© Science Publications, 2005 Adaptive Data Transmission in Multimedia Networks

This study suggests a method where the packet size of each source is adjusted according to 
the network bandwidth. A controller is used to trace the data transmission rate at the router. An 
algorithm is developed and coded in Tool Command Language. Simulation is performed on NS-2 
using 4 different test cases and the results show that the proposed algorithm avoids router queue 
overflow.


INTRODUCTION
Data loss due to overflow is presently a great challenge to network engineers. Therefore much attention is given to all these factors that contribute to such unwanted scenarios in the network transmission. When the total transmission rate is higher than the network bandwidth, packets are lost either because they are dropped when a router queue overflows or when a client buffer overflows [1,2]. Continuous media applications such as video and audio are sensitive not only to the packet loss probability but also to the correlation of packet losses [1,[4][5][6]. Since the data packet size is directly proportional to the transmission rate, packet size adjustment is recommended to control the traffic flow. Much advancement is made to tolerate the need in current multimedia systems. Self adjusted network transmission has been proposed to optimize the network utilization and also to avoid packet overflow at the client buffer [2,7,9,10]. Webcasting is introduced recently where producers bring real sound and vision to the web. Such broadcasting expects high Quality of Service (QoS) requirements regardless of network model, protocols or traffic agents used in a particular network [17,19]. Much attention is given to high bandwidth requirement and real time delivery constraint. QoS with congestion control mechanism and end to end service guarantee and others have been introduced [11][12][13][14]. To enable satisfactory haptic multimedia data transmission, asynchronous transfer mode (ATM) defines constant bit rate (CBR) to handle traffic at fixed rate and variable bit rate (VBR) service to handle varying traffic [3]. In this study, adaptive transmission rate is proposed using CBR traffic agent. A network model is designed with CBR and user datagram protocol (UDP) traffic agents with a single router. When the total transmission rate is greater than the network bandwidth, after the router, a control mechanism is used to determine the new packet size for each source and the transmission is reinitiated with total transmission rate that is equivalent to the network bandwidth. The simulation results showed that the proposed mechanism avoids router queue overflow.

Adaptive Transmission Rate
Network Model: A network model with 2 sources called node 0 and node 1 and two optional sinks called node 3 and node 4 are connected through a router called node 2. The links between sources and the router are set to be 2 Mbps with 10 ms of delay. The link between the router and the sink n3 is set to be 2 Mbps with 10 ms of delay and the other link to the sink n4 is set to be 3 Mbps with 10 ms of delay. The designed network model is shown in Fig. 1. Concept: Standard sizes of packets are set at the both CBR0 and CBR1. When the transmission begins, both source nodes will begin transmission simultaneously. When the controller notices a possibility of overflow occurrence, it adjusts the packet size to be transmitted by the sources and alert the sources. The sources will reinitiate their transmission with adjusted sizes of packets using CBR3 and CBR4. Pseudo code for rate adjustment for n number source nodes on the network is given as follows. The controller only adjusts the transmission rate when there is a possibility of overflow. D t is the total transmission rate, D s is the transmission rate at the source and BW is the network bandwidth. The Fig. 3 shows the proposed network model.

Adaptive Transmission Data Rate Network:
In this study, a simulation was performed to illustrate how the proposed adaptive transmission data rate mechanism is achieved in multimedia networks [15,16]. Simulation was carried out by first setting the node 3 as the sink and later by setting the node 4 as the sink. Simulation was run by setting the packet size that gives the accumulated data rate lower than the link bandwidth and the next by setting the packet size that gives the accumulated data rate higer than the link bandwidth.  Table 1 shows the test cases conducted on the same network. When n3 is set to be the sink, the network bandwidth is 2 Mbps and when n4 is set to be the sink, the network bandwidth is 3 Mbps. The total transmission rate should not exceed the network bandwidth. If the total transmission rate is higher than the network bandwidth, the transmission controller adjusts the packet size of the source nodes so that the transmission rate is equal to the network bandwidth. The plots show the total transmission rate vs. simulation time in the designed adaptive data transmission network. The plots for all 4 test cases are shown in Fig. 5-8.

Data Loss in a Non-Adaptive Transmission Data Rate
Network: When the first two cases in Table 1 were

CONCLUSION
There is a very high possibility for a router queue overflow to occur when video and audio data are transmitted over a network with low QoS management [8,18]. Router queue overflow occurs when the total data transmission rate is higher than the network bandwidth. This study, proposes an adaptive mechanism, that can avoid router queue overflow. A controller is used to monitor and adjust the data flow from the router to the sink. With such mechanism QoS requirements can be achieved by avoiding the occurrence of data losses.