Browse Prior Art Database

Automatic Selection and Distribution of a Network System Clock

IP.com Disclosure Number: IPCOM000115049D
Original Publication Date: 1995-Mar-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 4 page(s) / 138K

Publishing Venue

IBM

Related People

Cazaentre, JM: AUTHOR [+5]

Abstract

New voice and video applications require the network interface bandwidth managers to service information in a continuous bit stream. When continuous bit stream transmission is required, non-synchronous transmission methods are not sufficient to guarantee certain grades of service to the customer for transmission of synchronous information.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 48% of the total text.

Automatic Selection and Distribution of a Network System Clock

      New voice and video applications require the network interface
bandwidth managers to service information in a continuous bit stream.
When continuous bit stream transmission is required, non-synchronous
transmission methods are not sufficient to guarantee certain grades
of service to the customer for transmission of synchronous
information.

      Synchronous signals that are transported through the network,
require the distribution of a high-accuracy clock at both ends of the
network to synchronize the transmitter and receiver in a network.  If
the transmitter sends at a bit rate higher than the receiver, the
switching system will overflow causing a frame to be lost.  If the
transmitter sends at a bit rate lower than that used by the receiver,
the switching system underflows and requires retransmission of the
frame.  Either occurrence is called a controlled slip.  The impact of
slips are a reduction of throughput on digital data, a click on
voice, and a drop-out of a call on voice-band data or facsimile.

      In continuous bit stream transmission, the number of bits into
the network must be equal to the number of bits out of the network.
With stratum 3 frequency tolerances of 20 parts per million at both
the transmit and receiving ends of a network, this amounts to four
frame slips every second which is unacceptable.  This technical
report describes a method for ensuring that a highly accurate clock
is distributed to all bandwidth managers in a network.  It also
describes a method for central and distributed clock management in
case of failure.

      To ensure the delivery of a continuous stream of data the
bandwidth manager extracts a highly accurate clock from the network
and synchronizes all its clocks to it (Fig. 1).  Since the receiving
bandwidth manager and the transmitting bandwidth manager synchronize
clocks to the highly accurate clock, frame slips and loss of
information are virtually non-existent.

      As shown in Fig. 2, a central system clock card receives a
number of reference clocks from the various Line Interface Couplers
(LICs).  These reference clocks vary in accuracy depending on their
source.  In addition an external "office clock" can be received if
the accuracy of the networks clocks is not sufficient.  Since the
low-speed LICs can receive four different network clocks, logic on
the low-speed LICs chooses one-of-four clocks and sends it to the
system clock card as a reference.

      The system clock card then prioritizes the accuracy of the
clocks based on initial system configuration parameters.  When
operating in simplex mode with only one clock card installed, the
microcode specifies the system clock to be sent to all LICs.  When
operating in duplex mode with redundancy, the microcode specifies one
clock card as a primary and one clock card as a secondary.  In this
mode of operation, the primary clock card...