International Journal of Computer Applications (0975 – 8887) Volume 16– No.8, February 2011 1 Performance Evaluation of a Communication Network with Dynamic Bandwidth Allocation and Bulk Arrivals P.Srinivasa Rao Dept. of CS & SE, AUCE ANDHRA UNIVERSITY Visakhapatnam, INDIA Kuda Nageswara Rao Dept. of CS & SE, AUCE ANDHRA UNIVERSITY Visakhapatnam, INDIA K.Srinivas Rao Dept. of Statistics ANDHRA UNIVERSITY Visakhapatnam, INDIA ABSTRACT Communication networks had a separate deployment for each emerging service like Telephone networks, data networks and multimedia networks. In Integrated Services Network the packets are transmitted using efficient statistical multiplexing. In this paper, we develop and analyze a two-node communication network with dynamic bandwidth allocation having bulk arrivals. The performance of the statistical multiplexing is measured by approximating the arrival and the service process with Poisson process and the bulk size is uniform which is chosen such that several of the statistical characteristics of the communication network identically match with the Poisson process. Through mathematical modeling, the performance measures of the communication network like the joint probability generating function of the buffer size distribution, the average content of the buffers, the mean delays in transmission, the throughput of the nodes and utilization are derived explicitly under transient conditions. Through numerical studies, the sensitivity of the input parameters on the performance measures is also carried. It is observed that the dynamic bandwidth allocation strategy and bulk size distribution of the arrivals have significant effect on the performance measures. This network is much useful in satellite communications and Internet service providing, etc, Keywords Communication networks, Dynamic bandwidth allocation, Bulk arrivals, Performance measures 1. INTRODUCTION Much work has been reported regarding communication networks and their performance evaluation. Martin Reiser (1982) and Jaime Jungok Bae(1991) have reviewed the communication networks and analytical methods for their evaluation. Several authors developed various communication network models with several considerations in order to analyze the situation close to the reality. One of the important considerations in communication network model is transporting data/voice more effectively with a guaranteed Quality of Service (QoS). For efficient communication, different service models have been proposed a scalable traffic management mechanism to ensure QoS. However, in some situations at broadband integrated services, the digital network has a synchronized transmission mode. The output of one transmitter is usually the input of another transmitter. Due to the unpredicted nature of the transmission lines, congestion occurs in communication systems. In order to analyze the communication network efficiently, one has to consider the analogy between communication networks and waiting line models. Generally, the analysis in a communication system is mainly concerned with the problem of allocation and distribution of data or voice packetization, statistical multiplexing, flow control, bit dropping, link assignment, delay and routing etc. For efficient utilization of the resources, mathematical modeling provides the basic frame work in communication networks. The communication networks are modeled as interconnected queues by viewing the message as the customer, communication buffer as waiting line and all activities necessary for transmission of the message as service. This representation is the most natural with respect to the actual operation of such systems. This leads a communication network to view as a tandem or serial queuing network. Several authors have studied the communication networks as tandem queues (Kleinrock L. 1976; Yukuo hayshida, 1993; Paul Dupis et al, 2007). Because of the unpredicted nature of demand at transmission lines congestion occurs in communication systems. Statistical multiplexing is one of the major considerations for efficient utilization of the resources. With the statistical multiplexing load dependent communication network models have been generated to accommodate the bit dropping methodologies (Kin K. Leung, 2002). Bit dropping method can be classified as IBD (Input Bit Dropping) and OBD (Output Bit Dropping). Depending on the implementation of the actual algorithms, IBD or OBD performance is measured. As a result of the bit dropping or flow control strategies voice quality is expected to degrade gracefully when overload occurs. The extent of degradation of service quality is a function of the fraction of voice calls lost, which in turn depends on the load. To have an efficient transmission with high quality, it is needed to consider the variation on transmission rates based on the contents of the buffers. This is often referred as dynamic bandwidth allocation. Some algorithms have been developed with various protocols and allocation strategies for optimal utilization of bandwidth (Emre and Ezhan, 2008; Gundale and Yardi, 2008; Hongwang and Yufan, 2009; Fen Zhou et al. 2009; Stanislav, 2009). These strategies are developed based on arrival process of the packets through bit dropping and flow control techniques. It is needed to utilize the bandwidth maximum possible by developing strategies of transmission control based on buffer size. One such strategy is dynamic bandwidth allocation. In dynamic bandwidth allocation, the transmission rate of the packet is adjusted instantaneously depending upon the content of the buffer. Recently P.Suresh Varma et al (2007) has developed some communication network models using dynamic bandwidth allocation. However, they considered that the arrivals of packets to the buffer are single. But, in store-and-forward communication the messages are packetized and transmitted. When a message is packetized, the number of packets of that message is random having bulk in size. Hence, considering single packet arrival to the initial node may not accurately evaluate the performance of the communication network. Therefore, in this paper, a communication network with dynamic bandwidth allocation having bulk arrivals is developed