In the article we study a model of TCP connection with Active Queue Managementin an intermediate IP router. We use the fluid flow approximation technique to model the interactions between the set of TCP flows and AQM algoithms. Computations for fluid flow approximation model are performed in the CUDA environment.

In the article we study a model of network transmissions with Active Queue Management in an intermediate IP router. We use the OMNET++ discrete event simulator to model the varies variants of the CHOKe algoithms. We model a system where CHOKe, xCHOKe and gCHOKe are the AQM policy. The obtained results shows the behaviour of these algorithms. The paper presents also the implementation of AQM mechanisms in the router based on Linux.

The main idea of all Active Queue Management algorithms, is to notify the TCP sender about incoming congestion by dropping packets, to prevent from the buffer overflow, and its negative consequences. However, most AQM algorithms proposed so far, neglect the impact of the high speed and long delay links. As a result, the algorithms’ efficiency, in terms of throughput and/or queue stability, is usually significantly decreased. The contribution of this paper is twofold. First of all, the performance of the well known AQM algorithms in high speed and long delay scenarios is evaluated and compared. Secondly, a new AQM algorithm is proposed, to improve the throughput in the large delay scenarios and to exclude the usage of random number generator.

In the paper, we investigate queueing system M/G/∞ with non–homogeneous customers. As non–homogenity, we mean that each customer is characterized by some arbitrarily distributed random volume. The arriving customers appear according to a stationary Poisson process. Service time of a customer is proportional to his volume. The system is unreliable what means that all its servers can break simultaneously and then the repair period goes on for random time having an arbitrary distribution. During this period, customers present in the system and arriving to it are not served. Their service continues immediately after repair period termination. Time intervals of the system in good repair mode have an exponential distribution. For such system, we determine steady–state sojourn time and total volume of customers present in it distributions. We also estimate the loss probability for the similar system with limited total volume. An analysis of some special cases and some numerical examples are attached as well.

Queuing regime is one outstanding approach in improving channel aggregation. If well designed and incorporated with carefully selected parameters, it enhances the smooth rollout of fifth/next generation wireless networks. While channel aggregation is the merging of scattered TV white space (spectrum holes) into one usable chunk for secondary users (SU). The queuing regime ensures that these unlicensed users (SUs) traffic/ services are not interrupted permanently (blocked/dropped or forced to terminate) in the event of the licensed users (primary user) arrival. However, SUs are not identical in terms of traffic class and bandwidth consumption hence, they are classified as real time and non-real time SU respectively. Several of these strategies have been studied considering queuing regime with a single feedback queuing discipline. In furtherance to previous proposed work with single feedback queuing regime, this paper proposes, develops and compares channel aggregation policies with two feedback queuing regimes for the different classes of SUs. The investigation aims at identifying the impacts of the twofeedback queuing regime on the performance of the secondary network such that any SU that has not completed its ongoing service are queued in their respective buffers. The performance is evaluated through a simulation framework. The results validate that with a well-designed queuing regime, capacity, access and other indices are improved with significant decrease in blocking and forced termination probabilities respectively.

A wireless sensor system is an essential aspect in many fields. It consists of a great deal of sensor nodes. These sensor networks carry out a number of tasks, including interaction, distribution, recognition, and power supply. Data is transmitted from source to destination and plays an important role. Congestion may occur during data transmission from one node to another and also at cluster head locations. Congestion will arise as a result of either traffic division or resource allocation. Energy will be wasted due to traffic division congestion, which causes packet loss and retransmission of removed packets. As a result, it must be simplified; hence there are a few Wireless sensor networks with various protocols that will handle Congestion Control. The Deterministic Energy Efficient Clustering (DEC) protocol, which is fully based on residual energy and the token bucket method, is being investigated as a way to increase the energy efficiency. In the event of congestion, our proposal provides a way to cope with it and solves it using this method to improve lifespan of the sensor networks. Experiments in simulation show that the proposed strategy can significantly enhance lifetime, energy, throughput, and packet loss.

In the present paper, the model of multi–server queueing system with random volume customers, non–identical (heterogeneous) servers and a sectorized memory buffer has been investigated. In such system, the arriving customers deliver some portions of information of a different type which means that they are additionally characterized by some random volume vector. This multidimensional information is stored in some specific sectors of a limited memory buffer until customer ends his service. In analyzed model, the arrival flow is assumed to be Poissonian, customers’ service times are independent of their volume vectors and exponentially distributed but the service parameters may be different for every server. Obtained results include general formulae for the steady–state number of customers distribution and loss probability. Special cases analysis and some numerical computations are attached as well.

In the present paper, we analyze the model of a single–server queueing system with limited number of waiting positions, random volume customers and unlimited sectorized memory buffer. In such a system, the arriving customer is additionally characterized by a non– negative random volume vector whose indications usually represent the portions of unchanged information of a different type that are located in sectors of unlimited memory space dedicated for them during customer presence in the system. When the server ends the service of a customer, information immediately leaves the buffer, releasing resources of the proper sectors. We assume that in the investigated model, the service time of a customer is dependent on his volume vector characteristics. For such defined model, we obtain a general formula for steady–state joint distribution function of the total volume vector in terms of Laplace-Stieltjes transforms. We also present practical results for some special cases of the model together with formulae for steady–state initial moments of the analyzed random vector, in cases where the memory buffer is composed of at most two sectors. Some numerical computations illustrating obtained theoretical results are attached as well.

The article proposes a model in which Diffusion Approximation is used to analyse the TCP/AQM transmission mechanism in a multinode computer network. In order to prevent traffic congestion, routers implement AQM (Active Queue Management) algorithms. We investigate the influence of using RED-based AQM mechanisms and the fractional controller PI^γ on the transport layer. Additionally, we examine the cases in which the TCP and the UDP flows occur and analyse their mutual influence. Both transport protocols used are independent and work simultaneously. We compare our solution with the Fluid Flow approximation, demonstrating the advantages of Diffusion Approximation.

In the present paper, the most important aspects of computer algebra systems applications in complicated calculations for classical queueing theory models and their novel modifications are discussed. We mainly present huge computational possibilities of Mathematica environment and effective methods of obtaining symbolic results connected with most important performance characteristics of queueing systems. First of all, we investigate effective solutions of computational problems appearing in queueing theory such as: finding final probabilities for Markov chains with a huge number of states, calculating derivatives of complicated rational functions of one or many variables with the use of classical and generalized L’Hospital’s rules, obtaining exact formulae of Stieltjes convolutions, calculating chosen integral transforms used often in the above-mentioned theory and possible applications of generalized density function of random variables and vectors in these computations. Some exemplary calculations for practical models belonging both to classical models and their generalizations are attached as well.

Establishing the proper values of controller parameters is the most important thing to design in active queue management (AQM) for achieving excellent performance in handling network congestion. For example, the first well known AQM, the random early detection (RED) method, has a lack of proper parameter values to perform under most the network conditions. This paper applies a Nelder-Mead simplex method based on the integral of time-weighted absolute error (ITAE) for a proportional integral (PI) controller using active queue management (AQM). A TCP flow and PI AQM system were analyzed with a control theory approach. A numerical optimization algorithm based on the ITAE index was run with Matlab/Simulink tools to find the controller parameters with PI tuned by Hollot (PI) as initial parameter input. Compared with PI and PI tuned by Ustebay (PIU) via experimental simulation in Network Simulator Version 2 (NS2) in five scenario network conditions, our proposed method was more robust. It provided stable performance to handle congestion in a dynamic network.