Mine Caglar

A Long-Range Dependent Workload Model for Packet Data Traffic

We consider a probabilistic model for workload input into a telecommunication system. It captures the dynamics of packet generation in data traffic as well as accounting for long-range dependence and self-similarity exhibited by real traces. The workload is found by aggregating the number of packets, or their sizes, generated by the arriving sessions. The arrival time, duration, and packet-generation process of a session are all governed by a Poisson random measure. We consider Pareto-distributed session holding times where the packets are generated according to a compound Poisson process. For this particular model, we show that the workload process is long-range dependent and fractional Brownian motion is obtained as a heavy-traffic limit. This yields a fast synthesis algorithm for generating packet data traffic as well as approximating fractional Brownian motion.

Multiclass G/M/1 queueing system with self-similar input and non-preemptive priority

The Internet domains are tied together by service level agreements which are based on various QoS parameters such as delay, jitter, packet-loss rate, throughput and availability. To offer tighter and more comprehensive service level agreements, accurate modeling of IP traffic and its queuing behavior over the entire network domain is necessary. We present a novel analytical model of a single router which takes into account multiple classes of self-similar traffic based on G=M=1 queueing system with non-preemptive priority. Our long-range dependent traffic model is generated by infinitely many sources governed by a Poisson random measure. We derive exact expressions for the transition probabilities of the embedded Markov chain of G=M=1 by first deriving the interarrival distribution of the incoming traffic. Closed form expressions for the expected waiting time of multiple classes have been derived. The queuing system is evaluated numerically for a typical router to show the effect of the Hurst parameter on several performance measures. Such work forms a basis for modeling the behavior of self-similar traffic accurately through heterogenous network domains, eventually leading to the formation of realistic service level agreements. (Less)

Towards the formation of comprehensive SLAs between heterogeneous wireless DiffServ domains

Traffic patterns generated by multimedia services are different from traditional Poisson traffic. It has been shown in numerous studies that multimedia network traffic exhibits self-similarity and burstiness over a large range of time-scales. The area of wireless IP traffic modeling for the purpose of providing assured QoS to the end-user is still immature and the majority of existing work is based on characterization of wireless IP traffic without any coupling of the behaviour of queueing systems under such traffic conditions. Work in this area has either been limited to simplified models of FIFO queueing systems which do not accurately reflect likely queueing system implementations or the results have been limited to simplified numerical analysis studies. In this paper, we advance the knowledge of queueing systems by example of traffic engineering of different UMTS service classes. Specifically, we examine QoS mapping using three common queueing disciplines; Priority Queuing (PQ), Low Latency Queuing (LLQ) and Custom Queueing (CQ), which are likely to be used in future all-IP based packet transport networks. The present study is based on a long-range dependent traffic model, which is second order self-similar. We consider three different classes of self-similar traffic fed into a G/M/1 queueing system and construct analytical models on the basis of non-preemptive priority, low-latency queueing and custom queueing respectively. In each case, expressions are derived for the expected waiting times and packet loss rates of different traffic classes. We have developed a comprehensive discrete-event simulator for a G/M/1 queueing system in order to understand and evaluate the QoS behaviour of self-similar traffic and carried out performance evaluations of multiple classes of input traffic in terms of expected queue length, packet delay and packet loss rate. Furthermore, we have developed a traffic generator based on the self-similar traffic model and fed the generated traffic through a CISCO router-based test bed. The results obtained from the three different queueing schemes (PQ, CQ and LLQ) are then compared with the simulation results in order to validate our analytical models.

An analytical model based on G/M/1 with self-similar input to provide end-to-end QoS in 3G networks

The dramatic increase in demand for wireless Internet access has lead to the introduction of new wireless architectures and systems including 3G, Wi-Fi and WiMAX. 3G systems such as UMTS and CDMA2000 are leaning towards an all-IP architecture for transporting IP multimedia services, mainly due to its scalability and promising capability of inter-working heterogeneous wireless access networks. During the last ten years, substantial work has been done to understand the nature of wired IP traffic and it has been proven that IP traffic exhibits self-similar properties and burstiness over a large range of time scales. Recently, because of the large deployment of new wireless architectures, researchers have focused their attention towards understanding the nature of traffic carried by different wireless architecture and early studies have shown that wireless data traffic also exhibits strong long-range dependency. Thus, the classical tele-traffic theory based on a simple Markovian process cannot be used to evaluate the performance of wireless networks. Unfortunately, the area of understanding and modeling of different kinds of wireless traffic is still immature which constitutes a problem since it is crucial to guarantee tight bound QoS parameters to heterogeneous end users of the mobile Internet. In this paper, we make several contributions to the accurate modeling of wireless IP traffic by presenting a novel analytical model that takes into account four different classes of self-similar traffic. The model consists of four queues and is based on a G/M/1 queueing system. We analyze it on the basis of priority with no preemption and find exact packet delays. To date, no closed form expressions have been presented for G/M/1 with priority.

An analytical framework for self-organizing peer-to-peer anti-entropy algorithms

An analytical framework is developed for establishing exact performance measures for peer-to-peer (P2P) anti-entropy paradigms used in biologically inspired epidemic data dissemination. Major benefits of these paradigms are that they are fully distributed, self-organizing, utilize local data only via pair-wise interactions, and provide eventual consistency, reliability and scalability. We derive exact expressions for infection probabilities through elaborated counting techniques on a digraph. Considering the first passage times of a Markov chain based on these probabilities, we find the expected message delay experienced by each peer and its overall mean as a function of initial number of infectious peers. Further delay and overhead analysis is given through simulations and the analytical framework. The number of contacted peers at each round of the anti-entropy approach is an important parameter for both delay and overhead. These exact performance measures and theoretical results would be beneficial when utilizing the models in several P2P distributed system and network services such as replicated servers, multicast protocols, loss recovery, failure detection and group membership management.

Principles and performance analysis of SeCond: A system for epidemic peer-to-peer content distribution

We propose and design a peer-to-peer system, SeCond, addressing the distribution of large sized content to a large number of end systems in an efficient manner. In contrast to prior work, it employs a self-organizing epidemic dissemination scheme for state propagation of available blocks and initiation of block transmissions. In order to exploit heterogeneity of peers, enhance the utilization of system resources and for the ease of deployment, scalability, and adaptivity to dynamic peer arrivals/departures, we propose mechanisms for adjusting protocol parameters dynamically according to the bandwidth usages. We describe design and analysis details of our protocol SeCond. Comprehensive performance evaluations and comparison with the BitTorrent system model have been accomplished for a wide range of scenarios. Performance results include scalability analysis for different arrival/departure patterns, flash-crowd scenario, overhead analysis, and fairness ratio. The major metrics we study include the average file download time, load on the primary seed, uplink/downlink utilization, and communication overhead. We show that SeCond is a scalable and adaptive protocol which takes the heterogeneity of the peers into account. The protocol is as fair as BitTorrent although it has no explicit strategy addressing free-riding. We also illustrate the applicability of an analytical fluid model to the behavior of SeCond.

Traffic engineering and QoS control between wireless diffserv domains using PQ and LLQ

Numerous recent studies have proven that traffic patterns generated by multimedia services are different from traditional Poisson traffic. It has been shown that multimedia network traffic exhibits long-range dependency (LRD) and self-similar characteristics. The area of wireless IP traffic modeling in terms of providing assured QoS to the end-user is still immature and the majority of existing work is merely based on characterization of the wireless IP traffic without investigating the behavior of queueing systems under such traffic conditions. Work done in this area has been limited to simplified models of FIFO queueing systems which do not accurately reflect likely queueing system implementations or the results have been limited to simplified numerical analysis studies. In this paper, we contribute towards the solution of this problem by focusing on traffic engineering of different UMTS service classes by providing efficient QoS mapping using two common queueing disciplines; Priority Queueing (PQ) and Low Latency Queueing (LLQ), which are likely to be used in future all-IP based packet transport networks.The present study is based on a realistic traffic model which is long-range dependent and self-similar. We consider three different classes of self-similar traffic being fed to a queueing model of three queues based on a G/M/1 queueing system. We first make an analysis on the basis of non-preemptive priority and then on the basis of low-latency queuing and find closed form expressions of expected waiting times and packet loss rates of different traffic classes. In order to validate our models we conduct a series of experiments. We have developed a comprehensive discrete-event simulator for a G/M/1 queueing system in order to understand and evaluate the QoS behavior of self-similar traffic and carry out performance evaluations of multiple classes of input traffic in terms of expected queue length, packet delay and packet loss rate. Furthermore, we have developed a traffic generator to realize our self-similar traffic model and use it to feed traffic through a CISCO router-based test bed. The results obtained from the two different queueing schemes (PQ and LLQ) are then compared with the simulation results to ascertain the accuracy of our model.

Lyapunov exponents of Poisson shot-noise velocity fields

We consider the Lyapunov exponents of flows generated by a class of Markovian velocity fields. The existence of the exponents is obtained for flows on a compact set, but with the most general form of the velocity field. As a particular class, we study the homogeneous and incompressible flows. In this case, the exponents are nonrandom, free of the initial position of the particle path, and their sum is zero. We numerically compute the top Lyapunov exponent on for a range of parameters to conjecture that it is strictly positive.

Stepwise fair-share buffering for gossip-based peer-to-peer data dissemination

We consider buffer management in support of large-scale gossip-based peer-to-peer data dissemination protocols. Coupled with an efficient buffering mechanism, system-wide buffer usage can be optimized while providing reliability and scalability in such protocols. We propose a novel approach, stepwise fair-share buffering, that provides uniform load distribution and reduces the overall buffer usage where every peer has a partial view of the system. We report and discuss the comparative performance results with existing buffering approaches as well as random buffering which serves as a benchmark. We present separate evaluations of bufferer selection and gossip-based data dissemination. Reliability, content dissemination time, message delay, buffering delay, and minimum buffer requirements are considered as the key metrics investigated through simulations. The performance of our approach in the case of multiple senders, link failures with multiple bufferers, and scalability to larger networks are investigated. Several power-law and hierarchical overlay topologies are considered. Analytical bounds for reliability of dissemination are also provided.

Stepwise Fair-Share Buffering underneath Bio-inspired P2P Data Dissemination

We consider buffer management problem in support of large-scale bio-inspired peer-to-peer data dissemination services. Bio-inspired epidemic protocols have considerable benefits as they are robust against network failures, scalable and provide probabilistic reliability guarantees. Coupled with an efficient buffering mechanism, system wide buffer usage can be optimized while providing reliability and scalability in such protocols. We propose a novel algorithm, Stepwise Fair-share Buffering, that provides uniform load distribution in comparison to earlier approaches and reduces the overall buffer usage where every peer has the partial view of the system. We report and discuss the comparative performance results and provide an analytical evaluation of our approach.