User's Manual Part 2

Tech-X Flex
®
(NG2) Tech-X Flex User Guide - Firmware v06.50
6-49
Intro
Overview
Wi-Fi
Ethernet
System
IP/Video
MoCA
RF
Specs
measured loss is only accurate when consecutive loss events are smaller than the capacity of the
continuity counter, which is 0-15 (4 bits). In other words, the maximum amount of measurable
consecutive loss is 15 packets. Also, note that a packet with an errored sync byte or a transport error
indicator set will be discarded and considered lost for the purpose of this measurement.
The unit also uses continuity counter values to determine out-of-order packets and the counter range of
0-15 provides a related accuracy limitation. The basic unit behavior is to consider any late packet that
arrives within 7 packets of expected order as out-of-order, otherwise it is considered to be a member of
the next counter “set.” This behavior is best illustrated by an example, as follows...
Assume that all packets are arriving as expected, when packet 2 of a counter set goes missing (that is,
packet 3 arrives after packet 1). At that point, packet 2 is initially considered lost. If packet 2 finally arrives
sometime before packet 9, its status changes to out-of-order and the respective cumulative counts are
adjusted accordingly. However, consider instead a scenario where packet 2 arrives after packet 10. In
this case, the original packet 2 is considered permanently lost and the packet that arrives is considered
to be packet 2 of the next set, at which point the originally-expected packets 11, 12, 13, 14, 15, 0, and 1
are initially considered lost. If these packets then arrive normally, their status changes to out-of-order and
the respective counts are adjusted accordingly. When the “real” packet 2 arrives for the next set, the unit
has two “packet 2’s” in the buffer and must assume that the original packet 2 is out-of-order for some
unknown previous set, so it increments the out-of-order count again and resets the algorithm. In this
scenario, a single late packet has caused the lost count to increment by one and the out-of-order count to
increment by 8.
The DF, presented as a quantity of time, is the maximum observed imbalance in stream flow over the
measurement interval, with respect to the expected media payload rate. That is, it effectively reports how
much buffering would be required to fully compensate for network delay variation at the respective node.
As such, it also indicates the amount of latency that must be introduced in order to properly decode the
stream. To calculate the DF, the software uses a “virtual buffer” concept, using the ingress of packets
versus the expected “drain” rate (that is, the media rate) to determine the variance. In some respects, the
DF provides a high-level view of the delay variation experienced by packets transiting from source to
destination. It may be useful to quantify the performance of the audio/video streams and transport
network over time and to adjust equipment buffers accordingly.
For convenience, Spirent has implemented a proprietary algorithm to convert MLR and DF calculations
into a score that resembles a mean opinion score (MOS), as defined by the ITU-T. This scoring method,
referred to as “MDI-S,” uses a scale of 1 - 5 to indicate perceived viewer experience with the following
typical benchmarks:
Score Quality Human perception of degradation
5 Excellent Imperceptible. No degradation of video quality can be
detected by a human viewer.
Preliminary issue - Limited distribution only!