Friday, 29 December 2017

An Enhanced Sender-Based Packet Loss-Recovery Technique 
in VoIP

VoIP Voice over Internet Protocol (VoIP) is a telephony technology that commonly uses the real-time transport protocol (RTP) to transport voice packets over a IP based network. RTP runs on top of the user datagram protocol (UDP), and thus it is an unreliable delivery protocol. When a packet loss occurs, the quality of the audio at the receiving endpoint degrades because the receiving endpoint does not have voice data for regenerating the lost segment of the audio. Researchers proposed many techniques for improving quality of service (QoS) in the face of packet loss. Some of these techniques employ receiver-based packet loss concealment (PLC) approaches and others employ sender-based loss- recovery techniques (SBLR), where the sender assumes an active role to help the receiver recover lost data or improve QoS when packet loss occurs. Most SBLR mechanisms work by retransmitting voice data or by transmitting additional data. In what follow, this article briefly describe various sender-based loss- recovery techniques currently used.

Interleaving: This technique adds part of the same voice signal segment in different packets thus spreading the impact of loss over longer time period. Forward Error Correction (FEC) works by transmitting redundant packets for error correction. Redundant Data Transmission (RDT) works by transmitting audio data more than once. This technique includes previously transmitted audio data along with new audio data in a single IP packet. Duplicate Packet technique transmits the redundant data in separate IP packets and thereby increases bandwidth consumption by requiring additional data and header overhead.
Redundant and duplicate data transmission is shown to produce the best audio quality among other techniques in different packet loss situations (single packet, burst of 2,3… n packets). These approaches however, add overhead in term of more bandwidth requirement, CPU processing and packet transmission delay and may increase network congestion which lead to higher packet loss and may drops the voice call.
The proposed technique works toward improvement of QoS in VoIP system without adding excessive overhead in term of more network bandwidth requirement or processing/transmission delay.

The SBLR approach commonly is designed using several Operation factors. These include:

  • Degree of Redundancy :-This represents the amount of previously transmitted audio data to be retransmitted along with new audio data in a single IP packet.
  • P-persistence parameter :- Extensions of SBLR conventional redundant class works by randomly transmitting redundant audio with the redundancy rate depends on a pre-determined P-persistence parameter to improve QoS in VoIP system.
  • Threshold Value:-The SBLR technique redundancy may depend on fixed thresholds. When new packet arrives, the loss rate is compared to threshold if the loss is greater than the threshold, redundant data and new data will be included in the IP packet. Otherwise only new data will be sent.
  • Network QoS Factors:-The path taken by a VoIP packet traveling across the network depends on a large number of factors, including routing protocols and per network buffering policies. These factors may strongly impact the quality of VoIP.
  • SBLR technique is activated based on report about the actual network QoS measures. These include: packet Loss Rate, Packet Discard Rate -because of jitter and/or large delay, the distribution of lost and discarded packets, packets Round-Trip Delay corresponding to the packet path delay and End System Delay, which represents the delay that the VoIP endpoint adds (because of encoding, decoding and the jitter smoothing buffer).


The following measures to evaluate the performance of the proposed sender-based loss recovery scheme have been used:

  • Mean Opinion Score (MOS)
  • Throughput
  • Delay
  • Jitter
  • Cost of Technique
  • Power Ratio

MOS is a common benchmark introduced by ITU recommendation G.107 for measurement of the subjective quality of human speech, represented as a rating index with a maximum value equals to 5. MOS is derived by taking the average of numerical scores given by juries to rate quality and using it as a quantitative indicator of system performance. SBLR techniques generate more data bits than plain technique. The measure “Cost of Technique” can be defined as to compare the number of bits generated by specific technique to the number of bits generated by plain technique.

Power Ratio P: This measure helps in a collective analysis of different measures we used in this study. It is defined as a sum of functions of scaled performance measures including normalized throughput, normalized MOS, normalized delay, normalized jitter, cost of technique.


This technique works by incorporating an adaptive threshold for redundant transmission of voice audio data. The redundancy depends on two operation factors which are the actual current VoIP network packet loss and a threshold variable parameter set equal to the average network loss rate seen by the receiver. If network current loss rate is greater than the threshold, previous data and new data will be included in single IP packet. Otherwise only new data will be sent. The threshold average is updated whenever a new data segment arrives to the new loss average based on the current measured report of the loss rate seen by the receiver. Figure 1 shows the flow of Adaptive Threshold Redundant Delivery technique.
In mathematical notation;
Let threshold T = average loss. Let L= current loss rate.
For every packet
if L >T send packet j in frames i and i+1, otherwise send packet j in frame i
Update T to a new Average based on current and previous loss rates
Based on the above, the technique requires the knowledge of network loss rate to control sending redundant data. However, this requirement will not introduce extra overhead traffic to network because RTP Control Protocol, commonly used for VoIP calls packets transport, defines and reports a set of performance metrics such as Packet loss and discard rate, delay, and Call Quality (MOS). Thus, the proposed technique can be implemented with no additional changes to existing VoIP System.

Ms. Garima Verma
Assistant Professor
Dept of Information Technology

No comments:

Post a Comment