Discrete-time analysis of the leaky bucket scheme with threshold-based token generation intervals

In ATM networks, source policing functions are crucial for ensuring that user traffic does not exceed the negotiated quality of service. Many different source policing functions have been proposed, including the well known Leaky Bucket scheme [1]. However, recent studies have shown that there are a number of problems in using this scheme. One criticism is that it does not penalize violating traffic sufficiently to reduce its impact on non-violating traffic source [2]. The purpose of this paper is to evaluate modifications to the Leaky Bucket scheme which will enforce the policing function more strictly, and compare with some other proposed techniques such as the Peak Counter scheme [3]. Since analytical analysis is in most cases intractable, simulation is the main tool for this study.

In ATM networks, the time axis is divided into equal intervals called time slots, so most references assume that the Markov chain makes a transition from state i to state j in the boundary of time slots in view of Markov-modulated Bernoulli process (MMBP). This paper considers a more general situation, i.e., the Markov chain makes a transition from state i to state j in the boundary of frames whose length are an integral multiple of length of one time slot. Under this MMBP model, the paper develops a four-dimensional discrete-time Markov chain, and obtains results of the performance analysis of leaky bucket algorithms.

The cell loss suffered by a data stream in an asynchronous transfer mode (ATM) network is studied using its loss function L( mu , beta ). This is the number of cells lost if the stream were allocated a transmission bandwidth of mu cells per second and a buffer of beta cells. Properties of the loss function are derived. Two popular allocation schemes are studied: the fixed-rate-fixed-buffer scheme and the leaky bucket scheme. An example suggests how the loss function might be used to relate pricing of bandwidth and buffer resources to the resulting cell loss. >

We study the problem of dynamic resource allocation of a GPS server with two traffic classes when the leaky bucket scheme is employed as a traffic policing mechanism. Three popular input traffic models - independent Poisson arrival, autoregressive model, and partially observed traffic (hidden Markov model) - are investigated. Theoretically, optimal control can be obtained by a basic dynamic programming algorithm. However, such a solution is computationally prohibitive due to the curse of dimensionality. Instead, we propose several heuristic policies with improvements using rollout, parallel rollout, and hindsight optimization techniques under the aforementioned traffic models and show that these techniques can significantly reduce the penalty associated with delays and dropped packets.

Bandwidth allocation mechanism for LAN/MAN interworking with an ATM network

Forthcoming applications, especially in the multimedia area, require communication services with guaranteed quality of service (QoS). Due to dynamic changes in the network, it is crucial to monitor continuously the achieved QoS during data transmission. This paper focuses on QoS monitoring of ATM connections in end systems as well as inside the ATM network. QM-SSCS, a service specific convergence sublayer for AAL5 is presented which supports QoS monitoring. Moreover, the ATM Network Monitor ANEMON, an entity for collection of topology and QoS-related information within an ATM network, is described.

We present an approach to Service Provisioning with Quality of Service (QoS) requirements in ATM networks. The ATM network provides services to users every $\widehat{T}$ units of time. Each service is characterized by its one way route, its input traffic constraints and its provided QoS. The input traffic constraints consist of the maximum input rate and an upper bound on the burstiness curve. The QoS provided is defined by the maximum percentage of cell loss along the route and the maximum end‐to‐end cell delay. The service provisioning problem is defined as follows: determine the amount, price and required resources (bandwidth and buffers) for each type of service, that maximize a social welfare function that consists of the network’s revenue and the users’ surplus. This problem is solved every $\widehat{T}$ units of time and the allocation is made only over the available resources, that is, those that are not used by connections that are still active. We prove the existence of a solution to the Service Provisioning Problem and we also suggest an iterative procedure that interprets the solution. The network adjusts the prices to maximize the welfare function and also to guarantee that the allocated resources do not exceed the available resources. Based on these prices, users request a new allocation that minimizes their cost and the network adjusts again the prices based on the new allocation. The above procedure has the following features: (i) the network needs to know only the average request rate for each type of service, their route and the resources requested by the users; (ii) users need to know only their private information (input traffic constraints and QoS requirements), their route, and the prices for resources announced by the network.

Stochastic Petri Net-based performance evaluation of hybrid traffic for social networks system

Satisfying Quality of Service (QoS) requirements(e.g. bandwidth and delay constraints) for the different QoS based applications of WSNs raises significant challenges. Each algorithm that is used for packet routing in such applications should be able to establish tradeoffs between end to end delay parameter and energy consumption. Therefore, enabling QoS applications in sensor networks requires energy and QoS awareness in different layers of the protocol stack. In this paper, we propose a Reliable and QoS Aware Multipath routing algorithm in wireless sensor networks namely RQMP. This protocol uses control message together with a Forward Error Correction (FEC) technique to establish a reliable connection between source and destination. RQMP maximizes the network lifetime via data transmission across multiple paths as load balancing that causes energy consume uniformly throughout the network. RQMP uses the residual energy, available buffer size, Signal-to-Noise Ratio (SNR) and distance to sink to predict the best next hop through the paths construction phase. Also our proposed protocol employs a queuing model to handle both real time and non-real-time traffic. Simulation results show that our proposed protocol is more efficient than previous algorithms in providing QoS requirements and minimizing energy consumption.

Uncontrolled variable-bit-rate (VBR) coded video yields consistent picture quality, but the traffic stream is very bursty. When sent over ATM networks, cell losses may be incurred due to limited buffer capacity at the switches; this could cause severe picture quality degradation. Source rate control can be implemented to generate a controlled VBR bit stream which conforms to specified bit rate bounds and buffer constraints. However, source rate control could result in picture quality degradation too. Hence, for real-time video services, an important issue to address is whether the picture quality degradation incurred by source rate control is within acceptable levels or how to choose the appropriate coding parameters to make it so. We establish quantitatively the relationship between picture quality and source rate control for the case of guaranteed service with different combinations of allocated bandwidth, buffer size, and other key video-coding parameters of MPEG-2. In addition, quality control in the context of two-layered scalable video service (basic and enhanced quality) is also considered. Our study reveals that, in order to maximize both the basic and the enhanced quality, source rate control should be implemented on both layers. The relationships between the two types of quality and different combinations of allocated bandwidths, buffer sizes, and some key coding parameters are also established quantitatively for MPEG-2 SNR scalability.

As new computer network technologies emerge, the application designers and the application users expect an increasing level of quality of service from them. Hence, it is a common practice in the newer technologies to more Quality of Service (QoS) components. Until now, these QoS solutions have been both network-technology specific and network-oriented solutions. In this thesis, we present an application-oriented approach to design a high quality network which is independent of the underlying communication technology. In this thesis, we propose a QoS architecture to provide predictable performance to the end-to-end application users in a high quality networking environment. In our architecture, QUANTA (Quality of Service Architecture for Native TCP/IP over ATM networks), we integrate different application requirements and different existing native QoS architectures into a single end-to-end architecture. Through experimentation we identify the architectural issues and the different QoS components required. We propose solutions which include isolation of the applications and managing the knowledge of the applications. The issue of isolating an application is subdivided into classification and identification of the applications. In addressing these issues we propose a ripple-through classification mechanism and a Generic Soft State (GSS) identification mechanism. To manage the knowledge of the applications we propose different QoS components, such as a GSS negotiation mechanism, a GSS communication mechanism and a GSS monitoring mechanism (such as GSS Relays and GSS Agents). QUANTA's overhead is measured by running applications with different life-times and QoS requirements with and without QUANTA. For a transaction-oriented applications, the overhead induced by using QUANTA is larger than the benefit of using QUANTA. In high data rate applications and in long life time applications, a predictable performance to the applications is achieved with very low overhead by QUANTA. We demonstrate that QUANTA can manage and maintain Quality of Service for different classes of applications under varying host and network load conditions transparent to the application user. Using QUANTA, we can reach nearly 80% of the channel utilization under loaded conditions, whereas without QUANTA, the load on the network can reduce the channel utilization to 40%. Quanta has reduced a 350 msec delay under loaded conditions to less than a 10 msec delay. With the exception of short transition periods, QUANTA can sustain throughput to within the bounds of the specification of the user. We identify the limitations of QUANTA as currently proposed and discuss possible enhancements to it and other such architectures to remedy these limitations.

Internet Quality of Service (QoS) is an actual research topic. The need for more or less guaranteed transmission rates (e.g. for bulk-data transfer applications), upper bounds for transmission delay and jitter (e.g. for real time applications like IP telephony, video-conferencing) stimulated the development of QoS transmission technologies like ATM, IP-RSVP and the actual discussion about Diff-Serv mechanisms. This paper describes the implementation and the first measurement trials with the One-Way-Delay measurement components of CM-Toolset. The overall tool architecture consists of a distributed system of load generators and receivers (agents) and a measurement server, which stores the measurement results in a data-base. To measure the exact One-Way-Delay (OWD) the synchronised GPS clock system was integrated. The measurements were realised in a 4 hop heterogeneous IP and ATM network. The clock drift during the loop measurements was about 10 ms, linear and deterministic. So the drift could be eliminated by statistical computations. The measurements were made for UDP flows under different load situations( 1 UDP....5 UDP, TSDU size, interarrival time), and for n*UDP+1 TCP. The measurements show very clearly the interaction and dependencies between the number of flows, their load parameters and the QoS parameters like packet loss and one way delay. The measurements of UDP+TCP multiplexed flows explains, why the concept of flow separation is important for QoS guarantees. CM Toolset can be used by protocol engineers for the development of new QoS-based network technologies by collecting measurement data for modelling the network behaviour. Network managers can tune the operational network to improve the quality e.g. of IP-Telephon.

The statistics of cell loss at both burst and cell levels are studied by simulation based on the configuration of the Singapore One ATM network. A virtual connection through the Singapore One network is modelled as a tandem queueing network to which mixtures of homogeneous discrete-time burst–silence traffic with different burst lengths are offered. Attention is focused on a designated traffic stream as it goes through the tandem queueing network. In the simulation, the authors have not introduced the ‘independence assumption’ between the adjacent queues, i.e. the actual tandem queueing network is used to obtain the performance measures for the designated traffic stream. Cell loss probability and high-order cell loss distributions such as cell loss-free run distribution and cell loss pattern distribution are obtained. These results provide valuable information for the estimation of end-to-end overall quality of service (QoS), design of the cell policing mechanism and flow control schemes in the ATM networks.

We discuss the call admission control problem in ATM networks. In particular, we focus on a single node in isolation as a model for the ATM multiplexer, and consider the cell loss probability and the average cell queueing delay in the multiplexer as the quality of service (QoS) performance measures. Difficulties often arise in the analysis for such systems due to the so-called curse of dimensionality in calculating some integer-parametered functions, where the integer parameter represents the system size or dimension. Our basic idea is to develop global rational approximation for such systems. Various examples presented in the previous papers have demonstrated the power of the approach. In this paper we apply the rational approximation method to the call admission problem in ATM networks.

In ATM networks the virtual channel identifier (VCI) for a connection is used at switches to route cells belonging to the connection along the correct channel. The VCI is also used to provide the appropriate resources for the connection and to check that the connection does not exceed the resources allocated to it. The resource management functions in an ATM network are responsible both for allocation of resources and for guarding the network against badly behaved users. In addition they allow the network to decide whether it can enter into new connection “contracts” with its users and guarantee their required quality of service (QoS). The article highlights some of the resource management issues in ATM networks. (5 pages)