General operation of the simulation of speech background traffic generation in

Figure 3.6 depicts how the background traffic generator node is integrated into a simulation as well as the most relevant inputs. The background traffic generator node will receive three types of input to operate, they are: (1) attributes defined by the modeller in the simulation setup stage, (2) attributes embedded in the simulation and (3) feedback parameters obtained from the simulation. The shape of the traffic generated in this node as well as the destinations where such traffic will be sent are to be defined by these three set of inputs. Appendix A presents a detailed description of the background traffic generation node implemented in OPNET.

In Figure 3.7, a flow diagram describes the different stages involved in generating speech background traffic for the three groups of encoding algorithms. More detailed information regarding the process of calculating the frame size for each group will be discussed below.

Speech background traffic generation for fixed data rate codecs

1. The first step of the simulation is the definition of the simulation attributes by the modeller. The following parameters are defined at this stage:

1.1 In the background traffic generator node, one of the following codec/modes is selected: G.729, G.711, iLBC-20ms or iLBC-30ms. The selection of the codec will univocally define the encoded frame size and frame duration (see Table 3.1). 1.2 In the background traffic generator node, the IP address of the destination node and

the number of speech frames encapsulated per packet are defined (see Section 3.1.2 and Appendix A for more details).

Figure 3.6 Operation of Voice background traffic generator

1.3 In the background traffic generator node, the number of conversations per codec per destination is entered. Along with some timing settings, the number of conversations attribute defines the traffic scalability for each codec-destination. 2. Using the codec type attribute defined in the simulation setup stage and information

from Table 3.1, the frame size and frame duration for the selected codec are obtained. 3. Using the number of conversations attributes, separate independent streams are created,

each one representing a conversation.

4. For each conversation, using the number of voice frames per packet attribute and based on equation 3.10, the voice payload size and application layer packet size are calculated.

5. The application layer packet is sent down to all other layers and finally transmitted through the simulated network. Each packet is potentially exposed to latency, jitter and packet loss creating the simulation speech background traffic.

Figure 3.7 Traffic generation chart for each of the encoding algorithm groups encompassing all inputs and attributes

Speech background traffic generation for AMR-NB codec

1. The first step of the simulation is the definition of the simulation attributes by the modeller. The following parameters are defined at this stage:

1.1 In the background traffic generator node, AMR-NB codec is selected. The selection of the codec will define the frame duration of 20ms.

1.2 In the background traffic generator node, the IP address of the destination node and the number of speech frames encapsulated per packet are defined (see Section 3.1.2 and Appendix A for more details).

1.3 In the background traffic generator node, the number of conversations per codec per destination is entered. Along with some timing settings, the number of conversations attribute, defines the traffic scalability for each codec-destination. 1.4 In the client node, the RTCP generation frequency is defined in terms of number of

RTP packets received before one RTCP packet is sent back to the background traffic generator node (see Section 3.1.2 and Appendix A for more details).

1.5 In the client node, it is defined if the delay information to be sent back to the background traffic generator node is that measured from the last received RTP packet or the average of the last group of RTP packets received since the previous RTCP packet was sent.

2. Initially, since no information regarding the delay associated with the network path has been obtained, AMR-NB will start transmitting in the highest mode (12.2 kbps). 3. The client node, after receiving the number of AMR-NB frames defined in step 1.5,

will generate a control packet that will be sent back to the real voice traffic generator node. The RTCP packet will offer the real voice traffic generator node updated information regarding the delay on the path connecting the two nodes.

4. At the reception of the RTCP packet from the Client node, the background traffic generator node updates the M2E parameter based on the E2E delay obtained from the RTCP packet and using equation 3.5.

5. Using Table 3.2 and the M2E value obtained in step 4, the frame size for the subsequent frames is computed. At the reception of the next RTCP packet the process of adjusting the frame size will be repeated. During the interval between RTCP packets reception, the frame size is maintained Using the number of voice frames per packet attribute and based on equation 3.10, the voice payload size and application layer packet size are calculated.

6. The application layer packet is sent down to all other layers and finally transmitted through the simulated network. Each packet is potentially exposed to latency, jitter and packet loss creating the simulation speech background traffic.

Speech background traffic generation for Speex codec

1. The first step of the simulation is the definition of the simulation attributes by the modeller. The following parameters are defined at this stage:

1.1 In the background traffic generator node, Speex codec is selected (one of the two modes available). The selection of the codec will define the frame duration of 20ms.

1.2 In the background traffic generator node, the IP address of the destination node and the number of speech frames encapsulated per packet are defined (see Section 3.1.2 and Appendix A for more details).

1.3 In the background traffic generator node, the number of conversations per codec per destination is entered. Along with some timing settings, the number of conversations attribute, defines the traffic scalability for each codec-destination. 2. According to the parameter entered in step 1.4, a number of traffic streams are

generated.

3. For each conversation independently, a number of frame sizes equal to the parameter defined in step 1.3 are calculated using the methodology discussed in 3.1.1. Within each conversation, the frame size variable maintains its autocorrelation even when frames are encapsulated in different packets. Analogously, the frame size variable is independent between conversations.

4. Using the number of frames per packet attribute defined in the simulation setup stage and equation 3.10, the voice payload size and the application layer packet size are obtained.

6. The application layer packet is sent down to all other layers and finally transmitted through the simulated network. Each packet is potentially exposed to latency, jitter and packet loss creating the simulation speech background traffic.