Packet loss occurs when the jitter buffer of an endpoint does not receive packets or receives the packets too late for processing. A longer delay or disordered packets can also amount to packet loss. Also, the network might appear to be losing packets when the network intentionally discards the packets because of late arrival at the endpoint. Unintentional packet loss in the network and discarded packets in the jitter buffers of the receiving endpoints characterize the quality of IP networks.
Packet loss can be bursty or more evenly distributed. Bursty packet loss has a greater effect on voice quality than distributed packet loss. Therefore, a 1% bursty loss has a more adverse effect than a 1% distributed loss.
The following are some effects of packet loss on a VoIP network:
Every codec has a Packet Loss Concealment (PLC) method and because of the PLC, it becomes difficult for the network to detect packet loss. Therefore, a PLC-enabled compression codec, such as the G.729A, provides better voice quality than a full bandwidth G.711 codec without a PLC.
Packet loss is more noticeable for tones such as fax tones or modem tones (other than DTMF) than for voice. The human ear can most likely detect packet loss during a tone, which uses a consistent pitch, than during speech, which uses a variable pitch.
Packet loss is more noticeable for contiguous packet loss than for random packet loss over time. For example, the effect of losing 10 contiguous packets is worse than losing 10 packets evenly spaced over an hour.
Packet loss is usually more noticeable with larger voice payloads per packet than with smaller packets, because more voice samples are lost in a larger payload.
In the presence of packet loss, the time for a codec to return to normal operation depends on the codec type.
Even minimal packet loss such as 0.12% can greatly affect the capability of a TTY/TDD device meant for people who are hard of hearing.
Packet loss for signaling traffic increases network traffic substantially when the loss exceeds 3%, possibly impacting voice quality.