RU210691U1 - DEVICE FOR CALCULATION OF OPTIMUM PARAMETERS FOR SMOOTHING FRACTAL TRAFFIC BY CRITERION OF MAXIMUM CORRESPONDENCE OF AVERAGE RATE AND GREATEST SURGE OF INPUT AGGREGATED FLOW UNDER THE CONDITION OF NO PACKET LOSS DUE TO BUFFER OVERFLOW - Google Patents

Abstract

The utility model relates to the field of packet-switched data networks. The technical result is the creation of a device for calculating the optimal value of the smoothing rate and the volume of the smoothing buffer, ensuring the absence of losses due to buffer overflow and the maximum compliance with the average rate and the largest burst of the input aggregated packet stream. The device contains a signal input unit, equivalent to the value of the maximum packet size; a block for inputting a signal equivalent to an estimate value of the standard deviation of the rate of the input packet stream; input unit of a single signal level; a signal input unit equivalent to the value of the Hurst exponent; a signal input unit equivalent to the number of received packets; subtraction block; reciprocal calculation unit; five multiplication blocks; division block; logarithm block; two exponentiation blocks; module calculation unit; square root calculation block; a block for inputting a signal equivalent to the value of the average rate of the input packet stream; two addition blocks; display unit of the optimal smoothing speed; the block for displaying the optimal smoothing buffer, and organizing links between them, makes it possible to calculate the optimal parameters for smoothing fractal traffic according to the criterion of maximum correspondence between the average speed and the largest burst of the input aggregated stream, provided there are no packet losses due to buffer overflow. 2 ill.

Description

The utility model relates to the field of technology for packet-switched data transmission networks and can be used to calculate the optimal value of the smoothing rate and the size of the smoothing buffer, ensuring no loss due to buffer overflow and maximum compliance with the average speed and the largest burst of the input aggregated packet flow.

Known devices [1, 2], similar in purpose of the proposed device. The elements included in the devices [1, 2] are mainly aimed at optimizing the parameters of the communication network by changing the route or throughput, while the buffering process is not described.

The device [1] converts the MPEG transport stream into IP packets for broadcasting in a WLAN, and this device contains a packetizer, but the composition of the packetizer is not disclosed, but it is only stipulated that the formation of packets is carried out in accordance with certain internetwork protocols and program-specific information. Formation of packages, thus, does not allow to ensure the absence of losses due to buffer overflow and simultaneous smoothing of the input stream.

An apparatus [2] that implements a variable packet length approach for high packet data rate transmissions in an access network (AN) comprises a data rate control (DRC) node for receiving data rate requests from access terminals, with each rate request data indicates at least one transmission rate that the access terminal requests from the access network device to transmit data to the access terminal; a physical layer (PL) packet generation node configured to receive data rate requests from the DRC node based on the rate requests, select at least two access terminals to receive data from the multi-user PL packet, select a length from a set of lengths for the multi-user PL packet, selecting a transmission rate from a set of transmission rates for transmitting the multi-user PL packet, and generating a multi-user PL packet of the selected length; and a transmitter for transmitting the multi-user PL packet to the access terminals.

The disadvantage of this technical solution is the selection of the packet size based on the requested transmission rate, and the set of possible packet sizes corresponding to a specific transmission rate is predetermined, moreover, optimal lossless smoothing is not provided in conditions of a pulsating information flow.

The closest in technical essence and selected as a prototype is the device [3], which calculates the optimal packet size according to the criterion of minimum delay in the reproduction of digital compressed images under conditions of optimal lossless smoothing.

The technical result of the device-tax [3] is provided through the use of the block input value of the size of the service part of the package; block for entering the value of the maximum size of the compressed frame; block for entering the value of the frame rate; a block for inputting the value of the speed of the decompression device; block for inputting the value of the speed of the transmitter; block for inputting the value of the speed of the receiver; eight multiplication blocks; minimum calculation block; six division blocks; two blocks for calculating the reciprocal; square root calculation block; block rounding up to the nearest integer; block rounding down to the nearest integer; five addition blocks; comparison block; inversion block; the result display unit, wherein: the output 1 of the block 1 for entering the value of the size of the service part of the packet is connected to the input 1 of the multiplication block 7, to the input 1 of the multiplication block 22 and to the input 2 of the addition block 33; output 1 block 2 input values of the maximum size of the compressed frame is connected to the input 2 block 7 multiplication, input 1 block 8 multiplication, input 1 block 10 division and input 1 block 25 division; output 1 block 3 input frame rate is connected to the input 2 block 8 multiplication; output 1 block 4 input value of the speed of the decompression device is connected to the input 2 block 25 division and input 2 block 9 calculate the minimum; the output 1 of the block 5 input values of the speed of the transmitter is connected to the input 3 of the block 9 to calculate the minimum and to the input 1 of the block 12 to calculate the reciprocal; the output 1 of the block 6 input values of the speed of the receiver is connected to the input 4 of the block 9 to calculate the minimum and to the input 1 of the block 11 to calculate the reciprocal; output 1 block 7 multiplication connected to the input 1 block 13 division and input 1 block 16 division; the output 1 of the multiplication block 8 is connected to the input 1 of the block 9 for calculating the minimum; the output 1 of the block 9 calculation of the minimum is connected to the input 2 of the division 10, with the input 2 of the division 13 and with the input 2 of the multiplication block 15; output 1 block 10 division is connected to the input 3 block 26 addition and input 2 block 27 addition; the output 1 of the block 11 calculating the reciprocal is connected to the input 1 of the block 14 addition; the output 1 of the block 12 calculating the reciprocal is connected to the input 2 of the block 14 addition; output 1 of division 13 is connected to input 1 of division 19 and to input 1 of division 21; output 1 of block 14 addition is connected to input 1 of block 15 multiplication, with input 2 of block 24 of multiplication, with input 2 of block 23 of multiplication and with input 2 of block 22 of multiplication; output 1 block 15 multiplication is connected to the input 2 block 16 division; the output 1 of the block 16 division is connected to the input 1 of the block 17 for calculating the square root; the output 1 of the square root calculation block 17 is connected to the input 1 of the round-down block 20 and to the input 1 of the round-up block 18; output 1 of block 18 rounding up connected to input 2 of block 19 division, with input 1 of block 23 of multiplication and with input 1 of block 29 of multiplication; the output 1 of the block 19 division is connected to the input 5 of the block 26 addition; output 1 block 20 rounding down connected to the input 2 block 21 division, input 1 block 24 multiplication and input 2 block 31 multiplication; output 1 block 21 division connected to the input 4 block 27 addition; output 1 block 22 multiplication connected to the input 2 block 26 addition and input 3 block 27 addition; output 1 block 23 multiplication connected to the input 4 block 26 addition; output 1 block 24 multiplication connected to the input 5 block 27 addition; output 1 block 25 division is connected to the input 1 block 26 addition and input 1 block 27 addition; output 1 block 26 addition is connected to the input 1 block 28 comparison; output 1 block 27 addition connected to the input 2 block 28 comparison; output 1 block 28 comparison is connected to the input 2 block 29 multiplication and input 1 block 30 inversion; output 1 block 29 multiplication connected to the input 2 block 32 addition; output 1 block 30 of the inversion is connected to the input 1 of the block 31 multiplication; output 1 block 31 multiplication connected to the input 1 block 32 addition; output 1 block 32 addition is connected to the input 1 block 33 addition; output 1 block 33 addition is connected to the input 1 block 34 display the result.

The analogue device [3] has the following main disadvantages:

1. Do not optimize the real-time stream for which the traffic path is not known (accumulative function of the input stream arrival).

2. They do not provide a balanced ratio between the smoothing speed and the size of the smoothing buffer, in which an insignificant change in one parameter does not lead to a significant change in the other.

3. Do not provide maximization of the correspondence of smoothing parameters to the characteristics of the input stream.

4. Optimize only the process of transferring the media stream.

In the proposed utility model, the noted disadvantages are eliminated.

The purpose (technical result) of the utility model is to create a device for calculating the optimal parameters for smoothing fractal traffic according to the criterion of maximum compliance with the average speed and the largest burst of the input aggregated stream, provided there are no packet losses due to buffer overflow.

This goal is achieved by the fact that this device through the use of:

a signal input block equivalent to the value of the maximum packet size;

a signal input unit equivalent to an estimate value of the standard deviation of the rate of the input packet stream; input unit of a single signal level;

a signal input unit equivalent to the value of the Hurst exponent; a signal input unit equivalent to the number of received packets; subtraction block;

reciprocal calculation unit;

five multiplication blocks;

division block;

logarithm block;

two exponentiation blocks;

module calculation block;

square root calculation block;

block input signal, equivalent to the value of the average rate of the input stream of packets;

two addition blocks;

a block for displaying the optimal smoothing speed; of the block for displaying the optimal smoothing buffer, and:

output 1 block 1 input signal equivalent to the value of the maximum packet size is connected to the input 1 block 22 addition;

output 1 block 2 input signal, equivalent to the value of the estimate of the standard deviation of the rate of the input stream of packets, connected to the input 1 block 8 multiplication;

output 1 block 3 input of a single signal level is connected to the input 1 block 6 subtraction;

output 1 block 3 input of a single signal level is connected to the input 1 block 6 subtraction;

output 1 block 4 input signal, equivalent to the value of the Hurst exponent, connected to the input 2 block 6 subtraction, input 1 block 9 division and input 2 block 12 exponentiation;

output 1 block 5 input signal, equivalent to the number of received packets, connected to the input 1 block 7 calculation of the inverse value;

output 1 block 6 subtractor connected to the input 2 block 8 multiplication and input 2 block 9 division;

the output 1 of the block 7 calculation of the inverse value is connected to the input 1 of the block 10 of the logarithm;

output 1 block 8 multiplication connected to the input 1 block 17 multiplication and input 1 block 18 multiplication;

the output 1 of the division block 9 is connected to the input 1 of the squaring block 11 and to the input 1 of the exponentiation block 12;

output 1 block 10 logarithm is connected to the input 1 block 13 calculation module;

output 1 block 11 squaring connected to the input 2 block 17 multiplication;

output 1 block 12 exponentiation is connected to the input 1 block 15 multiplication;

output 1 block 13 calculation of the module is connected to the input 1 block 14 multiplication by two;

output 1 block 14 multiplication by two is connected to the input 2 of the block 15 multiplication;

output 1 block 15 multiplication connected to the input 1 block 16 calculate the square root;

output 1 block 16 calculate the square root is connected to the input 3 block 17 multiplication and input 2 block 18 multiplication;

output 1 block 17 multiplication is connected to the input 2 block 20 addition;

output 1 block 18 multiplication is connected to the input 2 block 22 addition;

output 1 block 19 input signal, equivalent to the value of the average rate of the input stream of packets, connected to the input 1 block 20 addition;

the output 1 of the addition block 20 is connected to the input 1 of the optimal smoothing speed display block 21;

output 1 block 22 addition is connected to the input 1 block 23 mapping the optimal smoothing buffer.

Thanks to a new set of features due to the additional introduction of: a signal input block equivalent to the value of the maximum packet size; a signal input unit equivalent to an estimate value of the standard deviation of the rate of the input packet stream; input unit of a single signal level; a signal input unit equivalent to the value of the Hurst exponent; a signal input unit equivalent to the number of received packets; subtraction block; logarithm block; two exponentiation blocks; module calculation block; block input signal, equivalent to the value of the average rate of the input stream of packets; a block for displaying the optimal smoothing speed; a block for displaying the optimal smoothing buffer, by removing: a block for entering the value of the size of the service part of the packet; block for entering the value of the maximum size of the compressed frame; block for entering the value of the frame rate; a block for inputting the value of the speed of the decompression device; block for inputting the value of the speed of the transmitter; block for inputting the value of the speed of the receiver; three multiplication blocks; minimum calculation block; five division blocks; reciprocal calculation unit; block rounding up to the nearest integer; block rounding down to the nearest integer; three addition blocks; comparison block; inversion block; of the result display unit and organization of connections between them, calculate the optimal parameters for smoothing fractal traffic according to the criterion of maximum compliance with the average speed and the largest burst of the input aggregated stream, provided there are no packet losses due to buffer overflow.

The utility model can be widely used in communication systems using synchronization mechanisms for encryption or error-correcting coding, as well as for streaming data transmission using smoothing buffers.

The use of the proposed device makes it possible to calculate such optimal values of the smoothing rate and smoothing buffer size, which ensure the maximization of the correspondence between the average speed and the largest burst of the input aggregated stream, provided there are no packet losses due to buffer overflow.

The device implements the following theoretical provisions.

At the end of the 90s, it was discovered that telecommunications traffic has a long-term dependence property; since that time, a large number of statistical models have been developed to describe a single and aggregated fractal multimedia, and not only, flow. In addition, since then, the impact of long-term memory, characteristic of real traffic, on the performance of a communication network has been widely studied. At the same time, another philosophy was actively developed to solve the problem of traffic description. It was based on the concept of a linearly bounded arrival process (LBAP), first introduced in [4]. An LBAP process is an incoming process bounded by a linear boundary of the total amount of data generated by a traffic source in any time interval. To shape traffic in accordance with the concept of the LBAP process, such profiling and traffic shaping algorithms as “token bucket” and “leaky bucket” were introduced, which in this case are considered equivalent. This approach allows only two parameters: the slope r of the straight line - the speed of the "markers" per unit time and the additive constant b - the size of the "bucket", which characterizes the burstiness of the flow, to obtain a complete characterization of the worst case for the source of telecommunication traffic. It is intuitively clear that for a given traffic source, the set of pairs (the rate of "markers" r and the size of the "bucket" b) that satisfy the LBAP constraint is generally infinite, and the determination of the minimum set of LBAP pairs, that is, its closure, requires offline processing of the entire route. In addition, the choice of a suitable pair on this curve depends on the conditions imposed on the delay when optimizing the parameters of telecommunication traffic. Summarizing the above, we can assume that the LBAP-curve describes a set of pairs (r,b), parameters of the bursty flow profiling device, for which there are no losses due to buffer overflow.

The LBAP curve describes the restriction of traffic generated by the source with the shaper according to the "token bucket" algorithm during any time interval Δt, by a linear function of this time interval.

If we denote by A (Δt), the amount of traffic transmitted by the source during the time interval Δt, then the traffic is called LBAP traffic, that is, having a linearly limited arrival process, if there is a pair (r, b) such that

where r is the long-term average source rate, and b is the maximum burst of traffic from the source that can be sent in any time interval of duration Δt. Note that b is the maximum deviation that a source can exhibit with respect to its long-term average behavior. An LBAP curve is a set of pairs (tuples) of r and b, under which the stream is linearized (smoothed) and no losses due to buffer overflow are ensured.

которое аппроксимирует множество минимальных пар (r,b), обеспечивающих передачу данных без переполнения буфера.One of the most common approaches to obtaining an LBAP-curve with parameters (r,b) for the source is to use a "marker bucket" controller. From a theoretical point of view, a pair (r,b) that satisfies the LBAP constraint is not unique. On the contrary, solutions (1) lie in a two-dimensional set whose boundary is a curve of infinite minimal pairs (r,b) that obey relation (1). This curve will be called the LBAP-curve and denoted by the equation

which approximates the set of minimal pairs (r,b) that ensure data transfer without buffer overflow.

Intuitively, as r approaches the average source velocity from the right, simple queuing reasoning says that the corresponding value of b increases rapidly to infinity, while as r increases until the source velocity peaks, b tends monotonically to zero. In FIG. 1 shows an example of an LBAP curve obtained from the analysis of a bursty aggregated flow.

While all pairs (r,b) lying on the LBAP curve satisfy the non-denial-of-service description of LBAP, one might consider using an additional degree of freedom to optimize according to some predetermined criteria. LBAP-curve is nothing else than a set of “marker bucket” parameters that are optimal according to the criterion of the absence of losses.

The studies carried out show that for many conventional real paths (traffic arrival functions), the LBAP curve shows a well-defined knee region, beyond which, each of the parameters r and b, respectively, increase rapidly with minor changes in the other parameter.

This fact prompts the idea of choosing as the most appropriate pair (r*,b*) a pair that lies in the vicinity of the bending region. It is worth noting that when the peak rate P _LBAP of the traffic source is known, it is possible to obtain a slightly refined characteristic, while understanding that the total amount of traffic generated by the source in this case is limited as follows

A source that satisfies this constraint is called a "double leaky bucket" regulator with parameters (r,b,P _LBAP ).

To take into account the influence of ripples in communication networks on the operation of traffic shapers of the "token bucket" type, we will generalize the terminology used and the fundamental concepts of short-wave and long-range reflection processes.

и автоковариационной функцией вида:Let A(t) be the total volume of traffic (in units of packets, bytes, or bits) generated by one or more sources over a time interval t. Let us divide the time interval t into single non-overlapping intervals of duration T _and . Let X _n be the amount of traffic (also called work) recorded in the nth unit of time, such that X _n is the increment process for A(t). Assume that X _n is a stationary stochastic sequence in the broad sense with mean value

and an autocovariance function of the form:

But since the size of the trace is finite, it is more appropriate to speak of a sample autocovariance function of the form:

where n is the number of intervals for splitting the bit stream of data arrival. The process X _n is called:

short-term dependent (SCD), or with a short memory, if

long-term dependent (LTD), or long-term memory if

Thus, in long-term dependent processes, the series associated with autocovariance functions diverge. In addition, it turns out that the autocovariance function itself decreases to zero as k increases, in accordance with the power form k ^-α , for some 0<α<1. This property allows you to get some visual results. For this, both visual inspection and quantitative analysis can be used, as a result of which it is possible to determine whether the process is long-term dependent and the degree of this long-term dependence.

The above characteristics of the TIR processes cause a number of problems with the description of the LBAP characteristics for TIR traffic. In particular, of scientific interest is the behavior of the LBAP curve in the presence of traffic that has the same macroscopic parameters in terms of average bit rate and peak bit rate, but at the same time it is a DVZ or SVZ process. It is intuitively clear that the pulsating nature of the traffic, which depends on the long-term dependence, implies that, for the same chosen average speed r, the corresponding value of the burstiness b for DVZ traffic will be higher than for SVZ traffic. Since the value of the b parameter is related to the buffer space that is pre-allocated (reserved) at each intermediate node on the flow path during the connection establishment by resource reservation protocols, a large value of b will generate excess buffer space, and this will lead to a large end-to-end delay. On the other hand, for a fixed choice of the token bucket size parameter b, DVD traffic is expected to require a higher token rate r to satisfy the LBAP constraint. Describing the qualitative characteristic of the traffic regulator in the form of an LBAP curve, it should be noted that it shows how r and b correlate, ensuring lossless transmission.

In most cases, the choice of a pair (r*,b*) of the “marker bucket” parameters is carried out by a heuristic method, using qualitative reasoning about the inflection area on the LBAP curve. Although such an approach is acceptable, a more formalized formulation and solution of this scientific problem is of scientific value. Despite the fact that all points of the LBAP curve ensure the absence of losses due to buffer overflow, the problem of choosing a point that would be optimal according to the criterion of maximum correspondence between the average speed and the largest burst of the input aggregated stream is relevant from the point of view of ensuring guaranteed communication quality.

.Thus, when choosing the optimal parameters of the "marker bucket" with a fractal input stream, we will use the maximum packet size

.

для LBAP-кривой. Результирующая точка (r*,b*), является оптимальным решением этой двухпараметрической задачи условной оптимизации. Если будет обеспечено выполнение математического выражения

то эта оптимизационная задача может быть решена хорошо известным методом неопределенных множителей Лагранжа.The idea is to find a correct objective function ϕ _u (r,b), which must be minimized, while satisfying the constraint

for the LBAP curve. The resulting point (r*,b*), is the optimal solution to this two-parameter conditional optimization problem. If the mathematical expression will be satisfied

then this optimization problem can be solved by the well-known method of indefinite Lagrange multipliers.

обеспечивает максимум соответствия средней скорости и наибольшему всплеску входного агрегированного потока и отсутствие потерь из-за переполнения буфера при пульсирующем входном потоке, то целевая функция, которую требуется рассматривать, примет вид:Taking into account the fact that the point of the LBAP-curve, which is located closest to the point

provides maximum compliance with the average speed and the largest burst of the input aggregated stream and the absence of losses due to buffer overflow with a pulsating input stream, then the objective function that needs to be considered will take the form:

- является Евклидовой нормой вектора, представляющего собой разность векторов (r,b) и

.where:

- is the Euclidean norm of the vector, which is the difference of the vectors (r,b) and

.

The physical meaning of minimizing the objective function given by expression (7) is to ensure that there are no losses due to buffer overflow and that the average rate and the largest burst of the input aggregated flow correspond.

Assume the following mathematical model describing the amount of aggregated traffic generated by various sources over the time interval [0,t][5]:

- это среднее значение объема трафика за интервал [0,t], а B^H (t) - это процесс ФБД:where

is the average value of the traffic volume over the interval [0,t], and B ^H (t) is the FWA process:

where:

1) 0<H<1 - Hurst exponent;

2) ƒ(t,y) - the core of the process, given by the expression:

where I _A - indicator function of the set A;

4) dB(y) - basic white Gaussian noise.

The FBD process is a nonstationary DVZ process with a Gaussian marginal distribution N(0,σ ² t ^2H ), stationary increments and B ^H (0)=0. The volume of traffic generated by the source for any time interval Δt is determined by an expression of the form:

а ниже будет показано, что полученное решение также является экспоненциально убывающим. В данном случае ограничение, выраженное через вероятность отказа в обслуживании, примет следующий вид [5]:When dealing with a stochastic process, the LBAP constraint can be expressed as a probability of mismatching the traffic profile (buffer overflow). For convenience, the LBAP-boundary is approximated by the exponential function

and below it will be shown that the resulting solution is also exponentially decreasing. In this case, the constraint, expressed in terms of the probability of denial of service, will take the following form [5]:

with the corresponding LBAP-curve of the form [275]:

Condition (14) can be transformed to the form [5]:

where for the probability of missing one packet P _otk (r,b) there is an asymptotic expression obtained by Norros in the form [5]:

at

, то есть ситуации отказа в обслуживании из-за переполнения буфера. Поскольку реальная LBAP-кривая строится для детерминированного ограничения (2), то параметр

выбирается в диапазоне значений, близких к абсолютному значению натурального логарифма минимальной вероятности несоответствия очередного пакета, установленному профилю трафика, оцениваемой по фактически измеренной трассе, то естьThe most important aspect is the choice of the maximum allowable probability of traffic inconsistency with the established profile.

, that is, denial-of-service situations due to buffer overflows. Since the real LBAP curve is constructed for the deterministic constraint (2), the parameter

is selected in the range of values close to the absolute value of the natural logarithm of the minimum probability of the next packet mismatch with the established traffic profile, estimated from the actually measured path, i.e.

where N _pack is the number of packets (frames, frames) in the measured trace.

Finally, the LBAP curve, taking into account the fact that during packet transmission, the buffer size cannot be zero and must provide the ability to store data of at least one full packet of maximum length, can be written for the case of transmission of aggregated packet traffic in the following form:

.This expression refines the mathematical model of the LBAP-Norros curve described in [5], which graphically represents the lifting of the tail of the exponential function by the value of the maximum burst

.

The refined mathematical model (19) more accurately describes the behavior of the LBAP curve under conditions of packet transmission of pulsating telecommunication traffic in communication networks.

Next, analytical expressions will be presented to obtain the optimal pair in accordance with the formulated optimality criterion. The solutions were obtained using the method of indefinite Lagrange multipliers. Namely, a function of the form was defined:

к нулю.and a system of three nonlinear algebraic equations is solved by equating the partial derivatives for the function F(r,b,λ) with respect to the parameters r, b and the conditions

to zero.

In accordance with the method of indefinite Lagrange multipliers, we find the partial derivatives of the function (20) with the objective function (7) with respect to the optimized parameters r and b. As a result, we obtain a system of equations of the form:

After elementary algebraic transformations, the solution will take the form:

The device that implements these theoretical provisions is shown in Fig. 2.

The device for calculating the optimal parameters for smoothing fractal traffic according to the criterion of maximum compliance with the average speed and the largest burst of the input aggregated stream, provided there are no packet losses due to buffer overflow, contains: block 1 for inputting a signal equivalent to the value of the maximum packet size; block 2 input signal, equivalent to the value of the estimate of the standard deviation of the rate of the input packet stream; block 3 input of a single signal level; block 4 input signal equivalent to the value of the Hurst exponent; block 5 input signal equivalent to the number of received packets; block 6 subtraction; block 7 calculating the reciprocal; blocks 8, 14, 15, 17, 18 multiplication; division block 9; logarithm unit 10; exponentiation blocks 11, 12; block 13 calculation of the module; block 16 calculating the square root; block 18 input signal equivalent to the value of the average rate of the input stream of packets; blocks 20.22 addition; block 21 display the optimal speed of smoothing; block 23 display the optimal smoothing buffer.

.The operation of the device is as follows. In block 1, a signal is injected that is equivalent to the value of the maximum packet size

.

.In block 2, a signal is introduced that is equivalent to the value of the estimate of the standard deviation of the rate of the input packet stream

.

In block 3 enter a single signal level.

In block 4 enter a signal equivalent to the value of the Hurst exponent N _sweat.

In block 5 enter the signal equivalent to the number of received packets N _pack .

To the input 1 of block 6 subtracting from output 1 of block 3 input of a single signal level and to input 2 of block 6 subtracting from output 1 of block 4 signal input equivalent to the value of the Hurst exponent, signals equivalent to 1 and H _sweat , respectively, and at output 1 subtraction block 6, a signal equivalent to 1-H _pot is generated.

.At the input 1 of the block 7 for calculating the inverse value from the output 1 of the block 5 for the signal input equivalent to the number of received packets, a signal equivalent to N _pack is supplied, and at the output 1 of the block 7 for calculating the inverse value, a signal is generated that is equivalent to

.

и 1-Н_пот соответственно, и на выходе 1 блока 8 умножения, формируется сигнал, эквивалентный

.At the input 1 of block 8 of multiplication from output 1 of block 2 of the input signal, equivalent to the value of the estimate of the standard deviation of the rate of the input packet stream, and at the input of 2 of block 8 of multiplication from output 1 of block 6 of subtraction, signals equivalent to

and 1-H _sweat , respectively, and at the output 1 of the multiplication block 8, a signal is generated that is equivalent to

.

To the input 1 of the division block 9 from the output 1 of the signal input block 4, equivalent to the value of the Hurst exponent, and to the input 2 of the division block 9 from the output 1 of the subtractor 6, signals equivalent to H _sweat and 1-H _sweat , respectively, are supplied, and at the output 1 of the block 9 divisions a signal equivalent to

, и на выходе 1 блока 10 логарифмирования формируется сигнал, эквивалентный

At the input 1 block 10 logarithm output 1 block 7 calculation of the reciprocal value is fed a signal equivalent to

, and at the output 1 of the block 10 of the logarithm, a signal equivalent to

и на выходе 1 блока 11 возведения в квадрат формируется сигнал, эквивалентный

.At the input 1 block 11 squaring output 1 block 9 division serves a signal equivalent to

and at the output 1 of the squaring block 11, a signal equivalent to

.

и Н_пот соответственно, и на выходе 1 блока 12 возведения в степень формируется сигнал, эквивалентный

.To the input 1 of the exponentiation block 12 from the output 1 of the dividing block 9 and to the input 2 of the exponentiation block 12 from the output 1 of the signal input block 4, equivalent to the value of the Hurst exponent, signals equivalent to

and H _sweat , respectively, and at the output 1 of the exponentiation block 12, a signal is generated that is equivalent to

.

, и на выходе 1 блока 13 вычисления модуля формируется сигнал, эквивалентный

To the input 1 of the block 13 of the module calculation from the output 1 of the logarithm block 10, a signal equivalent to

, and at the output 1 of block 13 of the calculation of the module, a signal equivalent to

и на выходе 1 блока 14 умножения на 2 формируется сигнал, эквивалентный

To the input 1 of the block 14 multiplication by 2 from the output 1 of the block 13 calculation of the module, a signal equivalent to

and at the output 1 of block 14 multiplication by 2, a signal equivalent to

и

соответственно, и на выходе 1 блока 15 умножения формируется сигнал, эквивалентный

.To the input 1 of the block 15 of multiplication from the output 1 of the block 12 of exponentiation and to the input 2 of the block 15 of the multiplication from the output 1 of the block 14 of multiplication by 2, signals equivalent to

and

respectively, and at the output 1 of the multiplication block 15, a signal is generated that is equivalent to

.

, и на выходе 1 блока 16 вычисления квадратного корня формируется сигнал , эквивалентный

.At the input 1 of the block 16 for calculating the square root from the output 1 of the multiplication block 15, a signal equivalent to

, and at the output 1 of block 16 for calculating the square root, a signal is generated that is equivalent to

.

,

и

соответственно, и на выходе 1 блока 17 умножения формируется сигнал, эквивалентный

.To the input 1 of the multiplication block 17, from the output 1 of the multiplication block 8, to the input 2 of the multiplication block 17, from the output 1 of the squaring block 11 and to the input 3 of the multiplication block 17, from the output 1 of the square root calculation block 16, signals equivalent to

,

and

respectively, and at the output 1 of the multiplication block 17, a signal is generated that is equivalent to

.

соответственно, и на выходе 1 блока 18 умножения формируется сигнал, эквивалентный

.To the input 1 of the multiplication block 18 from the output 1 of the multiplication block 8 and to the input 2 of the multiplication block 18 from the output 1 of the square root calculation block 16, signals equivalent to

respectively, and at the output 1 of the multiplication block 18, a signal equivalent to

.

.In block 19, a signal is introduced that is equivalent to the value of the average rate of the input packet stream

.

и

соответственно, и на выходе 1 блока 20 сложения формируется сигнал, эквивалентный

At the input 1 of the addition block 20 from the output 1 of the signal input block 19, equivalent to the value of the average rate of the input packet stream and at the input 2 of the addition block 20 from the output 1 of the multiplication block 17, signals equivalent to

and

respectively, and at the output 1 of the addition block 20, a signal equivalent to

, и производят его отображение на экране оператора. Полученный сигнал эквивалентен величине оптимальной скорости сглаживания, обеспечивающей отсутствие потерь из-за переполнения буфера.To the input 1 block 21 display the optimal rate of smoothing from the output 1 block 20 addition serves a signal equivalent to

, and display it on the operator's screen. The received signal is equivalent to the value of the optimal smoothing rate, which ensures the absence of losses due to buffer overflow.

и

соответственно, и на выходе 1 блока 22 сложения формируется сигнал, эквивалентный

.At the input 1 of the addition block 22 from the output 1 of the signal input block 1, equivalent to the value of the maximum packet size, and at the input 2 of the addition block 22 from the output 1 of the multiplication block 18, a signal equivalent to

and

respectively, and at the output 1 of the addition block 22, a signal equivalent to

.

, и производят его отображение на экране оператора. Полученный сигнал эквивалентен объему (размеру) оптимального буфера сглаживания, обеспечивающему отсутствие потерь из-за переполнения буфера.To the input 1 block 23 display the optimal smoothing buffer from the output 1 block 22 addition serves a signal equivalent to

, and display it on the operator's screen. The received signal is equivalent to the volume (size) of the optimal smoothing buffer, which ensures that there are no losses due to buffer overflow.

Thus, the use of the proposed device will allow us to calculate such optimal values of the smoothing rate and smoothing buffer volume, which ensure the maximization of the correspondence between the average speed and the largest burst of the input aggregated stream, provided that there are no packet losses due to buffer overflow.

This can be useful both in designing new communication systems and in optimizing the parameters of functioning of existing systems.

This eliminates the shortcomings present in the prototype [3], namely:

1. Optimize the real-time stream for which the traffic path is unknown (cumulative function of the input stream arrival).

2. They provide a balanced ratio of smoothing speed and smoothing buffer volume, in which an insignificant change in one parameter does not lead to a significant change in another.

3. Provide maximization of the correspondence of smoothing parameters to the characteristics of the input stream.

4. Optimize the process of transferring any type of traffic.

The description of the utility model presents a functional diagram of a device that contains

block input signal, equivalent to the value of the maximum packet size;

a block for inputting a signal equivalent to an estimate value of the standard deviation of the rate of the input packet stream; input unit of a single signal level;

a signal input unit equivalent to the value of the Hurst exponent; a signal input unit equivalent to the number of received packets; subtraction block;

reciprocal calculation unit;

five multiplication blocks;

division block;

logarithm block;

two exponentiation blocks;

module calculation unit;

square root calculation block;

a block for inputting a signal equivalent to the value of the average rate of the input packet stream;

two addition blocks;

display unit of the optimal smoothing speed;

block for displaying the optimal smoothing buffer, with the organization of links between them.

Each of these blocks is made structurally as part of the FPGA (Field-Programmable Gate Array, User-Programmable Gate Array, FPGA). As is known, an FPGA is a device that initially contains a plurality of logical or arithmetic blocks with the possibility of flexible switching, while changing the switching in accordance with FIG. 1 leads to the formation of a new device that is used to achieve a technical result, which consists in calculating the optimal smoothing rate and smoothing buffer size for various parameters of the communication channel and transmitted traffic of any type according to the criterion of minimum loss due to buffer overflow and maximum compliance with the characteristics of the input stream .

For the constructive implementation of the device for calculating the optimal parameters for smoothing fractal traffic according to the criterion of maximum compliance with the average speed and the largest burst of the input aggregated stream, provided there are no packet losses due to buffer overflow, Xilinx FPGA of the Kintex-7 series is used. The formation of the claimed device is carried out using the Quartus II design environment, where a graphic diagram of the device is created in full accordance with Fig. 1. In this case, each of the blocks used in the device is implemented, respectively, by a divider, multiplier, subtractor, adder, comparator, inverter, etc. All functional blocks included in the claimed device are implemented in the Quartus II environment, in the form of elementary arithmetic-logical units, which are placed on the working form during design with the organization of links between them in accordance with Fig. one.

After flashing the FPGA Kintex-7, a new device is obtained, in which each of its constituent blocks (Fig. 1) is structurally made as part of a gate array.

Initial data input is carried out by transmitting a signal corresponding to the decimal representation of the value of the input parameter, and the result of the device operation is displayed in decimal format on a TFT display connected to the Kintex-7 board.

Information sources

1. RU 2370907, 2004.

2. RU 2341903, 2004.

3. RU 205444 U1, 2021.

4. Jean-Yves Le Boudec and Patrick Thiran. Network Calculus: A Theory of Deterministic Queuing Systems for the Internet. Online version, December 13, 2019.

5. Norros, I. On the Use of Fractional Brownian Motion in the Theory of Connectionless Networks / I. Norros // IEEE Journal on Selected Areas in Communications. - 1995. - No. 13 (6). - P. 953-962.

Claims (1)

A device for calculating the optimal parameters for smoothing fractal traffic according to the criterion of maximum compliance with the average speed and the largest burst of the input aggregated stream, provided there are no packet losses due to buffer overflow, containing a signal input block equivalent to the value of the maximum packet size; a block for inputting a signal equivalent to an estimate value of the standard deviation of the rate of the input packet stream; input unit of a single signal level; a signal input unit equivalent to the value of the Hurst exponent; a signal input unit equivalent to the number of received packets; subtraction block; reciprocal calculation unit; five multiplication blocks; division block; logarithm block; two exponentiation blocks; module calculation unit; square root calculation block; a block for inputting a signal equivalent to the value of the average rate of the input packet stream; two addition blocks; display unit of the optimal smoothing speed; block display optimal buffer smoothing, and the output 1 block 1 input signal equivalent to the value of the maximum packet size, connected to the input 1 block 22 addition; output 1 block 2 input signal, equivalent to the value of the estimate of the standard deviation of the rate of the input stream of packets, connected to the input 1 block 8 multiplication; output 1 block 3 input of a single signal level is connected to the input 1 block 6 subtraction; output 1 block 3 input of a single signal level is connected to the input 1 block 6 subtraction; output 1 block 4 input signal, equivalent to the value of the Hurst exponent, connected to the input 2 block 6 subtraction, input 1 block 9 division and input 2 block 12 exponentiation; output 1 block 5 input signal, equivalent to the number of received packets, connected to the input 1 block 7 calculation of the inverse value; output 1 block 6 subtractor connected to the input 2 block 8 multiplication and input 2 block 9 division; the output 1 of the block 7 calculation of the inverse value is connected to the input 1 of the block 10 of the logarithm; output 1 block 8 multiplication connected to the input 1 block 17 multiplication and input 1 block 18 multiplication; the output 1 of the division block 9 is connected to the input 1 of the squaring block 11 and to the input 1 of the exponentiation block 12; output 1 block 10 logarithm is connected to the input 1 block 13 calculation module; output 1 block 11 squaring connected to the input 2 block 17 multiplication; output 1 block 12 exponentiation is connected to the input 1 block 15 multiplication; output 1 block 13 calculation of the module is connected to the input 1 block 14 multiplication by two; output 1 block 14 multiplication by two is connected to the input 2 of the block 15 multiplication; output 1 block 15 multiplication connected to the input 1 block 16 calculate the square root; output 1 block 16 calculate the square root is connected to the input 3 block 17 multiplication and input 2 block 18 multiplication; output 1 block 17 multiplication is connected to the input 2 block 20 addition; output 1 block 18 multiplication is connected to the input 2 block 22 addition; output 1 block 19 input signal, equivalent to the value of the average rate of the input stream of packets, connected to the input 1 block 20 addition; the output 1 of the addition block 20 is connected to the input 1 of the optimal smoothing speed display block 21; output 1 block 22 addition is connected to the input 1 block 23 display the optimal smoothing buffer.

Images