This paper presents a novel analytical model of Transmission Control Protocol (TCP) using a generalized stochastic Petri net (GSPN). Extensive simulation work has been done for the performance evaluation of TCP NewReno protocol. In view of the limitations of the simulation technique, we present an analytical approach using GSPN. A GSPN is a useful mathematical tool that solves continuous time Markov chains for complex systems and evaluates the stationary behavior. In this paper, we analyze the slow-but-steady variant of TCP NewReno. The model captures the behavioral aspects of the slow start and the congestion avoidance phase together with the fast retransmit and recovery capabilities of TCP NewReno. Performance metrics such as throughput, goodput, and task completion time of the system are obtained. The effect of variation in the model parameters on the performance is studied. The results are validated using the network simulator, and their accuracy is verified by evaluating the confidence interval. The performance of the proposed model is compared with that of TCP Reno. The performance of the proposed model is also compared with one of the previous models. The numerical illustrations and comparison of the proposed technique with simulation validates the accuracy, efficiency, and competence of the GSPN technique. While GSPN modeling for TCP is investigated in depth for the TCP NewReno and TCP Reno variant in this paper, other protocols could be also analyzed similarly. 4186 R. VINAYAK, D. KRISHNASWAMY AND S. DHARMARAJA with TCP Sack. Padhye et al. [8] analyzes the throughput of TCP alone and does not take into consideration the fast recovery mechanism. Padhye et al.[9] presents a model of the TCP Reno. The model captures the essence of TCP's congestion avoidance behavior. The paper analyzes the performance of the protocol in terms of rounds. A simulative approach is followed for the evaluation. However, the paper majorly ignores the effects of fast recovery and fast retransmits. Significantly, the paper concludes that time-outs have a prominent impact on the performance of the protocol. Parves et al.[10] also did the same for TCP NewReno, simulatively. In [11], the slow-but-steady variant of TCP NewReno is analyzed simulatively, using an approach similar to [9]. In [12], the authors develop a continuous time Markov chain model of TCP. However, they ignore the generation of partial acknowledgement (PartACK) and complete acknowledgement (CompACK). They only manipulate the congestion window size, cwnd, and the threshold value, thresh, when the system enters the fast retransmit and recovery (FRR) phase. They ignore the fact that more duplicate acknowledgements (ACKs) arrive, while a segment is in the process of retransmission, and with each incoming duplicate ACK, the window size is incremented by one. The scenario of delayed ACK has not been considered.[13] presents a discrete time Markov chain model for the window size of the TCP and evaluates the throughput and probability of loss for the corresponding window siz...
In this paper, we propose to study the performance of a multi-server queuing system in which a customer requires simultaneous service from a random number of servers with the queuing disciplines retrial and preemptive priority. The infinitesimal generator matrix is presented for the proposed model and steady state measures are discussed. In particular, analytical expressions are obtained for the case where a customer (single type) requires simultaneous service from a random number of servers. Also, the generalised stochastic Petri net (GSPN) is developed for the proposed model and the particular cases. The performance of the models is analysed and compared, in terms of average system size and throughput. It is found that queuing systems in which a customer requires simultaneous service from a random number of servers performs better when the queuing discipline of retrial is appended into the system.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.