing with each PCF. A PDSN Id number is selected from the table using the MS IMSI (International Mobile Station Identifier) as a key to perform the selection. The present invention methodology addresses the issue of forward and backward compatibility, scalability, reliability, and load sharing in the PDSN selection.
28. In a wireless communication system including one Packet Control Function (PCF) and at least one Packet Data Serving Node (PDSN), a method for determining which PDSN should be used for packet data communications with a mobile station (MS), the method comprising:
maintaining a configuration table at the PCF being communicatively coupled between the MS and at least one PDSN, said configuration table having at least one unique PDSN identification (Id) number listed therein, said at least one unique PDSN Id number being respectively associated with the at least one PDSN; and
selecting a PDSN in response to a number of the PDSN Id numbers in the configuration table.
29. The method of
assigning a unique MS Id number to the MS, wherein selecting the PDSN further includes selecting in response to the MS Id number.
30. The method of
31. The method of
32. The method of
33. The method of
34. The method of
35. The method of
36. The method of
37. The method of
38. The method of
39. The method of
40. The method of
41. The method of
42. The method of
43. The method of
44. The method of
45. A Packet Control Function (PCF) for selecting which one of n Packet Data Serving Nodes (PDSNs) should be used for packet data communications with a mobile station (MS), said PCF being communicatively coupled between the MS and at least one PDSN and being configured to:
maintain a configuration table, said configuration table having at least one unique PDSN identification (Id) number listed therein, said at least one unique PDSN Id number being respectively associated with the at least one PDSN; and
select a PDSN in response to a number of the PDSN Id numbers in the configuration table.
46. The PCF of
47. The PCF of
This invention generally concerns wireless communications and, more particularly, a system and method for assigning a Packet Data Serving Node (PDSN) to a wireless mobile station (MS), to provide packet data services in a third-generation (3G) code division multiple access (CDMA) communication system.
In the second generation CDMA systems specified by the IS-95B standard, uplink data bursts from a MS to base station are limited to a small number of frames, typically 8 or less. The data rate is limited to 4.8 kbps, i.e., inferior to the one used by the users active on the dedicated (voice) channels, which is 9.6 or 14.4 kilobits per second (kbps). The 3G CDMA, as described in IS-2000 body of standards, is being designed to accommodate both longer messages (more frames per message) and higher data rates to support the communication packetized data. Data rates of 384 kbps are envisioned. The ability of the 3G CDMA to communicate packet data will permit users to seamlessly access the Internet with a computer through a wireless telephone.
The transfer of large volumes of data between the MS and the PDSN portal to the Internet will be aided by changing the 3G CDMA system from the traditional PSTN circuit-switched system, where the data network communication must be routed through the mobile switching center (MSC), to a system where the PDSN can locally connect to a radio access network (RAN) base station controller (BSC) using frame relay, or some other link-layer transport capable of carrying Internet protocol (IP) data packets. The linkage between the wireless system and the Internet can be more peer-to-peer and, therefore, simpler and faster, if the packet data protocol can be maintained across the PDSN.
When a packet data call is first established, by an MS request for example, a BS and associated BSC are selected as elements in the link. Typically, the BSC is associated with a particular PCF, which may even be co-located at the BS. The PCF must be connected to a PDSN, or a PDSN must be (initially) selected for the MS. There is currently no IS-2000 standard for the initial PDSN assignment.
Furthermore, the packet data link between the MS and PDSN, once initially selected, may become dormant as the MS ceases to communicate packet data. In this situation the air interface traffic and control channels are released, but the point-to-point (PPP) connection between the PDSN and the MS are maintained. It is foreseeable that a dormant MS will roam. Regardless of the relationship between PCFs and BCSs, a dormant MS will roam to a different packet zone (i.e. a new PCF). The new PCF must be notified and it must then establish a new connection to one of the PDSNs. It is highly desirable that the new PCF selects the PDSN that was previously selected by the old PCF. In this manner, the overhead required to re-establish a PPP connection can be minimized, even as dormant handoff occurs between the old and new PCFs, where the previous PDSN already has the PPP connection.
The current 3G CDMA network architecture does not have an interface or a mechanism defined to inform the new PCF of the PDSN which was previously selected by the old PCF, nor is there a third entity that overlooks all resources in the entire network.
In would be advantageous if there was a way for the same PDSN to be selected for packet data communications to a MS in the dormant state, in the event that a new PCF must be selected.
It would be advantageous if the same PDSN could be selected in a MS dormant state reselection process using the existing network architecture.
It would be advantageous if the same PDSN could be selected in a MS dormant state reselection process without the PCF and PDSN units having special knowledge of each other, without special network communications, or without the use of an independent network entity to control dormant state reselections.
A method is provided for selecting a PDSN for packet data communications with a MS in a wireless communication system. The method comprises: generating a record of unique PDSN (identification) numbers; and selecting a PDSN identification (Id) number in response to the number of PDSN Id numbers in the record. The selection of the PDSN is also responsive to the MS International Mobile Station Identifier (IMSI).
In one aspect of the invention, the wireless communication system includes a plurality of m PCFs and a plurality of n PDSNs with unique IP addresses. Then, the method comprising: at each PCF, generating a record of PDSN Id numbers; and selecting the PDSN Id number in response to the total number (n) of PDSN Id numbers in the record, and in response to the MS IMSI.
The record of PDSN Id numbers is a table with an ordered sequence of the PDSN Id numbers cross-referenced to the PDSN IP addresses. The selection of the first PDSN Id number includes dividing the first MS IMSI by the number of (n) PDSNs in the table. Specifically, the selection of the first PDSN Id number includes selecting the PDSN with the number equal to the remainder in the quotient, when the first MS IMSI divided by n.
Not all the PDSNs listed in the selection table need be connected to the first PCF hosting the table. When the first PCF is not connected to the IP address associated with the first PDSN Id number, the connection cannot be made. Then, the method further comprises reselecting an alternate PDSN Id number, after the first PDSN Id number has been calculated. In one aspect of the invention, the reselection of an alternate PDSN Id number includes varying the first PDSN Id number, typically by adding “1”, and dividing that varied number by n. Then, the PDSN with the number equal to the remainder in the quotient is selected.
Alternately, the reselection of an alternate PDSN Id number includes: removing the first PDSN Id number and associated IP address from the table; dividing the MS IMSI by the number of PDSNs remaining in the table; and selecting the PDSN Id number equal to the remainder in the quotient.
A wireless communication system for communicating packet data is also provided. The system comprises a first mobile station (MS) having a unique identification number (Id), such as its IMSI, and a transceiver for wireless packet data communications. The system comprises a plurality of m PCFs. Each PCF has a port for packet data communications with MSs, and with at least one of a plurality of n PDSNs. Each PDSN has a port for packet data communications, a unique IP address, and a unique (identification) number. Each PCF includes a table with an ordered sequence of the n PDSN Id numbers cross-referenced with the n IP addresses. Because of the common ordering of PDSN Id numbers in the tables, each PCF selects the same PDSN Id number for packet data communications involving the first MS. The selection process is responsive to the first MS Id, and the number of PDSNs in the table.
In short, the heart of the invention is a methodology embodied in a set of machine executable instructions. Typically, the instructions are written in a medium which can be accessed as memory by a computer, or which can be read by a computer. The set of instructions are typically embodied as a computer application which orders the computer microprocessor to perform the series of steps needed to cause a PCF to select a PDSN in accordance with the algorithm described in detail below.
A PDSN table must be configured at PCF. Each PCF maintains a table that includes all PDSNs it is connected to. Each table entry must contain (but is not limited to) a PDSN identification (Id) number, and an IP Address associated with the PDSN. It is important that the PDSN Id numbers are assigned consistently, stored in a predetermined ordered sequence (ascending or descending) in all PCFs. An example of such a table, such as might be configured in a PCF is shown below (Table 1).
The PDSNs are (locally) numbered from 0 to (n−1) and are stored in ascending order based on PDSN IP addresses. It is assumed that aaaa<bbbb< . . . <xxxx, and that n is the total number of PDSNs available to this PCF. As explained in more detail below, a PDSN can be entered into a selection table, even if that PDSN is not connected to the PCF hosting the table.
The PCF determines which PDSN to select for a particular MS based on the unique identification number (Id) of the MS. Typically, the IMSI is used. Then, the following calculation is performed:
It is understood that modulo n is a computational process where the remainder of a quotient is used as the answer. In this case the quotient is the MS IMSI divided by n. Note that n is the total number of locally available PDSNs to this PCF, not that of the entire network. However, when the network is fully connected, as explained below, the “local” n becomes the total number of PDSNs in the entire network.
Assuming that the first MS 24 has an IMSI of 3,000,001, then the division of 3,000,001 by n (3) yields a remainder of “1”. PDSN 1 (14) is chosen having the IP address of “002”.
If the selected PDSN is not available, for reasons such as lost of connection, exceeding PDSN capacity, or simply no existing physical link, the PCF re-selects an alternative PDSN from the table, based on Mobile Station IMSI. This continuous selection process may iterate for a maximum of (n−1) times, until a PDSN is finally selected.
In one aspect of the invention, the continuous selection calculates a reselected PDSN as follows:
In terms of the example given above where n=3 and the IMSI is 3,000,001:
Now PDSN 2 (16) is chosen having the address of “003”. One problem with this reselection method is that there is a statistical tendency for MSs to reselect a common PDSN, causing a load imbalance in the system.
In another aspect of the invention the reselection process is handled differently. The number n is reduced by one, and the calculation repeated as follows:
However, this process does not always work. In terms of the example given above where n=3 and the IMSI is 3,000,001:
For example, assuming that the initial selection process selects PDSN 1 (14) (i.e. PDSN No.=IMSI % 3=1), but PDSN 1 (14) is not available. Then, the PDSN table is temporarily changed, for this mobile 24 only, to Table 3. Alternately stated, Table 3 is the PDSN selection table of PCF 1 (18), modified for the reselection process, when the initially selected PDSN is not available.
The old PDSN 1 (14) is temporarily removed from the table, and the old PDSN 2 (16) now becomes PDSN 1, as a result of keeping the PDSN Id numbers in an ordered sequence, or ascending re-ordering. The continuous selection uses the updated table to select between PDSN of address “001” and “003” using IMSI % (modulo) 2 in this example. After a successful continuous selection, the original PDSN table (with three entries in this example, Table 2) is restored for next normal Selection as described above.
With respect to
Assuming that the first MS IMSI is still 3,000,001, and n is equal to 4, the first PCF 18 selects PDSN 1 (14) as follows:
If the first MS 24 is later assigned to the second PCF 20, the selection process selects the same PDSN Id number, because the PCFs use the same sequence in their tables. Because the first PCF 18 and the second PCF 20 are both connected to PDSN 1 (14) the IP address is maintained without the requirement of a new PPP being established.
Table 5, below, depicts the PDSN selection tables configured in PCF 1 (18) and PCF 2 (20) of
The invention advantageously ensures that the first mobile's PPP connection is maintained for the duration of a packet call as best as possible. The likelihood of selecting the same PDSN when a MS roams to a new PCF is a function of the actual PCF-PDSN connectivity at the new and old PCF. Although the selection process does not guarantee consistent PDSN selections if the network is deployed such that PDSNs and PCFs are not fully connected, the probability of selecting the same PDSN increases as the connectivity increases. In the degenerate case (full connectivity), this selection algorithm always selects the same PDSN for a given MS Id, because every PCF has the same PDSN table and uses the same n (the “global” n) for modulo operation.
The unique selection process described herein is backward and forward compatible. The selection method can be implemented without a need for signaling or architecture changes to the existing 3G CDMA network architecture, so that it is backward compatible. The selection process affects only a single network component, the PCF. The software required to support enhancements and changes to future all-IP architecture is easily modified, so the PDSN selection is forward compatible.
Expanding cluster 2, where a new PDSN (PDSN 4) is being added, will not affect cluster 1. Only those PCFs that will add a direct link with the new PDSN are effected, such as PCF 2. Therefore, all MSs that roam within cluster 1 will not be affected.
In some aspects of the invention the system in Step 100 includes n PDSNs, with each PDSN in the system having a unique Internet protocol (IP) address. Then, the generation of the record of PDSN addresses in Step 102 includes creating a table with an ordered sequence of the PDSN Id numbers cross-referenced to the IP addresses. The selection of the PDSN in Step 106 includes dividing the first MS Id by the number of (n) PDSNs. More specifically, the selection of the PDSN in Step 106 includes selecting the PDSN Id number equal to the remainder in the quotient, when the first MS Id divided by n.
In some aspects of the invention the method further comprises Step 101, where the first MS communicates with a first PCF to request packet data communications. The generation of the table of PDSN addresses in Step 102 includes the first PCF having a table of cross-referenced PDSN Id numbers and IP addresses. The selection of the PDSN Id number in Step 106 includes using the table of the first PCF to calculate the PDSN Id number.
In some aspects of the invention Step 201 assigns a first PCF, to establish packet data communications between the first MS and a PDSN. Then, the selection of the first PDSN Id number in Step 206 includes the first PDSN Id number being selected, regardless of which PCF is assigned. Step 208 routes packet data communications between the first MS and the IP address corresponding to the selected PDSN Id number, through the assigned PCF.
The generation of the record of PDSN Id numbers in Step 202 includes creating a table with an ordered sequence of the PDSN Id numbers cross-referenced to the IP addresses. Then, the selection of the first PDSN Id number in Step 206 includes dividing the first MS Id by the number of (n) PDSNs in the table. Further, the selection of the first PDSN Id number in Step 206 includes selecting the PDSN Id number equal to the remainder in the quotient, when the first MS Id divided by n.
In some aspects of the invention, the system of PDSNs and PCFs are not fully connected, which complicates the PDSN selection process. That is, Step 200 includes the first PCF being connected to a IP address corresponding to the second PDSN Id number, but not to the IP address corresponding the first PDSN identification number. Then, Step 207 reselects an alternate PDSN Id number, after the first PDSN Id number has been calculated.
In some aspects of the invention, the reselection of an alternate PDSN Id number in Step 207 includes varying the first PDSN Id number, dividing that varied first PDSN Id number by n, and selecting the PDSN Id number equal to the remainder in the quotient. The first PDSN Id number can be randomly varied for the reselection of an alternate PDSN Id number, or varied by adding a “1”.
Alternately, the reselection of an alternate PDSN Id number in Step 207 includes: removing the first PDSN Id number and associated IP address from the table; dividing the first MS Id by the number of PDSNs remaining in the table; and selecting the PDSN Id number equal to the remainder in the quotient.
With respect to load balancing, the present invention selection method causes the load of services to be evenly partitioned among locally available PDSNs on the basis of the MS IMSI. When approaching PDSN-PCF full connectivity, the load is evenly distributed in the entire network. The present invention uses a simple and reliable algorithm. It is easy to administer, impacts only a single network entity, the PCF, and it requires few resources, just memory and computing power. Although specific examples have been presented to clarify the concept of the invention, the present invention is not limited to the particular embodiments mentioned above. Other variations and embodiments will occur to those skilled in the art.