Multiplex communications – Data flow congestion prevention or control – Control of data admission to the network
Reexamination Certificate
1999-06-07
2003-08-26
Nguyen, Steven (Department: 2663)
Multiplex communications
Data flow congestion prevention or control
Control of data admission to the network
C370S231000, C714S749000
Reexamination Certificate
active
06611495
ABSTRACT:
BACKGROUND
The present invention relates to systems and methods for data transfer in packet-switched communication networks. More particularly, the present invention relates to systems and methods for improving the data transfer rate in packet-switched communication networks.
Communication networks can be classified as either circuit-switched or packet-switched networks. Circuit-switched networks, traditionally used in voice communication networks such as telephone networks, operate by establishing a logical connection between two points on the network. Typically the connection is of a fixed bandwidth or capacity regardless of the desired data transfer rate of the sender. Circuit-switched networks are widely regarded as being inflexible in their operation and inefficient in their allocation of bandwidth, or capacity, throughout the network. By contrast, packet-switched networks, traditionally used in data communications, operate by carving information into packets, which are then sent across the network to a desired recipient node. Packet-switched networks may be connection-oriented, such that a logical connection is maintained between the sender and the receiver, or connection-less, such that no logical connection is maintained between a sender and a receiver on the network. Packet-switched networks are widely regarded as being more flexible and substantially more efficient in allocating bandwidth than circuit-switched networks. Due in part to these factors, demand for packet-switched network services has grown dramatically over the past two decades. It is anticipated that demand for packet-switched networks will continue its dramatic growth.
Packet-switched data networks are designed and based on industry-wide data standards such as the open system interface (OSI) model or an implementation of the Transmission Control Protocol/lnternet Protocol (TCP/IP) stack. According to the OSI model, a network system is typically represented by a multi-layered protocol stack. The TCP/IP protocols are commonly modeled as a four-layer protocol stack. Layer
1
, referred to as the link, data link, or network interface layer, manages the interfaces and device drivers for interfacing with the physical elements of the network. Layer
2
, referred to as the network layer or the internet layer, is responsible for routing data packets on the network. The Internet Protocol (IP) is a network layer protocol. Layer
3
, referred to as the transport layer, is responsible for carving the data to be transmitted into appropriately sized packets for transmission across the network. The transport layer also manages error recovery and congestion control processes, acknowledging received packets and setting timeouts to ensure that transmitted data is received by the intended receiver. The Transmission Control Protocol (TCP) is a transport layer protocol. Layer
4
, referred to as the application layer, is responsible for managing particular TCP/IP applications. Exemplary TCP/IP applications include SMTP for e-mail, FTP for file transfers, Telnet for remote login, etc.
The TCP/IP protocol suite is the primary communications protocol suite applicable to the transfer of digital information across multiple, connected computer networks like the internet. The operation of a network according to TCP/IP protocols may be explained, at a conceptual level, as follows: TCP applications pass data for transmission to processes at the transport layer. The transport layer receives data for transmission from an application such as, for example, e-mail, and carves the data into appropriately sized packets, referred to as segments. The transport layer then passes the data segments to a network layer protocol utility, which compiles data packets according to network layer protocols like IP, referred to as datagrams, and routes the datagrams across one or more networks to the intended receiver. The network layer, in turn, requests the services of link-layer protocol utilities to manage a particular physical medium such as, for example, an ethernet connection.
The IP enables a connectionless, unreliable delivery service that may be used by numerous transport-layer protocols, including TCP. The IP layer is connectionless in that the IP maintains no logical relationship between successive datagrams transmitted across the network. Each datagram is treated by the network as an individual piece of information. Thus, even though a large file transfer may require sending hundreds, even thousands of datagrams, the IP layer of the network is unaware during the transfer that the packets are related. Individual packets may be routed across different paths through the network and/or may arrive at the destination in any sequence. The IP is unreliable in that robust error recovery procedures are not implemented at the IP layer.
TCP is a connection-oriented protocol. At the TCP (transport) layer, a logical connection is established between a TCP sender and a TCP receiver. As discussed above, at the sender node the TCP layer is responsible for dividing data into appropriately-sized segments for transmission. At the receiver node the TCP layer is responsible for acknowledging received segments and for reassembling related segments. Additional functions implemented at the TCP layer include error recovery and congestion control. In broad terms, error recovery refers to the process of ensuring that packets transmitted by a sender were, at a minimum, received by the intended recipient. Error recovery may include additional algorithms to ensure that the recipient received the correct data.
TCP uses a retransmission function as one error recovery procedure. Pursuant to TCP, a TCP sender retransmits data packets for which an acknowledgment packet is not received within a predetermined time period. TCP maintains a retransmission timer (REXMT) to implement the error recovery procedure. For each acknowledgment (ACK) segment returned by the receiver, the sender reinitializes the REXMT with a retransmission timer value (RTO) that is a function of the round trip time (RTT). In many implementations of TCP, the RTT is continuously measured by the TCP sender. In alternate embodiments, the RTT may be updated based upon actual RTT measurements. Radio Link Protocol (RLP), a link-layer protocol for wireless communications also implements a retransmission timer-based error recovery procedure, similar to the routine implemented by TCP.
Although various implementations of the retransmission routine have been effective, they may unnecessarily reduce data transfer rates in the network. Thus, there is a need for improved systems and methods for data transfer in packet-switched networks, including TCP networks. The present invention uses new techniques for managing a REXMT that enhance the data transfer rate of a network.
SUMMARY
The present invention addresses these and other problems by providing improved systems and methods for implementing packet-switched data transport services. More particularly, the present invention provides a retransmissionbased error recovery procedure that facilitates improved data transfer rates in comparison to known retransmission timer error recovery procedures. According to one aspect of the invention, the data transfer rate may be improved by reinitializing the retransmission timer with a value that compensates for the elapsed time since the transmission time of a previously transmitted packet from the sender to the receiver. Advantageously, the system allows error correction to be managed on a per-packet basis, yet requires only a single timer to be implemented, thereby saving computational expense.
The present invention further provides improvements in data transfer rates by reducing the probability of unnecessary timeouts at a sender node. More particularly, the present invention provides a method for recalculating a retransmission timer value that determines whether a TCP sender has enough transmitted, but unacknowledged, segments outstanding to trigger a fast retransmit routine. If so, the RTO is calculated to include a factor that i
Ludwig Reiner
Meyer Michael
Duong Duc
Nguyen Steven
Telefonaktiebolaget LM Ericsson (publ)
LandOfFree
System and method for improved data transfer in... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for improved data transfer in..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for improved data transfer in... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3111657