Method for reducing unnecessary traffic over a computer network

Description Submit all comments and votes

FIELD OF THE INVENTION

This invention relates generally to data device networks and, in particular, to apparatus and methods for connecting nodes to wireless networks usibg standard network protocols.

BACKGROUND OF THE INVENTION

With the tremendous growth of data processing by means of independent localized data processing devices, such as personal computers and minicomputers, data networks have evolved to connect together physically separated devices and to permit digital communication among the various devices connected to the network.

There are several types of networks, including local area networks (LANs) and wide area networks (WANs). A LAN is a limited area network and the data devices connected to a LAN are generally located within the same building. The LAN typically consists of a transmission medium, such as a coaxial cable or a twisted pair which connects together various computers, servers, printers, modems and other digital devices. Each of the devices, which are collectively referred to as "nodes", is connected to the transmission medium at an address which uniquely identifies the node and is used to route data from one node to another.

The WAN is used to connect devices together which are located at distances that are typically larger than the distances spanned by LANS. WAN networks often utilize existing public telephone networks and, thus can connect nodes located at great distances.

LANs and WANS are often connected together in various configurations to form "enterprise" networks which may span different buildings or locations or extend across an entire continent. Enterprise networks are convenient for several reasons: they allow resource sharing --programs, data and equipment are available to all nodes connected to the network without regard to the physical location of the resource and the user. Enterprise networks may also provide reliability by making several redundant sources of data available. For example, important data files can be replicated on several storage devices so that, if one of the files is unavailable, for example, due to equipment failure, the duplicate files are available.

In enterprise networks information to be sent from one node to another is generally divided into discrete messages or "packets" and the packets are transmitted between nodes in accordance with a predefined "protocol". In this context a "protocol" consists of a set of rules or procedures defining how the separate nodes are supposed to interact with each other.

In order to reduce design complexity, most networks are organized as a series of "layers" or "levels" so that information passing from one node to another is transmitted from layer to layer. Within each layer, predetermined services or operations are performed and the layers communicate with each other by means of predefined protocols. The purpose for the layered design is to allow a given layer to offer selected services to other layers by means of a standardized interface while shielding those layers frown the details of actual implementation within the layer.

In an attempt to standardize network architectures (the overall name for the sets of layers and protocols used within a network), a generalized model has been proposed by the International Standards Organization (ISO) as a first step towards international standardization of the various protocols now in use. The model is called the open systems interconnection (OSI) reference model because it deals with the interconnection of systems that are "open" for communication with other systems. The proposed OSI model has seven layers which are termed (in the order which they interface with each other) the "physical", "data link", "network", "transport", "session", "presentation" and "application" layers. The purpose of the OSI model is to attempt to standardize the processes conducted within each layer.

In accordance with the OSI model, the processes carried out in the physical layer are concerned with the transmission of raw data bits over a communication channel. The processes carried out in the data link layer manipulate the raw data bit stream and transform it into a data stream that appears free of transmission errors. The latter task is accomplished by breaking the transmitted data into data frames and transmitting the frames sequentially accompanied with error correcting mechanisms for detecting or correcting errors.

The network layer processes determine how data packets are routed from the data source to the data destination by selecting one of many alternative paths through the network. The function of the transport layer processes is to accept a data stream from a session layer, split it up into smaller units (if necessary), pass these smaller units to the network layer, and to provide appropriate mechanisms to ensure that the units all arrive correctly at the destination, with no sequencing errors, duplicates or missing data.

The session layer processes allow users on different machines to establish "sessions" or "dialogues" between themselves. A session allows ordinary data transport between the communicating nodes, but also provides enhanced services in some applications, such as dialogue control, token management and synchronization. The presentation layer performs certain common functions that are requested sufficiently often to warrant finding a general solution for them, for example, encoding data into a standard format, performing encryption and decryption and other functions. Finally, the application protocol layer contains a variety of protocols that are commonly needed, such as database access, file transfer, etc.

The layers are arranged in order to form a protocol "stack" for each node and the stacks are connected together at the physical level end. Thus, data transmission through the network consists of passing information down through one stack across the physical communication link to another protocol stack and passing the information up the other stack to the second node.

While the enterprise network works well, the recent proliferation in small portable data processing devices has lead to the rapid evolution of the wireless WAN network which can connect small mobile terminals to a land-based station and, in turn, to one or more enterprise networks. A typical architecture of a wireless WAN network comprises a number of mobile terminals connected by radio links to a base station. The base station is then, in turn, connected to a host computer or to an enterprise network by means of land lines.

Since wireless WAN networks involve at least one radio link, they exhibit characteristics of relatively low bandwidth and throughput when compared to ordinary enterprise networks. In addition, due to the large error rates involved in the radio link and the consequent necessity for retransmissions, the latency of the wireless network (the time which is taken to transmit a signal from one data transmission terminal to another and return an acknowledgement) is also quite high. Due to these characteristics, it has been found that it is inefficient, and in some cases, impossible to use standard communications protocols used in enterprise networks over wireless wide area networks. In particular, attempts to use common enterprise network protocols which were typically designed for much faster networks have resulted in excessive network traffic over the wireless WAN.

The conventional solution to this problem is to use specialized protocols for those networks which involve connections between a wireless network and various nodes. However, when specialized protocols are used, the protocol is often dependent on the exact network configuration. Significant additional development time is then often required to connect nodes to wireless WANs because custom protocol converters or gateways are needed. In addition, the use of specialized protocols often means that end-to-end reliable communication services are not available. Finally, existing network applications must often be reworked in order to utilize the specialized protocols.

Consequently, a method and apparatus is needed for networks which involve wireless WANs which method and apparatus will allow the use of standardized protocols to interface nodes with the wireless network while taking into account the special characteristics of the wireless WAN.

Further, tire method and apparatus must operate transparently with respect to both application programs in the nodes and the wireless network. Such a method will allow existing application programs to be used unmodified with the wireless network.

SUMMARY OF THE INVENTION

The foregoing problems are solved and the foregoing objects are achieved in accordance with illustrative embodiments of the invention in which standard protocol stacks, such as those commonly used on enterprise networks, are used to interface each node to the wireless network. Within the appropriate protocol stacks, the standard protocols are optimized by filtering and discarding some protocol packets, generating and "synthesizing" the reception of other protocol packets, and removing and transforming protocol header fields. The optimized protocol stream can be transmitted over the wireless WAN without seriously affecting WAN efficiency.

The optimization is accomplished by inserting an additional optimization layer between, the standard protocol stack and the wireless network driver. The optimization layer accepts normal inputs from the protocol stack and the driver and generates outputs which appear to the protocol stack to have come from a standard network driver and to the wireless network driver to have come from a wireless protocol stack. Consequently, the optimization layer operates transparently with respect to the existing stack and driver and both the stack and the driver behave as if they were operating in the environment for which they were designed.

The data packet stream passing through the standard protocol stack is converted in the optimization layer to the wireless protocol stream using the aforementioned optimizations. In particular, in accordance with one embodiment of the invention, the optimization layer reduces the number and size of data packets transferred over the wireless network by intercepting and interpreting the data packets before transmission over the network. The interpretation process includes filtering and discarding of data packets, generating and synthesizing the reception of data packets and removing or transforming selected protocol header fields.

In accordance with another embodiment of the present invention, the data packets are optimized as previously discussed and, in addition, conventional data compression is applied to both the reduced headers and the data in each data packet.

The reduction in the number of packets, the header size and the data compression reduces the amount of data which must be sent over to the wide area network to the point where the network traffic is sufficiently minimized for efficient use of the wireless WAN network.

BRIEF DESCRIPTION OF THE DRAWINGS

The above and further advantages of the invention may be better understood by referring to the following description in conjunction with the accompanying drawings, in which:

FIG. 1 is a block schematic diagram of a wireless WAN connected to a LAN-based enterprise network in accordance with the inventive principles, with each network including a variety of nodes.

FIG. 2 is a block schematic diagram network of prior art protocol stacks used to transmit information between two nodes over a LAN.

FIG. 3 is a block schematic diagram of prior art protocol stacks used to transmit data information between two nodes over a wireless WAN.

FIG. 4 is a block schematic diagram of the protocol stacks used to transmit data information between two nodes over a wireless WAN and modified in accordance with the present invention to include an optimization layer.

FIG. 5 is a more detailed block schematic diagram of prior art protocol stacks used to transmit data between two nodes structured in accordance with the international standards organization OSI seven-layer model.

FIG. 6 is a block schematic diagram of protocol stacks used to transmit data between two nodes in which the OSI seven-layer model has been modified in accordance with the present invention to include a optimization layer for optimizing the standard protocols so that they operate efficiently over a wireless WAN.

FIG. 7 is a block schematic diagram of protocol stacks modified in accordance with the present invention and used in transmitting information between two nodes in which a subnetwork is used for connectivity.

FIG. 8A is a block schematic diagram illustrating prior art protocol stacks which might be used in transmitting between two nodes over a prior art wireless network.

FIG. 8B is a block schematic diagram illustrating protocol stacks used in transmitting between two nodes over the prior art wireless network and modified in accordance with the principles of the present invention.

FIG. 9 is a block schematic diagram illustrating the basic structure and arrangement of software programs included in a client node which processes data and performs data compression and optimization in accordance with the present invention.

FIG. 10 is a block schematic diagram illustrating the software structure and arrangement in an illustrative server node incorporating the data compression and optimization of the present invention.

FIG. 11 is a schematic diagram illustrating a standard protocol header for a data packet to be transmitted from a server node over a wireless WAN showing the header both in original form and in reduced form after optimization in accordance with the present invention.

FIG. 12A schematically illustrates the construction of a data packet arranged in a wireless network protocol for transmission from the server node over an illustrative wireless network which data packet incorporates the reduced header shown in FIG. 11 in the "data" portion of the packet.

FIG. 12B schematically illustrates the reconstruction of the standard protocol header for the data packet shown in FIG. 12A after the packet has been received at the client node.

FIG. 13 is a schematic diagram illustrating a standard protocol header for a data packet to be transmitted from a client node over a wireless WAN showing the header both in original form and in reduced form after optimization in accordance with the present invention.

FIG. 14A schematically illustrates the construction of a data packet arranged in wireless network protocol for transmission from the client node over an illustrative wireless network which data packet incorporates the reduced header shown in FIG. 13 in the "data" portion of the packet.

FIG. 14B schematically illustrates the reconstruction of the standard protocol header for the data packet shown in FIG. 14A after the packet has been received at the server node.

FIG. 15 illustrates another data packet header illustrating how certain header fields can be replaced by tokens in accordance with the present invention.

FIG. 16 is a schematic illustration of retransmission request filtering performed in accordance with one aspect of the present invention.

FIG. 17A is a schematic illustration of another example of packet filtering performed in accordance with one aspect of the present invention.

FIG. 17B is a schematic illustration of still another example of packet filtering performed in accordance with one aspect of the present invention.

FIG. 17C is a schematic illustration of a further example of packet filtering performed in accordance with one aspect of the present invention.

FIG. 18A is a schematic illustration of a data transmissions and acknowledgements in a transmission between a node and server where packet "synthesizing" is not preformed.

FIG. 18B is a schematic illustration of an example of packet "synthesizing" performed in accordance with one aspect of the present invention.

FIG. 19 is a block schematic diagram illustrating the various software programs running in a server optimization layer constructed in accordance with the present invention which performs data processing and optimization and compression to allow the data packets to be efficiently used over a wireless WAN.

DETAILED DESCRIPTION OF ILLUSTRATIVE EMBODIMENTS

FIG. 1 illustrates, in schematic fashion, a wireless WAN, schematically designated by dotted box 100, connected to a LAN-based enterprise network schematically illustrated by dotted box 102. The wireless WAN consists of a plurality of client nodes, such as mobile PCs, of which PCs 104-110 are illustrated. PCs 104-110 communicate over a radio link with a base station (not shown) and a network switch 112. The network switch 112 is, in turn, connected, via a land link 116, to a server 124 which is part of enterprise network 102. Enterprise network 102 consists of various segments, including segments 120, 123 and 126. Segments 120 and 123 are connected by a server 118; segments 123 and 126 are connected by a server 124. Each segment may have various data utilization devices attached. For example, PC computers 114 and 122 are connected to segment 123 and PC 128 and host 130 are connected to segment 126.

The mobile PCs 104-110 communicate with the network switch 112 by means of radio packet modems (not shown) which are conventional devices. The network switch 112 is connected to the enterprise network server 124 over a communication link 116 using any one of a number of conventional protocols, for example, the well-known X.25 protocol. In accordance with the inventive principles, the data packets may be compressed using a lossless compression algorithm at the client nodes and sent, via the network switch 112, over the X.25 link 116 to the server node 124 where the packets are decompressed and sent to the remainder of the network. Similarly, data packets sent to server 124 by the remainder of the enterprise network 102 which packets are destined for any of the mobile clients 104-110 are compressed by server 124, sent over the transmission link 116 and transmitted by the wireless network to the selected client. The client then decompresses the packets.

There are several types of communications that can occur in a combined network, such as that shown in FIG. 1. First, one of mobile client nodes 104-110 may communicate with another mobile client node --for example, PC 104 may communicate through the network switch 112 to PC 106.

Alternatively, one of the client nodes connected to the enterprise network, for example node 122, may communicate with another node on the enterprise network, for example node 128, via server node 124. Node 122 may also communicate with host node 130 via server node 124.

Further, node 104 on the wireless WAN 100 may also communicate through network switch 112 and server node 124 with node 122 on the enterprise network 102.

As previously mentioned, these communications and other generalized network connections can be modeled as a "protocol stack" of layers in which selected data processing operations are performed in each layer and the layers communicate via standard protocols. FIG. 2 is an illustrative protocol stack modeling a connection established between two nodes located on the enterprise network 102, for example nodes 114 and 122. Node 114, or STATION 1, interfaces with the enterprise network 212 by means of the protocol stack consisting of layers 200, 204 and 208. The first layer is the application layer 200 which, as previously mentioned, handles protocols and interface information that directly communicate with a client application program running at the station.

Application layer 200, in turn, interfaces with standard protocol layers 204 where the protocols used in these layers are generally determined by the LAN and are standard for each type of network. The standard protocol layers provided by the network, in turn, communicate with media specific layer 208 through a standard interface layer 205 which is independent of the exact physical characteristics of the enterprise network 212.

At the second station, the information coming over the enterprise network 212 is provided to a media specific layer 210 which is analogous to layer 208. Layer 210, in turn, interfaces, over a standard interface 207, with standard protocol layers 206 which again are determined by the network to which STATION 2 is connected and are analogous to layers 204. Finally, the information passes to the application layer 202 which directly interfaces with the application program running in the second node, STATION 2.

It is important to note that, in the configuration illustrated in FIG. 2, the protocols used to provide communication services used by the client application programs are provided by standard protocol layers and transmission over a particular medium is handled by the media specific layer which does not directly communicate with the application program. Thus, as long as the network defined protocols are compatible with the underlying media, the application program needs to be written only to communicate with the standard protocols regardless of the actual media in use. In addition, the standard protocol layers do not need to be "aware" of the specific physical medium because they interface the media-specific layer through a standard interface.

However, as previously mentioned, the FIG. 2 arrangement assumes that the network protocols are compatible with the transmission media; if the standard network protocols are not compatible with the underlying media, then the arrangement illustrated in FIG. 2 either will not operate properly or will use the underlying medium inefficiently. In this situation, specialized protocols have conventionally been used. FIG. 3 illustrates a typical prior art layered configuration used when two nodes communicate via a wireless WAN. Such a connection may, for example, occur during communications between two nodes located on wireless WAN 100, such as nodes 104 and 106, or may also occur when a server node, such as server node 124, which is connected to WAN 100, communicates with a node on the wireless network such as node 104.

As with the previous layered arrangement, the protocol stack for STATION 1 includes an application layer 300 which communicates directly with the application program running in STATION 2. Application layer 300 then communicates with a protocol layer 304, however, in order to insure that the wireless network is used efficiently, non-standard wireless network specific protocol layers are used to 10 provide the communication services used by the application. These non-standard layers then communicate a media-specific layer 308 which, in turn, communicates directly with the wireless network 312.

At STATION 2, the information from the wireless network communicates with the media-specific layer 310 which then communicates with non-standard protocol layers 306. The non-standard protocol layers 306, in turn, communicate with application layer 302.

As previously mentioned the non-standard wireless specific protocol layers, 304 and 306, are the source of many drawbacks since they are specific to the particular wireless network that is being used. For example, applications designed for standard LAN networks will generally not operate on the wireless networks because the protocols are not compatible and, thus, each application program must be modified to operate with the particular non-standard protocol layer used on a given network. Also, existing communications equipment (for example, routers) used in enterprise networks does not recognize these protocols. Finally, end-to-end reliable communications cannot be achieved if the enterprise network equipment does not support the non-standard protocol.

FIG. 4 illustrates a protocol stack used for communication over a wireless WAN which communications stack has been modified in accordance with the present invention. As with the prior art arrangement of FIG. 3, STATION 1 has an application layer 400. However, in this modified stack, application layer 400 interfaces with standard protocol layers 404, in an analogous manner to FIG. 2. These standard protocol layers may be the same protocol used on LAN networks and, thus, the same application can be connected to both a LAN network as well as a wireless WAN without modification.

In accordance with the invention, the standard protocol layers 404 interface with the lower layers through the standard interface 405. The invention inserts an optimizing layer 408 transparently between the standard protocol layer 404 and the media specific layer 412. The optimizing layer 408 interfaces to the standard protocol layers 404 through the standard interface 405 and converts the standard protocol stream into an optimized protocol stream which is suitable for transmission over the wireless WAN 416. The optimizing layer 408, in turn, interfaces with a media specific layer 412 which actually communicates with the wireless network 416.

At STATION 2, a media specific layer 414 interfaces with the wireless network 416. The media specific layer 414, in accordance with the invention, interfaces with a second optimizing layer 410 which converts the optimized protocol stream received over the wireless network 416 into a standardized protocol stream which is suitable for use with the standard protocol layers 406 through standard interface 409. Protocol layers 406, in turn, interface with application layer 402 at STATION 2.

As previously mentioned, data communications protocols are characterized by the exchange of protocol data units (PDUs) between protocol stacks and each optimizing layer 408 and 410 reduces the number and size of PDUs transferred over the wireless network by intercepting and interpreting the PDU's being sent. This interpretation process includes filtering and discarding PDUs, generating and synthesizing the reception of PDUs, removing or transforming protocol header fields and other traffic optimization techniques such as compressing the data. These operations make it feasible to use the standard protocols over the wireless network.

Since both application layers 400 and 402 interface only with standard protocol layers, the applications do not have to be reworked if they are connected to a LAN or a wireless network. In addition, since the standard protocol layers 404 and 406 interface only with the standard interface 405, they do not have to be reworked to run over different underlying transmission media. However, since the wireless network 416 sees only the optimized information, it operates efficiently as if non-standard protocol layers had been used.

More specifically, the optimizing layers 408 and 410 are designed in such a manner that the interface between layers 408 and 410 and standard protocol layers 404 and 406, respectively, is the same as if layers 404 and 406 were connected directly to media specific layers 412 and 414. Similarly, the interface between layers 408 and 410 and media specific layers 412 and 414 is the same as if layers 412 and 414 were connected directly to standard protocol layers 404 and 406, respectively. Therefore the optimizing layers 408 and 410 can be inserted into the protocol stack without changing either the standard protocol layers 404 and 406 or the media specific layers 412 and 414 and the operation of the optimizing layers is completely transparent to the network system.

The exact location of the optimizing layers used in accordance with the present invention is shown by a comparison of FIGS. 5 and 6. FIG. 5 is a diagram of a prior art protocol stack used to connect two nodes in accordance with the OSI standard seven-layer architecture. In accordance with the OSI standard, each protocol stack for a node consists of seven layers. For example, stack 536 for STATION 1 comprises an application layer 500, a presentation layer 502, a session layer 504, a transport layer 506, a network layer 508, a data link layer 510 and a physical layer 512. The operation and the purpose of each of these layers has been previously discussed. Similarly, stack 538 for STATION 2 consists of layers 522-534. Although the actual data communication occurs between the two physical layers 512 and 534 over a data link 520, when the stack is arranged as shown in FIG. 5, each layer can be thought of as communicating with its "peer" which is a layer at the same level as a given layer. For example, the application layers 500 and 522 can be thought of as communicating directly even though information passes through all of the layers 502-512, across data link 520 and back through the layers 534-524. This "peer-to-peer" communication is schematically illustrated by a dotted line 514. Similarly presentation layers 502 and 524 can communicate peer-to-peer as illustrated by dotted line 516 and network layers 508 and 530 can be also be thought of as communicating by shown schematically by dotted line 518.

FIG. 6 discloses an illustrative positioning of the inventive optimization layers in the OSI seven-layer structure. In particular, the optimization layer 610 is inserted into the protocol stack 640 between the network layer 608 and the data link layer 612. Similarly, optimization layer 634 is inserted in stack 642 between network layer 632 and the data link layer 636.

In the OSI seven-layer reference model, the optimization layers 610 and 634 intercept PDUs generated by network layers 608 and 632 which contain nested higher layer PDUs and optimization processing is performed on both the link and higher layer PDUs.

FIG. 7 shows a common case in which optimized PDUs produced by the optimization layers 710 and 738 are transmitted over a subnetwork using another protocol stack. The initial part of the protocol stack 748 comprises layers 700-708 and is the same as the normal OSI model. Similarly, the initial part of the protocol stack 750 comprises layers 728-736 and is the same as the normal OSI model. In this case, the optimization layers 710 and 738 typically interface to network layers 712 and 740 in the subnetwork. However, it is possible that other layers could also be used.

Although the previous description has concerned generic networks, for simplicity, the following description will assume an interconnection of a LAN and wireless WAN network in accordance with two more specific protocols. However, it will be immediately obvious to those skilled in the art that any existing networks can be used with optimization layers which utilize the principles and operations of the present invention. In the particular embodiments discussed below, the standard "LAN-type" protocol used in this network is called the TP/NP protocol and is used with the OSI layered network architecture. This protocol is typical of such LAN protocols as the NetWare network protocol used on NOVELL networks. The characteristics and operation of this network is well-known and will not be discussed further herein.

The wireless network used in the following description is a generic network similar to many wireless networks presently existing such as the ARDIS, RAM or CDPD networks. The characteristics and operation of these networks are well-known and will not be described further herein in detail.

The basic protocol stack diagram for the prior art wireless network is shown in FIG. 8A which illustrates a connection between a mobile client node (stack 838) and a server node on a LAN network (stack 846). The client node protocol stack 838 communicates with a radio packet modem 840 which, in turn, communicates with base station 842. Base station 842 communicates with a message switch 844 which, in turn, communicates to the LAN server stack 846. Application program 800 can thought of as communicating with its peer program application/gateway program 822 shown schematically by dotted line 816. Similarly, the non-standard protocol layer 802 can be thought of as interfacing directly with the non-standard protocol layer 824 via dotted line 818.

Protocol stack 838 in the mobile client node consists of the application layer 800, non-standard protocol layer 802, and a protocol layer 804 for the protocol used by the radio packet modem 840. The non-standard layer 802 is network specific and must be used by clients and hosts/gateways which access the wireless network. Protocol layer 802 provides the means whereby the mobile client node identifies the host to which it wants to communicate and other options, such as the use of acknowledgements.

The modem protocol layer 804 converts the non-standard protocol used in layer 802 to the radio modem protocol (RM) used to interface with the radio packet modem 840. This latter protocol is both network and modem specific.

The radio packet modem, in turn, communicates with the base station 842 by means of a radio protocol (RP). The modem/base station radio protocols generally include significant error correction overhead and, if retries and acknowledgements are taken into account, the effective throughput over the radio link is typically only 10% to 50% of the nominal throughput depending on the traffic being carried over the network.

The information is transferred over the radio link in the radio protocol and is received at the base station 842. At base station 842, yet another transformation is conventionally made to convert the information in the radio protocol 812 into an internal format 812 denoted as "X" in FIG. 8A. The information is then forwarded to the network message switch 844.

In switch 844, the information is converted from the internal "X" format 814 to a standard data link protocol 820, which may be, for example, the conventional X.25 format. The information, now in X.25 format, is typically forwarded to the server node stack 846 over a land link.

In the server protocol stack, the X.25 infor