P. T. Brady, "A statistical analysis of on-off patterns in 16 conversations," Bell System Technical Journal, vol. 47, no. 1, pp. 73--91, Jan. 1968.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....class of traffic profiles is quite general and can accurately represent a wide range of bursty network traffic, originating from various applications. Plain Markov modulated rate processes (featuring exponentially distributed state sojourns) have been successfully employed for representing voice [4,10], video [6,7,11,12] and data [3,13] traffic. The additional generality offered by the semiMarkovian models, namely the ability of specifying general probability distribution functions for the state durations, permits accurate modeling of traffic subject to shaping (e.g. leaky bucket based ....

P. T. Brady, "A statistical analysis of on-off patterns in 16 conversations". Bell System Tech. J ., Vol. 47, No. 1, pp.73-91, 1968.

....class II packet is permitted by the BS, depending on the available bandwidth left unused by class I traffic in our protocol. It is well known that voice traffic is characterized by an on off model since a speech signal is in either talk spurt or silent mode during the conversation of a mobile [2]. Each voice mobile is modeled by a two state Markov chain, where one state represents ON (transmitting at rate and the other represents OFF (idle) The utilization (i.e. fraction of time during which the voice mobile is active) is assumed to be 0.5. Given the number of voice mobiles, the number ....

P. T. Brady, "A statistical analysis of on-off patterns in 16 conversations, " Bell Syst. Tech. J., vol. 47, pp. 73--91, Jan. 1968.

....will be demanded in an ATM network : voice, video and data. For simulation purposes weinvestigated a link with a bit rate of 2.048 Mbps, although our results are expected to scale to higher link speeds. The voice model that we use is a standard model with two states, speaking and silence [30] [31]. During speaking periods, cells are produced by the source at the bit rate of the voice. During silent periods no cells are produced. The mean duration of the speaking ( or ON state ) is 0.352 seconds. If wecountthenumber of cells produced in this period it is found that this number is ....

P. T. Brady, `A Statistical Analysis of On-Off Patterns in 16 Conversations', Bell Syst. Tech. J.,pp 73-91, 1968.

....(on a per frame basis) during simulation statistically matches the number of cells per frame contained in the video data. This is presented in Section 5. Finally in Section 6 there is some discussion and plans for future work. 2. DATA CHARACTERIZATION Video, likevoice, is naturally bursty [1] [3] but can be manipulated to give a constant output bit rate. To gain the maximum efficiency from an ATM network only the real amount of information should be sent. This will change as the information contentofthe picture changes and so a variable bit rate output from the codec is expected. To ....

P. T. Brady, `A Statistical Analysis of On-Off Patterns in 16 Conversations', Bell Syst. Tech. J.,pp 73-91, 1968.

....getting the job done. D. Mixed Voice Data One of the more exciting of the potential applications that has been investigated by the DQRAP Research Group at IIT is the support of voice packet transmission. It has been known for years that the active period on voice channels is around 40 or less [16]. However, very little has been done to take advantage of this phenomenon. The reason of course is that, as discussed in section I, an effective multiple access protocol was not available. Research by Lin and Campbell [13] indicates that a common broadcast channel operating under DQRAP requires ....

P. Brady, "A Statistical analysis of on-off patterns in 16 conversations," The Bell System Technical Journal, Vol. 48, No. 2, pp. 2445-2472, 1968 .

....low. The source model parameters, and , are chosen to agree with a packet voice model, and therefore depend on the energy threshold level used in the silence suppression mechanism. We found several sets of parameters in the literature with speaker activity factors ranging from 33 to 48 [19,20,25,41,42,43].We use the median values with s = 40 voice activity. The simulation time is chosen large enough that the confidence intervals are generally negligible. For example, in Fig. 4 below, the 95 confidence intervals are only measurable at the knee of each curve, at about 1.5 dB. Atlowvalues of G the ....

P. T. Brady, "A statistical analysis of on-off patterns in 16 conversations," Bell System Technical Journal,vol. 47, pp. 73--91, January 1968.

....port of the switch, in which the on off characteristics of speech have been used for bandwidth compression, i.e. packet voice with silence detection. Each voice source is assumed to exhibit two states, active, and silent, representing the talkspurts and pauses present in conversational speech [36] In the active state each voice source generates packets at a regular rate representing 32 kbits s voice coding, 256 bit packets with a further 32 bits overhead, and a 20 MHz system clock. No packets are generated in the silent state. The two states are modeled by an exponential distribution ....

P. T. Brady, "A statistical analysis of on-off patterns in 16 conversations," Bell Syst. Tech. J., vol. 47, pp. 73--91, Jan. 1968.

....frequencies in different cells, sufficiently apart from each other in order to keep the interference at a tolerable level [2] Digital voice communication, however, has some features that make this philosophy not efficient. For example, voice activity is limited to about the 35 40 of the time [3]: in fixed assignment strategies the remaining time would be wasted. The implementation of high capacity voice communications systems calls for more sophisticated multiple access protocols. Recently, Gilhousen et al. 4] have proposed code division multiple access (CDMA) as a way to utilize more ....

P.T. Brady, "A statistical analysis of on-off patterns in 16 conversations", Bell Syst. Tech. J., vol. 47, pp. 73-91, Jan. 1968.

....slotted system, but both techniques make it possible to report state changes in a timely manner. The systems support both data and voice, and use variants of the MSTDM protocol. Bandwidth is the scarce resource in cellular radio networks. Since speech conversations are only active 40 of the time[11], a TASI[12] mode of operation is used, and transmission is suppressed during silent intervals. TASI is simpler to implement in MSTDM than in a circuit switched system because the unused bandwidth is automatically available for contention by other sources when a channel becomes idle. A problem ....

P. T. Brady, "A Statistical Analysis of On-Off Patterns in 16 Conversations," BSTJ, Jan. 1968, pp. 73-91.

....class II packet is permitted by the BS, depending on the available bandwidth left unused by class I traffic in our protocol. It is well known that voice traffic is characterized by an on off model since a speech signal is in either talk spurt or silent mode during the conversation of a mobile [2]. Each voice mobile is modeled by a two state Markov chain where one state represents ON (transmitting at rate R b ) and the other represents OFF (idle) The utilization ae vo (i.e. fraction of time during which the voice mobile is active) is assumed to be 0.5. Given the number K vo of voice ....

P. T. Brady, "A statistical analysis of on-off patterns in 16 conversations," Bell Syst. Tech. J., vol. 47, pp. 73--91, January 1968.

....bulk sizes for the two classes of traffic. We also assumed that arrivals were independent from slot to slot. In this section, we relax this independence assumption and study more complex traffic patterns. The particular traffic model we use follows a traditional approach from the literature [30, 31]. We consider each class of traffic to be a superposition of several voice sources. We model a single voice source as an on off source with a talkspurt mean of 352 milliseconds and a silence period mean of 650 milliseconds. In our discrete time model, this corresponds to a geometrically ....

Brady, P. T., "A Statistical Analysis of On-Off Patterns in 16 Conversations," Bell System Technical Journal, vol. 47, pp. 73--91, January 1968.

....the first part of this paper we introduce a new modeling technique to eliminate the unnecessary input state space explosion. Using the current modeling technique, each 2 state MC is used to characterize the ON OFF nature of an individual source or source element, like packetized voice and video [5, 6, 23]. As a result, this technique fails to apply in high speed networks, simply because of the exploded input state space when a large number of diverse sources are multiplexed on each link. Take a simple example of voice and video integration on a 100Mbps link. Let each voice source be modeled by an ....

....of the exploded input state space when a large number of diverse sources are multiplexed on each link. Take a simple example of voice and video integration on a 100Mbps link. Let each voice source be modeled by an i.i.d. 2 state MC at the average input rate 33Kbps, including ATM cell overhead [5]. Likewise, each video source is modeled by 10 i.i.d. 2 state MCs at its overall average input rate 4.7Mbps [23] Assume that the link carries 20 voice traffic and 80 video traffic at 70 utilization. Correspondingly, the link transmits 12 video and 425 voice calls, which in input modeling are ....

[Article contains additional citation context not shown here]

P. T. Brady, "A Statistical Analysis of On-Off Patterns in 16 Conversations," Bell System Technical Journal, Jan. 1968, pp. 73-91.

....is quite complicated. By definition, we have q = X r q r and q 2 = X r q 2 r . III. MULTI MEDIA TRAFFIC WITH DIFFERING TIME SCALES In multi media traffic modeling, a two state Markov chain is often used as a basic element to construct various sources, such as voice and video [15,17,18,21,22]. The Bernoulli process, which is often used to construct data sources, is an extreme case of two state Markov chains. This basic element alternates between ON and OFF states (Fig.1) The holding time on each state is geometrically distributed. While in the ON state a constant length cell is ....

....F. Numerical Studies In numerical studies, we first consider voice data integration on ATM, where the voice traffic comes from three different types of speech sources. Although each speech source can be properly modeled by a two state Markov process, we know from the experimental results [21 23] that the values of ffl and S at each source can be substantially different. They are dependent on the selected sensitivity of the speech activity detections, and are also distinct according to speaker s sex, age, language, and the contents of the dialogue. For simplicity, here we use ffl and S, ....

[Article contains additional citation context not shown here]

P.T. Brady, "A Statistical Analysis of On-Off Patterns in 16 Conversation," Bell Syst. Tech. J. Jan. 1968, pp. 73-91.

....on a set of heterogeneous two state MCs has been widely used in recent input correlated queueing analyses. Particularly in high speed telecommunication networking field, such a decomposition modeling technique is commonly used to characterize multimedia sources like voice, video and file transfers [2, 5, 9, 16]. A similar spectral decomposition technique is used in [3, 15, 6, 22] In a recent paper by Elwalid, Mitra and Stern [6] the solution of an MMPP M 1 queueing system is constructed where MMPP stands for Markov Modulated Poisson Process. The results presented in this paper is extended from the ....

Brady, P.T. "A Statistical Analysis of On-Off Patterns in 16 Conversation," Bell Syst. Tech. J. Jan. 1968, pp. 73-91.

....backlog exceeds the finite buffer capacity. More specifically, the packet loss behavior is the overall result of the interplays of the following factors: link utilization, time varying scale, source access rate, buffer size and source elements population. For two state Markovian source models [25, 7], the so called time varying scale, or time scale, refers to the mean holding times in ON and OFF states of a source element, denoted by TON and TOFF , respectively. in this case denotes the source access rate, or access intensity, while the source is in the ON period. Note that we use the ....

P.T. Brady, "A Statistical Analysis of On-Off Patterns in 16 Conversation," Bell Syst. Tech. J. Jan. 1968, pp. 73-91.

....protocols provide adequate performance for both voice and data transmission at the cost of occasional losses (but within strict delay limits) for the former and reasonable delays (but with no loss) for the latter traffic. Voice is statistically multiplexed using speech activity detectors (SAD) [3, 7], and a reservation is set up when the voice user is active (i.e. talkspurt state) and it is released when he she becomes inactive (i.e. silence state) during the transmission. Channel resources are used more efficiently This work was supported by the Bogazi ci University Research Fund under ....

P. T. Brady, "A statistical analysis of on-off patterns in 16 conversations", Bell Systems Technical Journal, vol. 47, January 1968, pp. 73-91.

....themselves) is application dependent. The statistics of these silence and talkspurt periods have been studied for conversational voice [7, 13 15] A number of Markovian models for voice packet generation have been proposed for interactive conversations, both for combined state of two parties [69,70] and for a single party [7,71] The most common model for a single party is that of a two state Markov chain representing silences and talkspurts [7, 14, 70] In these models, a speaker remains in the talkspurt state (generating packets periodically) for an exponentially distributed amount of ....

P. T. Brady, "A statistical analysis of on-off patterns in 16 conversations," Bell System Technical Journal, vol. 47, pp. 73--91, Jan. 1968.

....input power in the low frequency band has a dominant impact on queueing performance, whereas input power in the high frequency band can be neglected to a large extent. In practice, a large amount of input power for multimedia traffic in high speed networks is expected in the low frequency band [9] [11] As a result, the improvement in transmission efficiency by finite buffering is substantially limited. One basic element of the traffic descriptor in ATM protocol is the peak input rate. It is used by the usage parameter control (UPC) function to regulate certain traffic streams. By ....

P. T. Brady, "A Statistical Analysis of On-Off Patterns in 16 Conversation," Bell Syst. Tech. J., Jan. 1968, pp. 73-91.

No context found.

P. T. Brady, "A statistical analysis of on-off patterns in 16 conversations," Bell System Technical Journal, vol. 47, no. 1, pp. 73--91, Jan. 1968.

No context found.

P. T. Brady, "A Statistical Analysis of On-Off Patterns In 16 Conversations", Bell System Technical Journal, 47(1):73--91, January 1968.

No context found.

P.T. Brady, "A Statistical Analysis of On-Off Patterns in 16 Conversations, " BSTJ, pp. 73-91, January 1968.

No context found.

P. T. Brady, "A Statistical Analysis of On-Off Patterns in 16 Conversations, " Bell Syst. Tech. Journal, Vol. 47, pp. 73-91, Jan. 1968.

No context found.

P. T. Brady, "A Statistical Analysis of On-Off Patterns in 16 Conversations," BSTJ, Jan. 1968, pp. 73-91.

No context found.

P. T. Brady, "A Statistical Analysis of On-Off Patterns in 16 Conversations," BSTJ, Jan. 1968, pp. 73-91.

No context found.

Brady, P T "A Statistical Analysis of On-Off Patterns in 16 conversations," BSTJ, Vol.47, NO.1 (January 1968), pp. 73-91

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC