Abstract: A novel strategy of Measurement Based Admission Control (MBAC) for video distribution system has been developed, aiming at obtaining the needed Quality of Service (QoS), expressed in terms of probability for the video aggregation to exceed the ch
B-frames of MPEG video streams with different token buckets. Another way to improve QoS degree for video flows is to provide a constant level of perceptual video stream Quality of Service, as described in [9].
2. THE PROPOSED SYSTEM
In the MBAC algorithm proposed in this paper, the new flow requesting admission will be admitted and forwarded together with the rest of flow aggregation only overflow coming from the peak of the required bandwidth to deliver the composite video stream (smoothed signal).
The probability of the union of these events can be approximated with the sum of the relative probabilities. As we are searching for a conservative MBAC algorithm, we can assume valid the following relation:
Pq≈Ps+Pi. (3.1) if the QoS guarantees of every flow belonging to the aggregate (included the new one) will be satisfied.
The algorithm estimates the required bit rate margin with respect to the channel capacity, C, considering the statistics of N video composite stream.
The MBAC system is based on a stream adaptive filtering technique, based on Burg linear prediction algorithm, to split the bit rate required by the composite video stream into the slow varying bit rate and its quick-time variations [3].
Statistics of N stream composite videos are exploited to estimate a “strict” margin with respect to the channel capacity, C, and estimate the probability of freeze for the new admitted stream in the composite video.
To make the system adaptive to the experienced multiplexed data stream input, Burg AutoRegressive (AR) estimation has been used, to exploit all the predictable information about the required bit rate. The unpredictable error has been characterized in terms of positive peak value, to be used as a measure of the margin with respect to the channel capacity to obtain a low freeze probability in the system.
Below, the proposed technique is described in detail and the measures of the channel capacity overflow probabilities are computed.
2.1 The MBAC algorithm
To address the problem of admission control, the statistics of an N-streams composite video have been considered. Two parameters are searched to define the MBAC algorithm: the margin with respect to the channel capacity, that for a N-streams composite video has to be used; the overflow probability for the last user accepting the limited resources video streaming system.
We have considered the statistics of both the prediction error and the predicted (low pass) composite video stream for all the cases of N-streams composite video we can happen to manage.
Given a channel capacity C, as in figure 1, we have to compute a conservative but minimal margin with respect to C, to manage all the unpredictable bit rate requirements in the N-streams composite video.
Thus we have observed the statistics of the prediction error, first, to obtain information about the margin to be used.
The channel capacity overflow probability event depends on the union of two events:
overflow coming from a peak in the prediction error;
where Pq is the searched overflow probability, Ps is the
overflow contribute coming form the signal component and Pi is the overflow probability coming from the prediction error, from which derives the bandwidth margin to be used due to the unpredictability of N-stream aggregation bit rate.
As a further requirement for the MBAC system, we assume the Pi probability be a fraction of Ps to safely guarantee a sufficient bandwidth margin proportionally to the loss probability of the N-stream aggregate (the established QoS parameter):
Pi=α×Ps (3.2)
In this way, the system will be able to better absorb prediction error peaks. The choice of α will be established in the sequel. From (3.1):
Pq=(1+α) Ps (3.3) With such assumptions, we can draw the graphics for the
Ps as a function of the bit rate for several orders N of the composite video stream.
By drawing the polynomial fitting curves of the bit rate as function of the Ps, we can compute the bit rate of the composite video signal as a function of the overflow probability Ps.

Figure 1: bit rate time evolution and definition of the channel capacity and margin.
2.2 The MBAC implementation
The overflow probability can be used to define a MBAC system; the N-streams composite video global bit rate has to be considered; the two components bit rates respectively coming from the unpredictability of the bit rate and from the smoothed version of the bit rate time series have to be considered.
Two different measures have been carried out, one over the predicted signal and the other over the prediction error. The first addressed system parameter to define is the extension of the observation window to be used in this application. The second, the filter taps to be used in the LP filter. Simulations have shown that a fair choice of such parameters is 1000 samples for the processing window (corresponding to about 30 seconds at 30 fps), and a LP
A New MBAC algorithm for... 暂无评价 4页 免费...VBR Video Streaming in a Multicastbased Video-on-...algorithm with two other algorithms based on ...
Based on autoregressive AR estimation approach 2 , a new adaptive frequency estimation algorithm which gives direct and accurate frequency measurements is ...
Based on autoregressive AR estimation approach 2 , a new adaptive frequency estimation algorithm which gives direct and accurate frequency measurements is ...
热门文档