Specifications
Chapter 6 Diagnostics and Troubleshooting Installation and Operation Manual
6-10 Monitoring Performance IPmux-24 Ver. 1.5
Parameter Description
Jitter Buffer
Underflows
The number of seconds with jitter buffer underflows since the last clear.
IPmux-24 is equipped with a “Packet Delay Variation Tolerance” buffer, also called a “jitter buffer”,
responsible for compensating for IP networks delay variation (IP jitter). The jitter buffer is configured
in milliseconds units and exists for each bundle independently.
Explanation:
Packets leave the transmitting IPmux-24 at a constant rate, but the problem is that they are reaching
the opposite IPmux-24 at a rate which is NOT constant, due to network delay variation (caused by
congestion, re-routing, queuing mechanisms, wireless media, half-duplex media, etc.). The TDM
devices at both ends require a constant flow of data, so they can’t tolerate delay variation. Therefore
the jitter buffer is required in order to provide the TDM equipment with a synchronous and constant
flow.
This is done as follows:
• Upon startup, the jitter buffer stores packets up to its middle point (the number of packets
correlates to the buffer’s configured depth in milliseconds). Only after that point it starts
outputting the E1/T1 flow towards its adjacent TDM device. The stored packets assure that the
TDM device will be fed with data even if packets are delayed by the IP network. Obviously, if
packets are delayed too long, then the buffer is gradually emptied out until it is underflowed. This
situation is called buffer starvation. Each underflow event increases the jitter buffer underflow
counter by one and indicates a problem in the end-to-end voice/data integrity.
The second functionality of the jitter buffer is that in adaptive mode the jitter buffer is also a part of
a mechanism being used to reconstruct the clock of the far end TDM side.
An underflow situation can be a cause of:
• Buffer starvation: Packets delay variation causes the buffer to empty out gradually until it is
underflowed.
• Continuous Sequence Errors. The sequence error means a halt in the valid stream of packet arrival
into the jitter buffer.
• Packets are being stopped/lost/dropped.
• Too small jitter buffer configuration that can’t compensate for the network delay variation.
• When all system elements are not locked on the same master clock, it will lead to a situation in
which data is clocked out of the jitter buffer at a rate different from the one it is clocked into. This
will gradually result in either an overflow or underflow event, depending on which rate is higher.
The event will repeat itself periodically as long as the system clock is not locked.
• When an overflow (see below) situation occurs, IPmux-24 instantly flushes the jitter buffer, causing
a forced underflow. So when you need to calculate the real underflow events and not the
self-initiated ones, subtract the number of overflows from the total number of underflows counted
by the device.
Recommendations:
• Try increasing the jitter buffer size.
• Check reasons for sequence errors or lost/dropped packets (if present), system clocking
configuration, Ethernet environment (full duplex) and connection, packets drop/loss/ignore by
routers/switches or non-uniform packets output by routers/switches due to queuing mechanisms.
• Make sure the same amount of TS for bundle is configured on each side of the IPmux-24
application, and that the “TDM bytes in frame” parameter is identical in both IPmux-24 units.
• Make sure Ethernet/IP network provides priority (Quality Of Service) to the IPmux-24 traffic.
Priority may be achieved by three means: VLAN tagging, IP TOS marking or by using the constant
2142 decimal value at each IPmux “UDP destination Port” field.