Text preview for : 5965-5297E.pdf part of HP 5965-5297E HP Publikacje 5965-5297E.pdf
Back to : 5965-5297E.pdf | Home
Cell Transfer Time Measurement Using the
Parallel Cell / Traffic Generator and Analyzer
How to use the HP E4829B custom / UTOPIA Level 1
implementation for cell transfer time variation measurements
Product Note Product Number
HP E4829B
bandwidth which is normally Application
wasted when other techniques,
such as TDM, are used (see [2]). Network equipment
manufacturers
To measure cell transfer time
variation in a communication For hardware and software
system interacting with ATM cells, engineers in the communications
the HP E4829B test system is able industry, designing ICs and
to log and extract time-stamp modules for ATM switches, or
information flowing into and from working in system integration, the
the payload of ATM cells. This HP E4829B parallel cell / traffic
Figure 1: HP E4829B entry system
measurement can be made at generator and analyzer system,
Introduction parallel to parallel (UTOPIA) which can measure cell transfer
ports, either at the switch fabric time variation, is the basic tool for
In order to achieve quality of ports or at the ATM layer device the evaluation of complex
service in an ATM switch, any interfaces, as shown in figure 2. ATM designs at parallel interfaces
variations in the cell transfer time (UTOPIA).
of channels carrying voice data
must be closely monitored. When
the delay variation exceeds certain
limits, the quality of voice sent to
the user is reduced. The designer's
Switch, hubs, routers
goal, therefore, is to ensure that
any variation in the transfer time CPU
of the traffic at the switch fabric is 8/16 n bit PARALLEL
`enhanced cell'
minimized. The ATM forum also
recommends that the cell transfer Standard ATM cell: SERIAL data !!
ATM payload ATM header PHY ATM ATM PHY
time variation is carefully 46 byte 5 byte layer Switch layer
fabric
evaluated (see [1]).
PHY ATM ATM PHY ATM payload ATM header
layer layer
Current developments in the field 46 byte 5 byte
of voice compression have
resulted in a new adaptation class, ATM payload ATM header
46 byte 5 byte
the real-time variable bit rate UTOPIA Custom Custom UTOPIA
(VBR-RT), which stipulates that HP E4829B HP E4829B
the adaptation process must
guarantee defined end-to-end Internal data: 8 or 16 bit PARALLEL
delay-and-delay jitter. This Figure 2: Block diagram showing how the HP E4829B connects to an ATM switch with
protects up to 60% of the parallel interfaces
Cell Transfer Time Measurement, HP E4829B, Product Note, Page 1
Component Test
Semiconductor manufacturers
provide a large variety of off-the-
shelf devices for the ATM industry,
including line interface ASICs,
ATM layer controllers and chip
and chip sets for the switch fabric
itself.
In order to achieve high standards
of quality throughout the system,
the individual chip or chip set
must have already been character-
ized for its share of the overall
cell transfer time variation budget.
Instructions for use
Figure 3: Cell editor of the HP E4829B showing a cell including a time-stamp segment
Cell definition
Analysis ments, the transmitter and the
The time-stamp is a segment receiver queue paths are
available when the cell structure is The HP E4829B offers a new compensated and the time-stamp
defined. The length of the segment feature called `measure time,' counters are synchronized.
is fixed at 32 bits, which is which actually performs the cell
equivalent to four bytes in an transfer time variation The received cells are sorted into
ATM-8 or two words in an ATM-16 measurement. The receiver is three categories: early, in-time and
application, and it can be placed forced to generate a time-stamp late, although the user has to
anywhere within the payload of when the cell arrives, which is provide the values for both limits.
the cell. Figure 3 shows a cell then compared with the time- Figure 4 illustrates the principle of
including a time-stamp segment. stamp extracted from the cell's the system, whereas figure 5
payload. From this, the transfer shows the setup in the processing
The definition of cells on the time is calculated. To guarantee editor window.
transmitter and receiver is the validity of these measure-
identical. As soon as the cell is
generated, the time-stamp
segment, situated on the
transmitter, is filled with
bits generated by the time-
stamp counter. On the
receiver, it defines the
bytes / words which are
used for calculating the cell
transfer time. The trigger
cell can also be used to
specify other parameters,
such as the measurement TX time-stamp
of the transfer time of
specific cells (e.g. RX time-stamp
matching cells for a
dedicated VPI/VCI) within
the incoming cell stream. Figure 4: Principle of cell transfer time measurement
Cell Transfer Time Measurement, HP E4829B, Product Note, Page 2
Results
The counter window, illustrated in
figure 6, displays the results of a
cell transfer time variation
measurement. There are three
counters, each assigned to one of
the three categories (early, in-time
and late), which display the
number of cells generated
according to specific limits. Once
the receiver has started, the
counter window is updated in
specific measurement intervals, as
defined by the user.
Implementation
Cell transfer time variation Figure 5: Processing editor window
measurements can be performed
on the 8 and the 16 bit solution of Synchronization lines on the VXI back plane. It is
the HP E4829B, which is equipped also possible to select a trigger
with the necessary features Before starting a measurement, line for each module.
(including software revision the time-stamp generators of the
A.2.3.0) to carry out these transmitter and receiver need to External clock
measurements. be synchronized. This is achieved
by using the `sync' command for Any external clock can be used for
Time-stamp segment the TX and RX with one module. the TX or RX. However, to achieve
However, for a system containing the highest measurement