pith. sign in

arxiv: 1907.07645 · v1 · pith:BMFNCB4Unew · submitted 2019-07-17 · 💻 cs.NI · cs.MM· eess.IV

The Statistical Analysis of the Live TV Bit Rate

Iskandar Aripov This is my paper

Pith reviewed 2026-05-24 19:52 UTC · model grok-4.3

classification 💻 cs.NI cs.MMeess.IV
keywords live TV streamingvariable bit rategeneralized extreme value distributiont-location-scale distributionBayesian information criterionqueuing modelPID allocationstreaming server scheduling
0
0 comments X

The pith

Live TV bit rate data from 13 channels fits the generalized extreme value distribution best for most streams and t-location-scale for the full aggregate.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper examines bit rate allocation records from 13 live TV channels to identify which probability distributions most accurately capture the variable rates that streaming systems must handle. It tests twenty candidate distributions against the measured data and selects the best match for each channel and for the combined system by means of the Bayesian information criterion. The resulting models are offered as ready-to-use parameters for queuing analysis inside streaming servers. A reader would care because these distributions directly affect how servers decide when to schedule packets and how much bandwidth to reserve, which in turn influences delay, loss, and the cost of network resources in content delivery.

Core claim

The statistical analysis of PID allocation data shows that the generalized extreme value distribution supplies the most precise fit for the majority of the thirteen individual TV channels according to the Bayesian information criterion, while the t-location-scale distribution supplies the best fit when all channels are treated as a single system; these fitted distributions are then used to supply concrete parameters for a streaming server queuing model.

What carries the argument

Maximum-likelihood fitting of twenty probability distributions to channel bit-rate time series followed by Bayesian information criterion ranking to select the generalized extreme value distribution per channel and t-location-scale distribution for the aggregate.

If this is right

  • Queuing models for streaming servers can be parameterized directly with the reported shape, scale, and location values.
  • Scheduling policies can be tuned to the tail behavior captured by the generalized extreme value distribution.
  • Dynamic routing decisions at content delivery networks can use the aggregate t-location-scale model to anticipate total bandwidth demand.
  • Resource allocation in software-defined networks can be driven by the per-channel and system-wide distribution parameters.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same fitting procedure could be applied to short measurement windows to detect when a channel's rate statistics have shifted.
  • Capacity planning tools for edge caches might incorporate the reported distributions to size buffers for live events.
  • The approach supplies a concrete way to compare traffic models across different video codecs or resolutions without re-deriving the entire model.

Load-bearing premise

The bit rate measurements obtained from 13 TV channels via PID allocation data are representative of the statistical behavior of live TV streaming traffic in general.

What would settle it

Repeating the same distribution-fitting exercise on bit-rate traces from a different collection of live TV channels and obtaining a different top-ranked distribution under the Bayesian information criterion would falsify the reported best fits.

Figures

Figures reproduced from arXiv: 1907.07645 by Iskandar Aripov.

Figure 1
Figure 1. Figure 1: The bandwidth allocated by the multiplexer for the selected culture channel during [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The Probability density function of the histogram of the selec [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The bandwidth allocated by the multiplexer to every channel. Y [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The PDF of the histogram for all the channels and best [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: All channels synchronized bandwidth allocation [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The bandwidth allocated by the multiplexer to 13 channels synchronized during 23 minutes. The Y-axis is in Kbps. Now computing the histogram of our multiplexor in [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: The PDF of the multiplexer’s bandwidth allocation. The X [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: [2]: Multiplexer controls channels bandwidth allocation [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗
Figure 8.2
Figure 8.2. Figure 8.2: Streaming server shows channels to differe [PITH_FULL_IMAGE:figures/full_fig_p009_8_2.png] view at source ↗
read the original abstract

This paper studies the statistical nature of TV channels streaming variable bit rate distribution and allocation. The goal of the paper is to derive the best-fit rate distribution to describe TV streaming bandwidth allocation, which can reveal traffic demands of users. Our analysis uses multiplexers channel bandwidth allocation (PID) data of 13 TV live channels. We apply 17 continuous and 3 discrete distributions to determine the best-fit distribution function for each individual channel and for the whole set of channels. We found that the generalized extreme distribution fitting most of our channels most precisely according to the Bayesian information criterion. By the same criterion tlocationscale distribution matches best for the whole system. We use these results to propose parameters for streaming server queuing model. Results are useful for streaming servers scheduling policy design process targeting to improve limited infrastructural resources, traffic engineering through dynamic routing at CDN, SDN.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper analyzes multiplexed PID allocation data from 13 live TV channels, fits 20 distributions (17 continuous, 3 discrete), and uses the Bayesian information criterion (BIC) to identify the generalized extreme value distribution as the best fit for most individual channels and the tlocationscale distribution for the aggregate. These fitted distributions are then proposed as parameters for a streaming-server queuing model to support scheduling, traffic engineering, and CDN/SDN design.

Significance. If the PID measurements capture content-driven bit-rate variability rather than allocation policy, the BIC-based ranking supplies concrete, reusable parameters for live-TV traffic models. The systematic comparison across 20 distributions and explicit use of BIC for selection is a methodological strength that could aid reproducible queuing analysis in network information literature.

major comments (2)
  1. [Abstract and implied data/results sections] Data collection and preprocessing (implicit in the abstract and results sections): no sample counts per channel, observation window, handling of zero-rate intervals, or comparison to direct streaming traces are reported. Without these, it is impossible to determine whether the BIC-selected GEV and tlocationscale fits describe intrinsic VBR statistics or merely the static/policy-driven PID allocations performed by the multiplexer.
  2. [Results and conclusion] Use of fitted distributions for queuing-model parameterization (results and conclusion): the manuscript states that the GEV and tlocationscale parameters can be used directly for server scheduling, yet provides no derivation or mapping from the fitted shape/scale/location values to the queuing-model inputs (e.g., service-time moments or arrival-process parameters). This step is load-bearing for the claimed utility in traffic engineering.
minor comments (2)
  1. [Abstract] The abstract lists “17 continuous and 3 discrete distributions” but does not name the full set or the software used for maximum-likelihood fitting and BIC computation; adding this list and a reproducibility note would strengthen the work.
  2. [Results] Notation for the tlocationscale distribution is non-standard in some statistical packages; a brief definition or reference to its location-scale-t form would remove ambiguity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and will revise the paper accordingly to improve clarity and completeness.

read point-by-point responses

  1. Referee: [Abstract and implied data/results sections] Data collection and preprocessing (implicit in the abstract and results sections): no sample counts per channel, observation window, handling of zero-rate intervals, or comparison to direct streaming traces are reported. Without these, it is impossible to determine whether the BIC-selected GEV and tlocationscale fits describe intrinsic VBR statistics or merely the static/policy-driven PID allocations performed by the multiplexer.

    Authors: We agree that these details are essential for readers to evaluate the data. In the revised manuscript we will add a new subsection on data collection that reports the exact sample counts per channel, the observation window length and sampling rate, the handling of zero-rate intervals (excluded as they indicate no active transmission), and a short discussion of how the PID allocation data from live channels reflects content-driven bit-rate variability rather than purely static policy. revision: yes

  2. Referee: [Results and conclusion] Use of fitted distributions for queuing-model parameterization (results and conclusion): the manuscript states that the GEV and tlocationscale parameters can be used directly for server scheduling, yet provides no derivation or mapping from the fitted shape/scale/location values to the queuing-model inputs (e.g., service-time moments or arrival-process parameters). This step is load-bearing for the claimed utility in traffic engineering.

    Authors: The referee is correct that an explicit mapping is missing. We will revise the results and conclusion sections to include the required derivations: specifically, closed-form expressions for the mean and variance of the generalized extreme value and tlocationscale distributions, which can be directly substituted into standard queuing models (e.g., M/G/1 service-time moments) for scheduling and traffic-engineering calculations. revision: yes

Circularity Check

0 steps flagged

Empirical BIC-based distribution fitting to observed PID data is self-contained; no reduction to inputs by construction

full rationale

The paper's core procedure is to collect bit-rate allocation samples from 13 channels, test 20 candidate distributions, and rank them by BIC; the reported GEV and tlocationscale selections are direct outputs of that comparison on external data. No equation defines a quantity in terms of the eventual best-fit parameters and then re-derives it, no self-citation supplies a uniqueness result, and the queuing-model parameters are simply the fitted values themselves rather than a claimed prediction. The analysis therefore contains no load-bearing step that collapses to its own inputs.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The central claim rests on the empirical fitting process, the representativeness of the 13-channel PID dataset, and the appropriateness of BIC for model selection in this domain.

free parameters (2)
  • parameters of generalized extreme value distribution
    Fitted separately to each of the 13 channel bit-rate traces.
  • parameters of tlocationscale distribution
    Fitted to the pooled data across all channels.
axioms (1)
  • domain assumption Live TV bit-rate traces can be adequately described by one of the 20 tested continuous or discrete distributions
    Invoked by applying the library of distributions without additional justification for their suitability to video traffic.

pith-pipeline@v0.9.0 · 5668 in / 1228 out tokens · 26346 ms · 2026-05-24T19:52:39.467528+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

18 extracted references · 18 canonical work pages

  1. [1]

    Fit All Valid Parametric Probability Distributions to Data

    Sheppard, Mike. “Fit All Valid Parametric Probability Distributions to Data.” MathWorks File Exchange, 1.4, MathWorks, 4 Apr. 2012

    work page 2012

  2. [2]

    Reflex™ by Ericsson, Remote Reflex™ by Ericsson

    Ericsson “Reflex™ by Ericsson, Remote Reflex™ by Ericsson” Statistical Multiplexing datasheet, Aug 2011

    work page 2011

  3. [3]

    Reflex Configuration Guidlines

    Ericsson “Reflex Configuration Guidlines” System Release R2C-FD3 datasheet, 5 Aug 2011

    work page 2011

  4. [4]

    IEEE J.Selected Areas in Commun.,13,1028-1038

    Tse,D.N.C.,Gallager,R.E.and Tsitsiklis,J.N.(1995).Statistical multiplexing of multiple time-scale Markov streams. IEEE J.Selected Areas in Commun.,13,1028-1038

    work page 1995

  5. [5]

    Statistical Multiplexing and Mix-Dependent Alternative Routing in Multiservice VP Networks

    Su C. and de Veciana G. "Statistical Multiplexing and Mix-Dependent Alternative Routing in Multiservice VP Networks " IEEE/ACM Trans. Networking vol. 8 No. 1 pp. 99-108 Feb. 2000. Multiservice VP Networks " IEEE/ACM Trans. Networking vol. 8 No. 1 pp. 99-108 Feb. 2000

    work page 2000

  6. [6]

    Teixeira, Luis & Corte-Real, Luis. (2009). Statistical Multiplexing of H.264 Video Streams Using Structural Similarity Information. J. Inf. Sci. Eng.. 25. 703-715

    work page 2009

  7. [7]

    Joint Video Coding and Statistical Multiplexing for Broadcasting Over DVB-H Channels,

    M. Rezaei, I. Bouazizi and M. Gabbouj, "Joint Video Coding and Statistical Multiplexing for Broadcasting Over DVB-H Channels," in IEEE Transactions on Multimedia, vol. 10, no. 8, pp. 1455-1464, Dec. 2008

    work page 2008

  8. [8]

    High efficiency video coding (HEVC): Replacing or complementing existing compression standards?,

    J. P. Henot, M. Ropert, J. Le Tanou, J. Kypréos and T. Guionnet, "High efficiency video coding (HEVC): Replacing or complementing existing compression standards?," 2013 IEEE International Symposium on Broadband Multimedia Systems and Broadcasting (BMSB), London, 2013, pp. 1-6

    work page 2013

  9. [9]

    Generalized Hurst exponent

    Tomaso, Aste. “Generalized Hurst exponent.” MathWorks File Exchange, 1.4, MathWorks, 31 Jan. 2013

    work page 2013

  10. [10]

    Analysis of an ATM buffer with self-similar (“fractal

    N. Likhanov, B. Tsybakov and N. D. Georganas, "Analysis of an ATM buffer with self-similar (“fractal”) input traffic," INFOCOM '95. Fourteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Bringing Information to People. Proceedings. IEEE, Boston, MA, 1995, pp. 985-992 vol.3

    work page 1995

  11. [11]

    L. Vu, I. Gupta, J. Liang, and K. Nahrstedt. Mapping the pplive network: Studying the impacts of media streaming on p2p overlays. Technical report, UIUC, 2006

    work page 2006

  12. [12]

    L. Vu, I. Gupta, J. Liang, K. Nahrstedt, Measurement and modeling of a large-scale overlay for multimedia streaming, in: Proceedings of the International Conference on Heterogeneous Networking for Quality, Reliability, Security and Robustness, QSHINE’07, 2007

    work page 2007

  13. [13]

    Florin Antone, Alexandru & Arsinte, Radu. (2014). A Study on the Optimal Implementation of Statistical Multiplexing in DVB Distribution Systems. Informatics and IT Today

    work page 2014

  14. [14]

    Analysis of Processing in Digital Terrestrial Television with Statistical Multiplexing for Standards DVB-T

    Espinosa Buitrago, Monica, and Ricardo Cesar Gomez Vargas. "Analysis of Processing in Digital Terrestrial Television with Statistical Multiplexing for Standards DVB-T." Symposium of Signals, Images and Artificial Vision - 2013: STSIVA - 2013 (2013): 1-5. Web

    work page 2013

  15. [15]

    Proceedings of SPIE - The International Society for Optical Engineering 9014 (2014): 90140O-90140O-13

    McCarthy, S. Proceedings of SPIE - The International Society for Optical Engineering 9014 (2014): 90140O-90140O-13. Web

    work page 2014

  16. [16]

    Ultra-HEVC Using Frame Frequency Error Optimization Technique for IPTV Realization

    Arun, M., and S. Selvakumar. "Ultra-HEVC Using Frame Frequency Error Optimization Technique for IPTV Realization." Multimedia Tools and Applications (2018): 1-15. Web

    work page 2018

  17. [17]

    Cross-layer packet prioritization for error-resilient transmission of iptv system over wireless network

    K. Go, S. Kang, Y. Yoon, M. Kim, B. Lee, "Cross-layer packet prioritization for error-resilient transmission of iptv system over wireless network", Proc. Multimedia Systems Conference (MMSys '14), pp. 180-190, Mar. 2014

    work page 2014

  18. [18]

    687–692 Author Iskandar Aripov is a Master of Science student in computer science at Georgia Institute of Technology with a focus on computing systems

    Hassane Khabbiza E, Alami REI, Qjidaa H (2016) A new method to reduce the bandwidth of IPTV systems, in 5th International Conference on Multimedia Computing and Systems(ICMCS), pp. 687–692 Author Iskandar Aripov is a Master of Science student in computer science at Georgia Institute of Technology with a focus on computing systems. He holds a Bachelor of S...

    work page 2016