20.11.2012 Views

Contents Telektronikk - Telenor

Contents Telektronikk - Telenor

Contents Telektronikk - Telenor

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

150<br />

ATM cell<br />

header<br />

2.1.1 Measuring the cell error ratio<br />

using the AAL-3/4 protocol<br />

In order to calculate the CER a test traffic<br />

of ATM cells with AAL3/4 protocol<br />

should be generated. An overview of this<br />

protocol structure is shown in Figure 5.<br />

See [6] for more information on the<br />

AAL3/4 protocol.<br />

The AAL-3/4 protocol includes a cyclic<br />

redundancy code (CRC)-10 field which<br />

should detect single bit errors in the entire<br />

ATM cell payload. The error detection is<br />

performed on a cell basis, therefore lost<br />

or misinserted cells should not influence<br />

the number of CRC errors (i.e. the number<br />

of errored cells) detected.<br />

During this measurement period the following<br />

primary measurement values<br />

were recorded:<br />

- Selected cell bandwidth (used to derive<br />

the total number of user cells)<br />

- Duration of measurement period<br />

- Number of CRC-10 errors.<br />

The following secondary measurement<br />

values were also recorded for verification<br />

of the functionality:<br />

- Number of corrected headers (single<br />

bit header errors)<br />

- Number of bad headers (multiple bit<br />

header errors of multiple consecutive<br />

single bit header errors)<br />

- Number of lost cells.<br />

By calculating the relationship Number<br />

of CRC10 Errors to Total Number of<br />

User Cells we obtain a figure for the cell<br />

error ratio. If a secondary measurement<br />

value was found to be larger than expected,<br />

then the measurement was deemed<br />

invalid.<br />

2.1.2 Measuring the cell error ratio<br />

using the AAL5 protocol<br />

In order to calculate the cell error ratio a<br />

test traffic of ATM cells with AAL5 protocol<br />

should be generated. An overview<br />

of this protocol structure is shown in Figure<br />

5.<br />

ATM cell payload<br />

HEADER SN SNP INFORMATION PAYLOAD<br />

Figure 6 ATM cells supporting the AAL1 protocol<br />

The AAL5 protocol includes a CRC-32<br />

field which should detect single bit errors<br />

in the entire ATM cell payload. The error<br />

detection is performed on a protocol data<br />

unit (PDU) basis, but by knowing the<br />

PDU length (i.e. the number of ATM<br />

cells used to transport each PDU) we can<br />

here too calculate the cell error ratio. In<br />

addition, if we generate PDUs which are<br />

carried by a single cell (i.e. as shown in<br />

the figure above) then the likelihood of<br />

multiple errors occurring in a single PDU<br />

is reduced.<br />

During this measurement period the following<br />

primary measurement values<br />

must be known or recorded:<br />

- Selected Cell bandwidth (used to<br />

derive the total number of cells)<br />

- Duration of measurement period<br />

- Number of CRC-32 errors<br />

- PDU length (number of cells used to<br />

carry each PDU).<br />

The following secondary measurements<br />

values were also recorded for verification<br />

of the functionality:<br />

- Number of corrected headers (single<br />

bit header errors)<br />

- Number of bad headers (multiple bit<br />

header errors of multiple consecutive<br />

single bit header errors)<br />

- Number of lost cells.<br />

By calculating the relationship Number<br />

of CRC32 Errors to Total Number of<br />

Cells and having knowledge about the<br />

length of the segmentation and reassembly<br />

(SAR)-PDU we obtain a figure<br />

for the cell error ratio. If a secondary<br />

measurement value was found to be<br />

larger than expected, then the measurement<br />

was deemed invalid.<br />

2.2 Cell loss ratio<br />

The cell loss ratio (CLR) is defined as<br />

the ratio between the number of cells lost<br />

to the number of user cells transmitted in<br />

a defined period of time.<br />

At least three possible causes of cell loss<br />

are identified. The cell delineation mechanism,<br />

which is based on the verification<br />

of the header error control (HEC) field<br />

may correct single bit errors and detect<br />

multiple bit errors. Double bit errors lead<br />

to a cell discarding, while more severe<br />

errors may lead to even worse conditions<br />

such as cell misinsertion. Due to the statistical<br />

nature of ATM traffic, buffer<br />

overflow may also occur. Finally, the<br />

user parameter control (UPC) function<br />

may discard cells if the agreed traffic<br />

contract parameters are violated.<br />

A fourth condition which will cause cell<br />

loss is the occurrence of severely errored<br />

cell blocks, though this condition should<br />

be filtered from any measurement period.<br />

Due to the difficulty in filtering this<br />

occurrence and the fact that these should<br />

occur relatively seldom, measurements<br />

may be necessary to perform several<br />

times if the results indicate that such an<br />

event has taken place.<br />

Normally, severely errored cell blocks<br />

should cause an Out of Cell Delineation<br />

(OCD) event and result in a Loss Of Cell<br />

Delineation (LOC) alarm. By measuring<br />

the total duration in the OCD state during<br />

the measurement period (if such an event<br />

should occur), we may calculate the<br />

number of cells which should have<br />

passed during the OCD state and exclude<br />

these from the measurement results.<br />

Two methods of measuring cell loss are<br />

possible; cell losses detected using the<br />

sequence number, and cell losses detected<br />

using cell blocks. Because this second<br />

method was not supported by the test<br />

equipment obtained, the first method was<br />

chosen. Two different protocols for testing<br />

the CLR have been looked at and are<br />

described in the following sections.<br />

2.2.1 Measuring the cell loss ratio<br />

using the AAL-1 protocol<br />

In order to calculate the cell loss rate a<br />

test traffic of ATM cells with AAL-1<br />

protocol should be generated. An<br />

overview of this protocol structure is<br />

shown in Figure 6.<br />

The AAL-1 protocol includes a Sequence<br />

Number (SN) field and a Sequence Number<br />

Protection (SNP) field, thereby providing<br />

a very reliable cell surveillance<br />

mechanism. Lost cells should only<br />

remain undetected when a multiple of<br />

eight cells disappear. Though if a multiple<br />

of eight cells are lost the switching<br />

system should also report a Loss Of Cell<br />

Delineation and this may constitute a

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!