BACKGROUND: There currently exists the concept of a TCP connection that is set up over a wireless link, such as in a cellular network. Wireless TCP (WTCP) is a TCP protocol that is optimized for use in wireless environments. Wireless links are prone to long latencies and high error rates. These factors can make the conventional TCP perform poorly in the wireless link environment, as the end-to-end TCP dynamics is usually not fast enough to track the variations in the wireless link quality and thus the wireless link, which usually is the bottleneck in the end-to-end path, is not optimally used. In an attempt to overcome these problems the Internet Engineering Task Force (IETF, http ://ietf. org/html. charters/ pilc-charter. html), as well as the Open Mobile Alliance (OMA, http ://www. openmobilealliance. org/tech/affiliates/wap/wap-225-tcp-20010331-a. pdf) have recommended a number of enhancements to conventional TCP to make it more efficient over wireless links.

One proposed enhancement recommends splitting the end-to-end TCP connection at certain intermediate network entities. For example, and referring to Fig. 1, if a mobile station (MS) 1 is to set up a TCP connection with application web server 2 in the wired Internet 4, it is recommended that one TCP connection be set up over a wireless access network 5 between the MS 1 and an intermediate network entity, sometimes referred to as a performance enhancing proxy (PEP) 3, while another TCP connection is set up between the PEP 3 and the web server 2. This enables the connection between the MS 1 and the intermediate network entity (the PEP 3) to run a version of TCP that is optimized for wireless links (also referred to as wireless profiled TCP), while the other connection runs the conventional (wireline) TCP. The intermediate network entity or PEP 3 can reside on a WAP proxy or on a gateway between the wireless access network 5 and the Internet 4.

As may be appreciated, an appropriate signaling mechanism is required to set up a split TCP connection in a cellular communications network, such as in the CDMA network architecture or in the GPRS/tJMTS network architecture. Further, in the case of a mobile-to-mobile session, the PEPs 3 at each of two ends may need to communicate so that an appropriate split TCP connection can be set up.

However, while it is known in the prior art (e. g., IETF's Performance Implication of Link Characteristics (PILC) working group and the WAP 2.0 protocol) to recommend the use of the split TCP connection for performance enhancement, no mechanisms are available or have been proposed, to the knowledge of the inventors, to actually set up such split TCP connections in a cellular environment. Note in this regard that if the PEP 3 is to reside on a priori known network elements, such as on a WAP Proxy, then no specific mechanism is needed other than to have. address of the WAP Proxy configured in the mobile station or other type of terminal. However, if the PEP 3 is to be hosted on a network element such as the GGSN/PDSN, these elements are not a priori known to the mobile station, as the specific GGSN/PDSN that the terminal connects to depends on the location of the mobile station.

SUMMARY OF THE PREFERRED EMBODIMENTS The foregoing and other problems are overcome, and other advantages are realized, in accordance with the presently preferred embodiments of these teachings.

This invention provides a mechanism to set up a split TCP connection in a cellular network, and involves identifying the PEP and informing the PEP about the required connection parameters, such as the IP address and TCP port number of the remote end of the connection. The PEP, in response, sends certain parameters to the MS so that a split TCP connection can be set up.

Disclosed herein is a method, and a system that operates in accordance with the method, to set up a TCP session between a mobile terminal (MT) or mobile station (MS) and an end point destination via a wireless network and the Internet. The method includes sending a split TCP request from the MS to a PEP located in the wireless network where the split TCP request includes information for identifying a network address of the MS and a network address of the end point destination. The method further includes, in response to receiving the split TCP request from the MS, establishing a split TCP connection that includes a wireless TCP (WTCP) connection between the MS and the PEP, and a TCP connection at least part way between the PEP and the end point destination. In one embodiment the end point destination is an application server that is coupled to the Internet, and the TCP connection is established between the PEP and the application server. In another embodiment the end point destination is a second MS that is coupled to a second PEP in a second wireless network, and the TCP connection is established at least as far as the second PEP, and then as a WTCP connection from the second PEP to the second MS.

In another embodiment the MSA and the MSB are served by the same PEP using two different WTCP connections.

In one exemplary embodiment the wireless network is a CDMA network, and the PEP is embodied in or is co-located with a Packet Data Support Node (PDSN) in the CDMA network. In another exemplary embodiment the wireless network is a General Packet Radio Service (GPRS) or Universal Mobile Telecommunications System (UMTS) network, and the PEP is embodied in or is co-located with a Gateway GPRS Support Node (GGSN) in the GPRS or UMTS network.

BRIEF DESCRIPTION OF THE DRAWINGS The foregoing and other aspects of these teachings are made more evident in the following Detailed Description of the Preferred Embodiments, when read in conjunction with the attached Drawing Figures, wherein: Fig. 1 is a diagram that illustrates a split TCP connection as proposed in the prior art; Fig. 2 shows an embodiment where a PEP is located with a PDSN in a CDMA cellular architecture, and is useful in describing a MS-web server WTCP embodiment; Fig. 3A shows an embodiment where a PEP is located with a GGSN in a GPRS/UMTS cellular architecture, and is also useful in describing the MS-web server WTCP embodiment; Fig. 3B shows an embodiment where a first PEP is located with a first GGSN in a first GPRS/UMTS cellular architecture and where a second PEP is located with a second GGSN in a second GPRS/UMTS cellular architecture, and is useful in describing aMS- MS WTCP embodiment ; Fig. 3C shows an embodiment where a single PEP is located with the GGSN in the GPRS/UMTS cellular architecture, and is useful in describing a MS-MS WTCP embodiment ; Fig. 4 is a logic flow diagram that illustrates a method for accommodating the MS-web server embodiment of Fig. 3A; Fig. 5 is a logic flow diagram that illustrates a method for accommodating the MS-MS multi-PEP embodiment of Fig. 3B ; and Fig. 6 is a logic flow diagram that illustrates a method for accommodating the MS-MS single PEP embodiment of Fig. 3C.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Described herein are several embodiments of split TCP connections for the case of the MS-server connection and for the case of the MS-MS connection. The case of the MS- MS connection is based on, and extends, the solution for the MS-server connection case.

The general messaging and message flow is described in the context of embodiments for CDMA and GPRS/UMTS architectures. For the case of the MS-server connection the term"server"in a generic fashion, and as used herein implies that this end point of the connection is a node that does not require wireless TCP (WTCP) service.

In embodiments of this invention the end point of the connection may not be the web server 2 that is connected to the conventional wired Internet 4, but it maybe another MS, thereby requiring a MS-to-MS connection to be established as described below in relation to Fig. 3B. Of course, the second MS may be functioning as a web server, just not one that is connected the wired Internet.

Case (i): MS-web server connection Assume that the MS 1 wishes to initiate a wireless TCP connection with the web server 2. The following conventions can be used in the description of this case: xl, yl: IP address and port number at the MS 1 x2, y2 : IP address and port number at the PEP 3 that the MS 1 communicates with x3, y3: IP address and port number at PEP 3 that the server 2 communicates with x4, y4: IP address and port number at the server 2 that the PEP 3 communicates with Referring also to Fig. 4, this embodiment of the invention performs the following procedure : 4A. The MS 1 sends a message to the PEP 3 specifying the need for split TCP connection. This split TCP connection request message includes, or is sent in conjunction with, information that is descriptive of (x4, y4) and (xl, yl). Note that the (xl, yl) information may be implicit in the split TCP connection request message as it, is the <BR> <BR> address and port of the origin of the message (i. e. , the address and port of the MS 1).

4B. The PEP 3 selects (x2, y2) and (x3, y3) to be used in the split TCP connection. This may be accomplished by a table lookup procedure.

4C. The MS 1 and the PEP 3 establish a WTCP connection between (xl, yl) and (x2, y2).

For this purpose the PEP 3 may first send a message to the MS 1 informing the MS of (x2, y2), and the MS 1 may then initiate the TCP connection setup (for example, by the use of a conventional SYN, SYN ACK and ACK transaction). In an alternate embodiment, the PEP 3 may initiate the TCP connection setup with the MS 1.

4D. The PEP 3 and the server 2 setup a TCP connection between (x3, y3) and (x4, y4) (for example, by using the conventional SYN, SYN ACK and ACK handshake used to establish TCP connections).

Embodiment for CDMA2000 architecture: Referring to Fig. 2, the PEP 3 may reside on or be co-located with a Packet Data Support Node (PDSN) 6 in the CDMA network architecture defined by 3GPP2. The PEP 3 may be implemented as a software module that is executed by a data processor of the PDSN 6, or by another data processor in the CDMA network architecture. The PEP/PDSN 6 is reached from the wireless link 5 via a radio network (RN) element 7 and an A10/A11 interface 8.

For accessing the Internet 4 over the CDMA network, the MS 1 first establishes a conventional Point-to-Point Protocol (PPP) connection with the PDSN 6. If using simple IP, the MS 1 obtains its IP address during the PPP connection establishment. For mobile IP, the PDSN 6 and the MS 1 perform agent advertisementlsolicitation signaling for the IP address assignment. The established PPP connection can then be used to carry out Step 4A from the MS 1 to the PDSN 6. A Resv message defined in 3GPP2 Specification X. P0011. *, Wireless IP Network Standard, February 2003, can be used in a modified form to encode the request to establish the split TCP connection and to convey the reverse link (x4, y4) information. The (x2, y2) information in Step 4C can be conveyed from the PDSN 6 to the MS 1 in a ResvConf message, also defined in 3GPP2 Specification X. P0011. *, Wireless IP Network Standard, February 2003, which functions basically as an acknowledgment to the Resv message.

Embodiment for GPRS/UMTS architecture: Further by example, and referring to Fig. 3A, the PEP 3 can reside (in a General Packet Radio System or GPRS embodiment) on or be co-located with a Gateway GPRS Support Node (GGSN) 13 in the GPRS, or in an Universal Mobile Telecommunications System (UMTS) network architecture, as defined by 3GPP. The PEP 3 maybe implemented as a software module that is executed by a data processor of the GGSN 13, or by another data processor in the GPRS/UMTS network architecture. In this embodiment the PEP/GGSN 13 is reached from the wireless link 5 via a base station system (BSS) 9 that is connected to a conventional mobile switching center (MSC) 10 and to a serving gateway support node (SGSN) 11 that in turn connects to an Internet Protocol (IP) core network 12 and thence to the GGSN 13.

When accessing the Internet 4 over the GPRS network, the MS 1 first establishes a primary Packet Data Protocol (PDP) context with the GGSN 13. The MS 1 obtains its IP address during the PDP context establishment procedure, as defined in 3GPP Technical Specification TS 23.060, General Packet Radio Service (GPRS); Service description; Stage 2 (Release 1999), Version 3.10. 0, January 2002. While setting up the PDP context, the MS 1 also specifies the desired service from the GPRS/UMTS access network. This can be accomplished by the use of an access point name (APN) field in a PDP Activate message. As in Step 4A for setting up the split TCP connection, the MS 1 may initiate the PDP context (primary or secondary) with the GGSN 13 by providing an APN value that is predetermined to be indicative of the request for WTCP service. The MS 1 preferably also includes the (x4, y4) information along with this message. The (x2, y2) information in Step 4C can be conveyed from the GGSN 13 to the MS 1 in an acknowledgment to the PDP Activate message, as describedin 3GPP Technical Specification TS 23.060, General Packet Radio Service (GPRS); Service description; Stage 2 (Release 1999), Version 3.10. 0, January 2002.

Case (ii) : MS-MS connection The foregoing discussion of Figs. 2,3A and 4 considered the case in which the MS 1 establishes a WTCP connection with the server 2. What is now described is the case, shown for convenience and not by way of limitation in a GPRS architecture system in Fig. 3B, in which a MS 1, say MS_A, establishes a WTCP connection with another MS 1, say MSB. The following naming convention is used during the description of this case: xl, yl : IP address and port number at MS A x2, y2: IP address and port number at PEP (PEP_A) that MS A communicates with x3, y3: IP address and port number at PEP A that the PEP for MSB (PEPB) communicates with x4, y4: IP address and port number at PEP B that PEP A communicates with x5, y5: IP address and port number at PEP B that MSB communicates with x6, y6: IP address and port number at MSB In this embodiment it may be assumed that both MSA and MSB desire to have a WTCP connection. The associated PEPs 3 are PEP A and PEPB, respectively.

Referring also to Fig. 5, this embodiment of the invention performs the following procedure: 5A. The MSA sends a message to PEP A specifying a request for a split TCP connection to be set up, along with information about MSB (at least (x6, y6)) and also the IP address and port number at MSA (xl, yl). The (xl, yl) information may be implicit in the message, as it is the address and port of the origin of the message.

5B. The PEP A selects (x2, y2) and (x3, y3) to be used in the split TCP connection, such as through a table look up procedure.

5C. The MSA and PEP A establish a WTCP connection between (xl, yl) and (x2, y2).

For this purpose the PEP A may first send a message to the MS A informing MSA of (x2, y2), and MSA may then initiates a WTCP connection setup. In an alternate embodiment, PEP A initiates the WTCP connection setup.

5D. PEP A begins the process of establishing a TCP connection with MSB. Once MSB receives a TCP SYN message from PEP A, it begins the procedure of establishing a WTCP connection with PEP B.

5E. MSB sends a message to PEP B conveying its WTCP set up request, along with the appropriate end-point information, (x3, y3) and (x6, y6). The (x6, y6) information may be implicit in the message, as it is the address and port of the origin of the message.

5F. The PEP B selects (x4, y4) and (x5, y5) to be used in the split TCP connection, such as through a table look up procedure. The MSB and PEP B establish a WTCP connection between (x6, y6) and (x5, y5). For this purpose the PEP B may first send a message to the MSB informing MSB of (x5, y5), and MS B may then initiates a WTCP connection setup. In an alternate embodiment, PEP B initiates the WTCP connection setup.

5G. PEP B, from (x4, y4), also sends a TCP-SYN ACK to PEP A, at address (x3, y3), and encapsulates MSB's information (x6, y6) in this TCP-SYN ACK packet. When PEP A sees the encapsulated TCP-SYN ACK packet, it understands that it is a special case where PEP-B is acting as a proxy for MSB for this particular TCP-SYN request.

5H. PEP A initiates a TCP connection with PEP B with end-points (x3, y3) and (x4, y4), respectively, using a conventional TCP SYN, SYN ACK and ACK handshake.

As can be appreciated, after performing steps 5A-5H there are WTCP connections established between MS-A and PEPA, and between MSB and PEP B, while there is a conventional TCP connection established between PEP A and PEP B.

The process by which the MSs A and B acquire IP addresses in the CDMA2000 and GPRS/UMTS architectures, in this case, is similar to what is described in the previous case.

The foregoing description has provided by way of exemplary and non-limiting examples a full and informative description of the best method and apparatus presently contemplated by the inventors for carrying out the invention. However, various modifications and adaptations maybecome apparent to those skilled in the relevant arts in view of the foregoing description, when read in conjunction with the accompanying drawings and the appended claims. As but some examples, the use of other similar or equivalent system architectures, message types and signalling protocols maybe attempted by those skilled in the art.

Furthermore, it is possible that MSA and MSB may be served from a single PEP 3, such as when they are two mobile stations connected to the same network operator.

Reference in this regard can be made to Fig. 3C for showing the exemplary GPRS embodiment (note that the CDMA embodiment of Fig. 2 can be employed as well, where MSA and MS-B are both served from the same PEP 3/PDSN 6). In the embodiment of Fig. 3C it is assumed that both MSA and MSB desire to have a WTCP connection.

Also in this embodiment it is assumed that (x3, y3), the IP address and port number at PEP A that PEP S communicates with, and (x4, y4), the IP address and port number at PEP B that PEP A communicates with, are not used. Further, it is assumed that (x5, y5), the IP address and port number at PEP B that MSB communicates with, is modified so as to represent the IP address and port number at PEP 3 that MSB communicates with.

Alternatively, parameters (x7, y7) could be defined for this purpose, it being realized that the embodiments of Figs. 3B and 3 C could very well both be used together (depending on where MSB happens to be located). Referring to Fig. 6, this embodiment of the invention performs the following procedure: 6A. The MSA sends a message to PEP 3 specifying a request for a split TCP connection to be set up, along with information about MS B (at least (x6, y6)) and also the IP address and port number at MSA (xl, yl). The (xl, yl) information may be implicit in the message, as it is the address and port of the origin of the message.

6B. The PEP 3 selects (x2, y2) and modified (x5, y5), shown as (x5, y5) M in Fig. 6, or (x7, y7), to be used in the split TCP connection, such as through a table look up procedure.

6C. The MSA and PEP 3 establish a WTCP connection between (xl, yl) and (x2, y2).

For this purpose the PEP 3 may first send a message to the MSA informing MSA of (x2, y2), and MSA may then initiate a WTCP connection setup. In an alternate embodiment the PEP 3 initiates the TCP connection setup.

6D. PEP 3 begins the process of establishing a TCP connection with MS_B. Once MSB receives a TCP SYN message from PEP A, it begins the procedure of establishing a WTCP connection with PEP 3.

6E. The MSB and PEP 3 establish a WTCP connection between (x6, y6) and the modified (x5, y5). For this purpose the PEP 3 may first send a message to the MSB informing MSB of (x5, y5), and MS_B may then initiate a new TCP connection setup. In an alternate embodiment the PEP 3 initiates the TCP connection setup.

As can be appreciated, after performing steps 6A-5E there are two WTCP connections established, one between MSA and PEP 3 and one between MSB and PEP 3. In operation, the PEP 3 operates to copy from port-to-port As a further exemplary and non-limiting modification to the foregoing teachings the MS 1 need not send (x4, y4), in the embodiment of Fig 4, or (x6, y6), in the embodiment of Figs. 5 and 6, but may instead send a URL (such as http://www. someserver. com or http ://www. somemobile. someoperator. com), and the corresponding address information can be provided using a conventional domain name server (DNS) procedure. However, all such and similar modifications of the teachings of this invention will still fall within the scope of this invention.

Furthermore, some of the features of the present invention could be used to advantage without the corresponding use of other features. As such, the foregoing description should be considered as merely illustrative of the principles of the-present invention, and not in limitation thereof.