Multiplex communications – Data flow congestion prevention or control – Control of data admission to the network
Reexamination Certificate
1998-09-24
2001-05-15
Ton, Dang (Department: 2661)
Multiplex communications
Data flow congestion prevention or control
Control of data admission to the network
C370S236000, C370S401000, C370S410000, C370S445000, C370S469000, C370S471000
Reexamination Certificate
active
06233224
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a communication method with a data communication protocol for packet transfer between a transport layer and a data link layer, and a data communications terminal using the data communication protocol for data communication.
2. Description of Related Art
The data communication protocol being the interfacing condition for data communications, typically the TCP/IP (Transmission Control Protocol/Internet Protocol) and OSI (Open Systems Interconnection) protocols, has a hierarchical structure as shown in FIG.
1
. As shown, the hierarchical structure consists, at the level of operations executable by application programs, of an application layer
16
containing characteristics of data communications terminals, coding control function, etc., presentation layer
15
containing communication-oriented profiles, document-oriented profiles, etc., session layer
14
having a control function for data transmission and reception IDs, etc., transport layer
13
for carrying out a transport procedure independent of any network because the transport layer
13
is an upper layer, network layer
12
for control of call setting, disconnection, etc. for each network, data link layer
11
for establishing, releasing, error control, etc. of a data link, and a physical layer
10
for setting a user and network voltage physical interface conditions.
For packet transmission, a packet generated by a software of the application layer
16
(also including the presentation layer
15
and session
14
) is transferred to a software which implements the transport layer
13
, and then sequentially from the transport layer
13
to the network layer
12
, from the network layer
12
to the data link layer
11
, and further to a lower layer. Finally, the packet is represented in the form of an electric or light pulse for transmission from a data communications terminal.
In the conventional packaged data communications terminal, the packet is copied from a memory only twice. That is, the memory copying is done when the packet is moved from the application layer
16
to the transport layer
13
in a main memory
1
and when the packet is moved from the main memory
1
of the data communications terminal to a packet buffer
3
in a network interface module
2
via a bus
5
, as shown in FIG.
2
. The packet is moved between other layers higher than the data link layer
11
just by passing a pointer to a location in the main memory
1
where the packet is stored.
In the memory copying from the application layer
16
to the transport layer
13
, a flow control is made to avoid any data loss due to a packet buffer overflow in the transport layer
13
. Namely, when the packet buffer in the transport layer
13
is about to overflow, such a control will be done that the application layer
16
suspends packet copying. However, data flow between the data link layer
11
and a higher layer will not be controlled. The packet buffers of the data link layer
11
include one in the network interface module
2
and one in the main memory
1
. Since no flow control is done, however, even if both the buffers overflow, the higher layer will continuously transfer packets to the data link layer
11
with the result that the packets will be discarded.
The reason for the above lies in the basic design policy for the conventional data communications protocol that the reliability of data communications is not assured for the layers lower than the transport layer
13
and the interaction between the layers is minimized.
The reliability of the data communications is assured only by packet retransmission from the transport layer
13
. Therefore, even if there is no space for packets in the packet buffers of the data link layer
11
, the protocol in the transport layer
13
does not suspend the packet transmission, thus the packets are lost and thereafter packets are retransmitted.
Conventionally, however, a packet is generated in the application layer
16
, and transferred to the data link layer
11
via the transport layer
13
and network layer
12
at a lower rate than the rate of packet transmission from the network interface module
2
to outside the terminal, so that there occurs little packet loss due to overflow of the packet buffer in the data link layer
11
.
The conventional data communication protocol is likely to be much deteriorated in throughput of data communications as the terminal performance and network function have become higher. More particularly, since the performance of the terminal components such as CPU, etc. has rapidly become improved, the rate of packet transfer to the data link
11
from a higher layer is higher than that of the packet transmission from the network interface module
2
to outside the terminal in some cases. In such a case, there is a possibility that the packet buffers of the data link layer
11
overflow, resulting in a packet loss even in the terminal.
Also, the data link layer
11
has functionally been improved to freely change the packet transmission rate as in the ATM (asynchronous transmission mode). In this case, there is a possibility that the rate of packet transmission from the data link layer
11
to outside the terminal is controlled to a low one depending upon how much congested the network is. As the packet transmission rate is thus low, the difference between the low rate of packet transmission from the data link layer
11
to outside the terminal and the rate of packet transfer to the data link layer
11
from a higher layer is larger, the possibility of the packet loss will be larger.
When a packet is lost, the transport layer
13
has to retransmit a packet, so that the data transmission throughput will be lower and the bandwidth of the network be wasted. Particularly in a transport protocol for congestion control such as TCP, when a packet loss is detected, the packet transmission rate is autonomously controlled to a low one by reducing the transmission window size or the like, the data transmission throughput will be remarkably low.
FIG. 3
shows the characteristic curve showing TCP throughput vs. maximum transmission window size.
The round trip time (RTT) between a transmitting terminal and a receiving terminal is 20 ms. With a large RTT, the transmission window size has to be increased in order to provide a higher data transmission throughput by TCP. The transmission window size referred to herein means a maximum data length that can be batch-transmitted with response of no reception acknowledgment. If the transmission window size is increased excessively, the packet buffers of the data link layer
11
overflow, resulting in a packet loss. This is caused by the fact that a higher layer will batch-transfer data in an amount for the transmission window size to the data link layer
11
. If a packet is lost, the availability control of TCP is put into action to largely lower the data transmission throughput. The conventional data communication protocol is disadvantageous in that the to-be-avoided packet loss inside the terminal lowers the data transmission throughput and network availability.
SUMMARY OF THE INVENTION
Accordingly, the present invention has an object to overcome the above-mentioned drawbacks of the prior art by providing a communication method and data communications terminal in which a packet transferred between a transport layer and a data link layer is not discarded and thus a packet is lost inside the terminal, thereby preventing the data transmission throughout and network availability from being lower.
The above object can be accomplished by providing a communication method in which data flow between a transport layer and a data link layer is controlled. That is, a packet transfer from the transport layer is controlled to prevent packet buffers of the data link layer from overflowing by any of the following two methods:
In the first method of flow control, when the packet buffers of the data link layer are so highly occupied that the buffers are likely to overflow, it i
Shionozaki Atsushi
Tanaka Hiroyuki
Utsumi Shusuke
Yamashita Kei
Crosby, Heafey Roach & May
Sony Computer Laboratory, Inc.
Ton Anthony
Ton Dang
Wigert, Jr. J. William
LandOfFree
Communication method and data communications terminal, with... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Communication method and data communications terminal, with..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Communication method and data communications terminal, with... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2486629