Assessment of VoIP Service availability in the current internet
Voice over IP (VoIP)
Major findings:
- packet losses are not rare events, and it is generally worse on international paths.
- Network outages make up a non-negligible portion of packet losses
- Although research networks such as Internet2 has much lower delay and loss than the public Internet, the effect of network outages on both types of networks is almost the same.
Availability is the proportion of time that a service is available for use.
Reliability measures how long a service can stay up before it is disrupted, which is measured in terms of Mean Time Between Failure (MTBF) and Mean Time To Restore (MTTR)
In the context of telephony:
Service availability is an important concept in telephony for several reasons.
- The telephone service has become a vital and integral component of our daily life, upon which businesses and individuals depend.
The most important factor causing low quality, namely packet loss, can be ameliorated using quality improvement techniques such as FEC. (Wenyu Jiang and Henning Schulzrinne. Comparison and optimization of packet loss repair methods on voip perceived quality under bursty loss.)
Assessment of VoIP quality over internet backbones
Our assessment is based on delay and loss measurements taken over wide-area backbone networks, considers realistic VoIP scenarios and uses quality measures appropriate for voice. Our findings indicate that although voice services can be adequately provided by some ISPs, a significant number of paths lead to poor performance even for excellent VoIP end-systems. This makes a strong case for special handling of voice traffic on those paths. Even on the good paths, rare loss events can occasionally cause perceptible degradation of voice quality. The appropriate choice of the playout buffer scheme for each path was found to be of critical importance for the perceived quality.
We identify those aspects that may lead to poor voice quality. Three main characteristics:
- We use delay and loss measurements collected by sending probes between measurement facilities at five different US cities.
- Use appropriate voice quality measures that take into account various transmission impairments.
- We take into account the effect of the different components of the VoIP system, with emphasis placed on the playback buffer component.
Indeed, a large number of the paths performed poorly for VoIP traffic, mainly due to high delay and large delay variability that hurt voice much more than data traffic. Furthermore, if more stringent communication requirements, such as interactivity levels suited for business conversations, are imposed, these paths become totally unacceptable for telephony use.
-VoIP system
Encoder: periodically samples the original voice signal and assigns a (usually fixed) number of bits to each sample, creating a constant bit rate stream. The traditional sample-based encoder G.711 uses Pulse Code Modulation (PCM) to generate 8 bits samples per 0.125 ms, leading to a data rate of 64Kbps.
Further reduction in the data rate can be achieved if no signal is encoded during silence periods, a technique known as voice activity detection (VAD).
The packetizer follows the encoder and encapsulates a certain number of speech samples or a certain number of frames into packets of equal sizes and adds the RTP header.
As the voice packets are sent over an IP network, they are subject to variable delays and network drops.
An important component at the receiving end, is the playback buffer whose purpose is to absorb variations in delay and provide a smooth playout. The playback buffer may operate in one of two modes: fixed or adaptive.
A fixed scheme schedules the playout of a packet after a fixed (network and buffering) delay from its sending time, the same for all packets.
scheduling using timescale modification in packet voice communications”
The playout buffer delivers a continuous stream of packets to the depacketizer and eventually to the decoder which reconstructs the speech signal.
Decoders often implement Packet Loss Concealment (PLC) that produces a replacement for a lost packet, similar to the original one, by filling in silence or noise, by interpolating or even by regenerating the packet form the surrounding ones.
A survey of packet loss recovery techniques for streaming audio”
Each of the above components along the path of the packetized voice, may introduce delay and loss. The components of the end-to-end delay are the following
1) Encoding and packetization delay at the sender
2) Propagation, transmission and queuing delay in the network
3) Buffering and decoding delay at the receiver
Distortion of the original voice signal may occur:
1) At the low rate encoder
2) In the network due to loss
3) At the receiver due to drops in the playback buffer
Another important impairment is echo, the reflection of the participants’ signals, perceived as delayed and attenuated versions of their own voices. The larger the end-to-end delay, the more annoying is the echo.
Real-time voice over packet-switched networks
Quality bounds for packetized voice transport
- VoIP Quality assessment methodology
Network performance is usually presented in term of delay and loss statistics. A commonly used subjective metric is the Mean Opinion Score (MOS)