Browse Prior Art Database

Output Throttle Mechanism

IP.com Disclosure Number: IPCOM000100437D
Original Publication Date: 1990-Apr-01
Included in the Prior Art Database: 2005-Mar-15
Document File: 6 page(s) / 205K

Publishing Venue

IBM

Related People

Bederman, S: AUTHOR

Abstract

This article describes an "output throttle" mechanism that controls the flow of traffic from two types of output queues (such as voice and data) into a B-Channel that supports LAPE. The types of Layer 2 virtual circuits that are supported are: . Multiple, connection-oriented, circuits that carry data traffic by means of I-frame packets. . A single, connectionless circuit that carries voice traffic by means of UI-frames.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 32% of the total text.

Output Throttle Mechanism

       This article describes an "output throttle" mechanism
that controls the flow of traffic from two types of output queues
(such as voice and data) into a B-Channel that supports LAPE.  The
types of Layer 2 virtual circuits that are supported are:
.   Multiple, connection-oriented, circuits that carry data traffic
by means of I-frame packets.
.   A single, connectionless circuit that carries voice traffic by
means of UI-frames.

      Each of the above types of virtual circuits is assigned a small
pool of buffers that are dedicated to the circuit.

      The "output throttle" causes queued frames to be extracted from
the various buffers in a manner that provides guaranteed bandwidth
for the voice packets, and fair access for the data packets.

      Voice packets that are exchanged during a conversation between
two individuals are required to meet the following objectives in
order to ensure satisfactory comprehension by the listeners:
.   The time for packets to travel from a speaker to a listener must
be kept short.  (This article assumes that a delay of approximately
36 msec is acceptable in a system that employs digital telephones.)
.   The variability in the arrival time between successive packets
must be kept small enough such that the output stream of voice
samples appears to be synchronized with the input stream of voice
samples.
.   Accurate delivery of voice packet information, while desirable,
is not a stringent requirement.  (This article assumes that
connectionless Layer 2 protocols will be used for voice packets.  UI
frame formats with 7 octets of overhead are used for these packets.)

      Data packet transmission can be characterized as follows:
.   Data packets, unlike voice packets, can tolerate considerable
variability in the time required for transit through a packet
network.  Data packets can be flow controlled in a manner that allows
the flow rate to increase or to decrease as bandwidth availability
fluctuates.
.   Accurate delivery of data packets can be a stringent requirement.
 (This article assumes that data packets will be transmitted using
connection-oriented Layer 2 protocols.  I-frames, with 8 octets of
overhead are used for these packets.)

      Refer to Fig. 1.  The transmission time on a B-Channel can be
divided into 12 msec intervals, with 7 msec of this time reserved for
the possible transmission of a voice packet.

      A 64 Kbps B-Channel can transmit 8 octets per msec. Thus, in 7
msec, 56 octets could be transmitted.  This article assumes that a
voice packet contains 55 octets (including 7 overhead octets).  Such
a voice packet requires slightly less than 7 msec for transmission.

      Fig. 1 shows approximately 5 msec being allotted for a "maximum
length" I-frame.  During this time interval, 40 octets could be
transmitted on a B-Channel.  Assuming 8 overhead octets in an
I-frame, this implies that the maximum all...