Loss probability calculations and asymptotic analysis for finite buffer multiplexers

Kim, Han Soo; Shroff, Ness B.

doi:10.1109/90.974529

Cited by 126 publications

(125 citation statements)

References 36 publications

Supporting

Mentioning

113

Contrasting

Unclassified

Order By: Relevance

“…Again, the genetic approach outperforms the EM approach which is in correspondence with the observations for the probability mass functions. It is well known that the decay rate of the probability mass function of the queue content of a queue is a good estimator for the decay rate of the packet loss ratio (as a function of the queue size) [33]. Hence, an accurate assessment of the packet loss ratio requires an accurate match of the probability mass function.…”

Section: Numerical Resultsmentioning

confidence: 99%

A genetic approach to Markovian characterisation of H.264 scalable video

Fiems

Steyaert

Bruneel

2011

Multimed Tools Appl

View full text Add to dashboard Cite

We propose an algorithm for multivariate Markovian characterisation of H.264/SVC scalable video traces at the sub-GoP (Group of Pictures) level. A genetic algorithm yields Markov models with limited state space that accurately capture temporal and inter-layer correlation. Key to our approach is the covariance-based fitness function. In comparison with the classical Expectation Maximisation algorithm, ours is capable of matching the second order statistics more accurately at the cost of less accuracy in matching the histograms of the trace. Moreover, a simulation study shows that our approach outperforms Expectation Maximisation in predicting performance of video streaming in various networking scenarios.

show abstract

Section: Numerical Resultsmentioning

confidence: 99%

A genetic approach to Markovian characterisation of H.264 scalable video

Fiems

Steyaert

Bruneel

2011

Multimed Tools Appl

View full text Add to dashboard Cite

show abstract

“…In recent work by Pihlsgard (2005), the PLR is studied in the very general setting of a GI/G/1/N continuous-time queue. Kim and Shroff (2001) further observe that the curve representing the PLR as a function of the buffer capacity N always has a similar shape as the probability of exceeding the level N in an infinite buffer. Hence, various authors have used tail probabilities of the buffer content distribution of an infinite capacity buffer to approximate the PLR (see e.g.…”

mentioning

confidence: 87%

Packet loss characteristics for M/G/1/N queueing systems

et al. 2008

View full text Add to dashboard Cite

In this contribution we investigate higher-order loss characteristics for M/G/1/N queueing systems. We focus on the lengths of the loss and non-loss periods as well as on the number of arrivals during these periods. For the analysis, we extend the Markovian state of the queueing system with the time and number of admitted arrivals since the instant where the last loss occurred. By combining transform and matrix techniques, expressions for the various moments of these loss characteristics are found. The approach also yields expressions for the loss probability and the conditional loss probability. Some numerical examples then illustrate our results

show abstract

“…It can be observed that the simulated loss probabilities, except for the highest priority class, are smaller but very close to the QoS requirements used for the effective bandwidth calculation. For the highest priority classes, the simulation results are much smaller than the QoS specification because of the finite buffer effect [19].…”

Section: A Single Flowmentioning

confidence: 91%

“…In fact, the resource utilization can be further improved if the finite buffer effect can be included in the buffer optimization and effective bandwidth calculation. A possible approach is to use (7) for PLPs in the buffer partition regions except the last one, and to use the techniques presented in [19] to determine the PLP in the last region. Further research on this topic is necessary.…”

Section: Conclusion and Discussionmentioning

confidence: 99%

“…(a) call level average (CLA), using (19) where is the total number of accepted sources; (b) packet level average (PLA), using (20) where is the minimal effective bandwidth of a type source, and ; (c) bursty-weighted call level average (BWCLA), using (21) where (peak rate/average rate) for a type-source, referred to as bursty coefficient, and . The CLA scheme is based on the intuition that the overall optimal partition vector should be a compromise among the optimal partition vectors for different traffic types.…”

Section: Dynamic Buffer Partitioningmentioning

confidence: 99%

See 1 more Smart Citation

Effective bandwidth of multiclass Markovian traffic sources and admission control with dynamic buffer partitioning

Cheng

Zhuang

Global Telecommunications Conference, 2002. GLOBECOM '02. IEEE

View full text Add to dashboard Cite

Abstract-We investigate the statistical multiplexing and admission control for a partitioned buffer, where the traffic is generated by multiclass Markov-modulated fluid sources. Each of the sources has ( 1) classes at each state. The quality of service (QoS) is described by the packet loss probability for each class. The buffer is partitioned with 1 thresholds to provide the loss priorities. Extending the effective bandwidth concept to such a buffer system is a challenging topic. In this paper, we find the minimal effective bandwidth in the asymptotic regime of large buffers and small loss probabilities by optimally setting the partition thresholds. The minimal effective bandwidth achieves efficient resource utilization and can be used to do admission control for heterogeneous multiclass Markovian sources in an additive way. The buffer partition thresholds are dynamically adjusted according to the input traffic load to guarantee QoS. Numerical analysis and simulation results verify the QoS satisfaction and the obvious improvement of resource utilization compared with previously published results, when the minimal effective bandwidth is used for resource allocation with the proposed dynamic buffer partitioning techniques.Index Terms-Differentiated services (DiffServ), dynamic buffer partitioning, effective bandwidth, loss priorities, multiclass Markovian traffic sources.

show abstract

Loss probability calculations and asymptotic analysis for finite buffer multiplexers

Cited by 126 publications

References 36 publications

A genetic approach to Markovian characterisation of H.264 scalable video

A genetic approach to Markovian characterisation of H.264 scalable video

Packet loss characteristics for M/G/1/N queueing systems

Effective bandwidth of multiclass Markovian traffic sources and admission control with dynamic buffer partitioning

Contact Info

Product

Resources

About