Добавил:
Upload Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:

Diss / 10

.pdf
Скачиваний:
143
Добавлен:
27.03.2016
Размер:
18.05 Mб
Скачать

Design Principles of Complex Algorithm Computational Process in Radar Systems

243

means that the operation of one of those algorithms with the functions αij 0 is possible after the algorithm Ai operation. Equation 7.5 is called the transition formula for the algorithm Ai. These formulas can be designed for all elementary DSP algorithms of the complex algorithm given by the matrix flowchart. Thus, in the case of matrix form (7.4), we obtain the following system of transition formulas:

A0 A1;

 

 

A1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

P1P3 A1 + P1A2 + P1P3 A3;

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

P2P3 A1 + P2P3 A3 + P2 A4

;

 

A2

(7.6)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A4;

 

 

 

 

A3

 

 

A4 P4 As +

P4 Ak ;

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ak .

 

 

A4

 

 

The matrix flowchart of the complex algorithm allows us to design the table reflecting both information and controlling association between the elementary DSP algorithms. We may change all the matrix elements by magnitudes

1,

if

αij 0 in (7.2);

 

 

(7.7)

lij =

 

0,

if

αij = 0 in (7.2).

 

 

 

As a result, we obtain the adjacency matrix that reflects formally an information association between the elementary DSP algorithms. For instance, the algorithm given by the matrix in (7.4) has the following adjacency matrix:

 

 

 

 

A1 A2

A3

A4

A5

Ak

 

A0

 

 

 

1

0

0

0

0

0

 

 

 

 

 

 

 

 

 

 

A1

 

 

 

1

1

1

0

0

0

 

 

 

 

 

A = A2

 

 

 

1

0

1

1

0

0

 

 

 

.

(7.8)

A3

 

 

 

0

0

0

1

0

0

 

 

 

 

 

A4

 

 

 

0

0

0

0

1

1

 

 

 

 

 

A5

 

 

 

0

0

0

0

0

1

 

 

 

 

 

We can design the graph flowchart of the complex algorithm of computational process based on the adjacency matrix.

7.2.2  Algorithm Graph Flowcharts

The algorithm graph flowchart is the finite oriented graph satisfying the following conditions:

There are two marked nodes in the graph: the input node corresponds to the operator “Start” and the only arrow is directed from this operator; and the output node corresponds to the operator “Stop” and no arrow is directed from this node.

244

Signal Processing in Radar Systems

One arrow (the node A) or two arrows (the node P) are directed from each node excepting the input and output nodes; the arrows from the node P are marked by the signs “+” and “−” or by the digits “1” and “0.”

The elementary DSP algorithm Ai is correlated with the A-node and the logical operator Pl is matched with each P-node; in the algorithm graph flowcharts, the A-nodes and the input and output nodes are depicted by circles and the P-nodes are depicted by jewel boxes.

The algorithm graph flowchart i.e., equivalent to the given matrix flowchart, i.e., with the same operation sequence, is designed in the following way:

The subgraphs equivalent to transition formulas of the given matrix flowchart are designed.

The equivalent branches of subgraphs are united.

The same operators are united and the final graph flowchart is formed.

To obtain the graph flowchart with the minimal or near minimum number of P-nodes, there is a need to draw the subgraphs with the minimum number of P-nodes, too, using the given transition formulas. The number of P-nodes can be reduced under union of the equivalent subgraph branches. The procedure to design and transform the algorithm graph flowchart can be illustrated by an example considering the matrix flowchart of algorithm given by (7.4) as the initial form. The transition formulas given by (7.6) are initial to design the graph flowchart, based on which the subgraphs for each elementary DSP algorithms A0, A1, A2, A3, A4, A5 must be drawn. The subgraph design process is started from the transition formula transformation to the corresponding form consisting of logical function expansion with respect to each variable. For example, the transition formula for the elementary DSP algorithm A1 given by (7.6) takes the following form:

A1 P1A2 +

 

 

 

 

(7.9)

P1(P3 A3 + P3 A1).

The formula for the elementary DSP algorithm A2 given by (7.6) takes the following form:

A2 P2 (

 

 

 

 

(7.10)

P3 A1 + P3 A3 ) + P2 A4.

Other transition formulas given by (7.6) do not require representation. Now, it is easy to draw the subgraphs of each elementary DSP algorithm (see Figure 7.3).

The next stage is the search and union of equivalent graph branches. In our case, the equivalent branches are the branches starting from the operator P circled by the dash line. After union of the

 

 

A1

 

 

A2

 

 

 

 

 

 

1

P1

0

0

P2

1

 

 

A4

 

 

 

 

 

 

 

A0

A2

1

P3

A4

1

P3

A3

 

 

A5

 

0

 

0

 

P4 0

 

 

A3

A1

 

A3

A1

 

1

 

A1

 

 

A4

A5

Ak

As

 

 

 

 

 

 

(a)

(b)

 

 

(c)

 

 

(d)

(e)

 

(f)

FIGURE 7.3  Subgraphs of each elementary DSP algorithm: (a) the algorithm A0; (b) the algorithm A1;

(c) the algorithm A2; (d) the algorithm A3; (e) the algorithm A4; and (f) the algorithm A5.

Design Principles of Complex Algorithm Computational Process in Radar Systems

245

A1

P1 0

1

1

A2 A3

 

A2

 

1

P3

P2

0

0

A1

A4

FIGURE 7.4  The subgraph uniting equivalent branches.

0

P1

1 A2

P2

1

P3

0

1 A3

 

 

 

0

 

 

 

 

 

 

 

 

As

A1

 

A4

 

P4

0

A5

 

 

 

 

 

1

 

A0

FIGURE 7.5  Final graph flowchart equivalent to the given matrix diagram.

equivalent branches, we obtain the subgraph depicted in Figure 7.4. There are no other equivalent branches for the considered example. Finally, after union of the same counting operators, we obtain the end graph flowchart of algorithm equivalent to the given matrix block diagram (see Figure 7.5). The considered example to design the graph flowchart by the given matrix block diagram allows us to minimize the number of logical operators of complex algorithm, which means to simplify it. The stage of minimization of the operator numbers is the necessary stage for DSP.

Under analysis of features and quality of algorithms presented by the graph flowcharts, it is worthwhile to transform them further with the purpose of uniting the elementary DSP algorithm with the logical operators (in pairs or some elementary DSP algorithm with a single logical operator) into nodes that are used for realization more or less components of the complex algorithm. Two arrows come out from such united node: if the test condition is satisfied as a result of node operation, the arrow is denoted by the sign “+” or “1,” otherwise by the sign “−” or “0.” For instance, in the case of the algorithm with flowchart presented in Figure 7.5, we can unite the algorithms into blocks (the input and output nodes are

not united): a1 ~ A1P1, a2 ~ A2P2 , a3 ~ P3, a4 ~ A1P1, a1 ~ A4P4, a5 ~ A3 A4P4, a6 = A5. The obtained flowchart is depicted in Figure 7.6, where the nodes are designated by the light circles.

 

a2

 

 

0

 

0

1

as

 

 

a4

 

a6

 

1

1

 

1

 

 

0

1

0

a0

 

a5

a1

a3

 

 

 

0

 

 

FIGURE 7.6  Union of algorithms into blocks.

246

Signal Processing in Radar Systems

7.2.3  Use of Network Model for Complex Algorithm Analysis

The main problem solved by the graph flowcharts of complex DSP algorithms is a definition of rational ways to present these problems and a choice of computational software tools and microprocessor subsystems to realize the complex DSP algorithm of a radar system. In short, the problem of optimization of computational process is assigned. The solution of this problem allows us to reduce significantly a realization time and to simplify the complex DSP algorithm of the radar system. Similar problems are the problems of network planning and control [1,2].

To design the network model of computational process under DSP in CRSs, there is a need to transform the graph flowchart of complex algorithm into the network graph that satisfies the following requirements, namely, the network graph cannot consist of contours, that is, the case when the initial top is matched with the final top, and there must be a strict order of top precedence in accordance with the condition that the number of the top i is less than the number of the top j, that is, (i < j) if there is a transition from ai to aj. Henceforth, we call a network of elementary DSP algorithms organized by the corresponding way for CRS signal processing and satisfying all the requirements mentioned earlier as the network model of the complex DSP algorithm. The network tops are interpreted as an “operation” of the corresponding elementary DSP algorithm expressed by the number of computational operations. The network arcs can be interpreted as a sequence of elementary DSP algorithm operations. Network transitions can be deterministic (planned) and stochastic. In the last case, the network model is called the stochastic network model.

The deterministic network model cannot present a complex algorithm functioning in CRSs since it is impossible to predict before a set of elementary DSP algorithms and sequence of realizations for each practical situation. Therefore, the stochastic network model, in which the transitions in the network graph are defined by the corresponding probabilities of transitions given by specific conditions of CRS functioning, is more suitable to image and analyze a realization of the complex DSP algorithm by microprocessor subsystems. When the network model has been constructed, the problem of estimating the time to complete all operations, that is, the time to finish all operations by microprocessor subsystems with the given effective speed of operations, arises. This time cannot be higher than a total duration to finish a complex DSP algorithm operation defined by the most unfavorable way from the initial graph top an to the final graph top ak, that is, along such route that generates a maximal duration of operations. This route is called the extreme route. The extreme route in the stochastic network model cannot be presented in clear form as, for example, in the network model with a given structure. Because of this, the problems of defining the average time or average number of operations required to realize the complex algorithm are assigned under analysis of stochastic network models. To illustrate some principles of designing the stochastic network graph, we consider the following example.

Let the digital signal reprocessing algorithm depicted in Figure 5.13 be considered as the complex DSP algorithm. The network graph of the considered algorithm is shown in Figure 7.7. Special transformations of the algorithm logical flowchart are not needed to design the network graph. Algorithm functioning is started from selection of the immediate target pip of current scanning in the buffer memory device (the elementary DSP algorithm an). Later, the algorithm of target pip coordinate system transformation from the polar coordinate system to Cartesian one (the algorithm a1) is realized. The next stage is a comparison of new target pip coordinates with coordinates of extrapolated target pips of target tracking trajectories (the algorithm a2). If the target pip is inside the gate of selected target tracking trajectory, then it is considered as a prolongation of this trajectory. The updated target tracking trajectory parameters (the algorithm a3) are made more exact, and preparation to send the required information to the user (the algorithm a4) is produced. After that, the graph goes to the final state (the algorithm ak). If the target pip is outside of the gates of none of target tracking trajectories, then the target pip is checked whether

Design Principles of Complex Algorithm Computational Process in Radar Systems

247

a0

a1

 

a2

 

a5

 

a8

 

 

a10

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

P1

 

 

1 – P1

 

 

 

1 – P2

 

1 – P3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

P2

 

 

 

P3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a3

a6

 

 

 

a9

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a4

a7

 

 

 

 

 

 

as

 

 

 

 

 

 

 

 

 

 

FIGURE 7.7  Flowchart of network graph.

it belongs to the detected target tracks (the algorithm a5). If a new target pip is able to confirm one of the detected target tracks, we carry out an accurate definition of target track parameters (the algorithm a6) and test the detection criterion (the algorithm a7). If the target pip is outside of the gates of none of detected target tracks, we assign it to beginning new target tracks (the algorithm a8). When the target pip is caught by one of the primary lock-in gates, the beginning of a new target track is carried out and initial values of the new target tracking trajectory are defined (the algorithm a9). Finally, if the target pip is outside the existing primary lock-in gates it is recorded as the initial point of the new target track (the algorithm a10).

The considered network graph is stochastic without doubt, and there is a need to define the probabilities of transitions between the network graph tops. Let the probability that the selected target pip belongs to the target tracking trajectories be denoted by P1 = Ptr and processed by the elementary DSP algorithms a1, a2, a3, a4, ak. Then,

P1′ = 1P1

(7.11)

is the probability that the target pip does not belong to the target tracking trajectories and has to be processed further. Let PD be the probability that the target pip belongs to the detected target track. Then, the probability that the target pip processing is finished by realization of the algorithms a1, a2, a5, a6, a7, ak is defined as

P2 = (1 − Ptr )PD

(7.12)

and the probability to proceed with a target pip processing further is determined in the following form:

P2′ = (1 − Ptr )(1 − PD ).

(7.13)

In an analogous way, we can obtain

 

P3 = (1 − Ptr )(1 − PD )Pbeg

(7.14)

and

 

P3′ = (1 − Ptr )(1 − PD )(1 − Pbeg ),

(7.15)

where

Pbeg is the probability that the target pip belongs to beginning the new target track

P3is the probability that the target pip will be assigned as the beginning point of the new target track

248

Signal Processing in Radar Systems

The probabilities Ptr, P0, Pbeg depend on the number of target tracking trajectories, detected target tracks, and beginning points of target tracks under processing. For the algorithm of target pip beginning by the criterion (see Chapter 4) we have

The average number of target pips subjected to process under each scanning is determined as

 

 

 

 

 

 

 

 

N

Σ = Nfalse + Ntrue ,

(7.16)

where

Nfalse is the average number of false target pips

Ntrue is the average number of true target pips appearing within the limits of the scanning period

The average number of true target pips coming in the target tracking gates under each scanning is given by

 

 

trgate ,

 

ntruescan = PDgate N

(7.17)

where PDgate is the probability of detection of true target pips within the limits of the target tracking gates, which is the same for all target tracks

The average number of false target pips inside the target tracking gates under each scanning is determined in the following form:

m+ n+ kth 1

 

nfalsescan =

PFscanj Nscanj ,

(7.18)

j= m+ n

where

PFscanj is the probability of the false target pip hit into the jth target tracking gate

Nscanj is the average number of jth target tracking gates for all true and false target tracking trajectories

Taking into consideration (7.16) through (7.18), the probability of an arbitrary selected target pip belonging to one of the target tracking trajectories can be expressed in the following form:

Ptr =

ntruescan

+ nfalsescan

.

(7.19)

 

 

 

NΣ

 

 

In an analogous way, we obtain the formula for the probability of target pip belonging to detected target tracks:

PD =

nDtrue + nDfalse

,

(7.20)

 

 

 

 

NΣ

 

where

 

 

Dtrue

 

nDtrue = PDD N

(7.21)

Design Principles of Complex Algorithm Computational Process in Radar Systems

249

and

m+ n1

 

nDfalse = PDfalsej ND j ,

(7.22)

j= m

where

PDfalsej is the probability of the false target pip hit into the detected target track gates with the number j

PDD is the probability of the true target pip detection into the detected target track gates

ND j is the average number of gates with the number j formed by all the detected target tracking trajectories

NDtrue is the average number of the true target tracks being in the detection process

The probability of target pip belonging to the started target tracks is determined as

 

mn PFlockj

 

 

 

 

 

 

 

 

 

-in N

lockj -in + PDlock-in Ntruelock-in

 

Pbeg =

j=1

 

 

 

 

,

(7.23)

 

 

 

 

 

 

 

NΣ

 

where

PFlockj -in is the probability of false target pip hit into the primary lock-in gates with the number j N lockj -in is the average number of the primary lock-in gates with the number j

PDlock-in is the probability of the true target pip detection into the primary lock-in gates Ntruelock-in is the average number of the primary lock-in gates of true target tracks

Thus, if the statistical characteristics and parameters of noise and target environment inside the radar coverage are known and the target pip beginning algorithm parameters are selected, in the considered case, we are able to determine the probability of transition in the network graph of the complex DSP algorithm. However, we cannot say that this possibility exists forever. In some cases, the probability of transition can be only estimated as a result of computer simulation of the complex DSP algorithm.

7.3  EVALUATION OF WORK CONTENT OF COMPLEX DIGITAL SIGNAL PROCESSING ALGORITHM REALIZATION

BY MICROPROCESSOR SUBSYSTEMS

7.3.1  Evaluation of Elementary Digital Signal Processing Algorithm Work Content

We consider the DSP algorithms realizing the main operations and control in a CRS as the elementary DSP algorithms:

The signal preprocessing stage: DSP algorithms of matched filtering in time or frequency domains, cancellation of passive interferences by the digital moving target indicator (MTI), detection and estimation of target return signal parameters, recognition of type and estimation of interference and noise parameters and ranking samples, etc.

The reprocessing stage: DSP algorithms of target track detection, selection of target pips and their beginning to the target tracking trajectories, target track parameters filtering, transformation coordinate system, and so on.

The CRS control process: DSP algorithms of the scanning signal parameter determination under target searching and tracking, determination of power balance between the modes of CRS operations, etc.

250

Signal Processing in Radar Systems

Each of the aforementioned DSP algorithms is characterized by the work content expressed by the number of arithmetical operations required to realize it. Preliminary computation of the required number of arithmetical operations is possible only if there is an analytical function between the algorithm input and output. In the case of other algorithms possessing a character of logical operations, mainly, the required number of operations can be obtained only as a result of realization based on microprocessor subsystems.

Results of analytical calculation of the required number of arithmetical operations are obtained individually by the number of additions, subtractions, products, and divisions. Henceforth, there is a need to determine the number of reduced arithmetical operations. As the reduction operation, as a rule, an addition is used (short operation). The number of reduced arithmetical operations is determined for each microprocessor subsystem taking into consideration the known ratio between the time to carry out the ith long and short operations, that is, τilong /τshorti . However, the computation of the work content is not finished at this stage since we must take into consideration other nonarithmetical operations.

The DSP of target return signals has a pronounced information-logical character. Logical operations and transition operations are for about 80% of the total number of elementary DSP operations (cycles) in the process of the complex DSP algorithm realization required for CRS functioning. For example, under realization of the digital signal preprocessing algorithm of twocoordinate radar system by microprocessor subsystems with permanent memory, the following number of operations in percentage is required: forwarding or transferring—45%; reduced arithmetical operations—23%; control transfer—17%; shift—5%; logical operations—3%; information exchange—2%; other operations—5%. Consequently, under computation of the work content of the elementary DSP algorithms there is a need to take into consideration the nonarithmetical operations, too. For this purpose, we can introduce the coefficient Kna and write the following relation:

 

 

 

 

 

 

N

i = Nai Kna , Kna > 1,

(7.24)

where Nai is the average number of arithmetical operations required for the ith algorithm. The number of microprocessor operations depends on a programming mode. Under the use of highlevel programming languages, a program length is two to five times greater than that of the optimal program. To take into consideration this fact, we introduce the coefficient Kprog ≈ 2. Thus, (7.24) can be written in this form:

 

 

 

 

 

 

N

i = Nai Kna Kprog.

(7.25)

This work content definition will be used in our next computations.

7.3.2  Definition of Complex Algorithm Work Content Using Network Model

Under analysis of the work content of complex DSP algorithms, we can use the Markov model of computational process or the stochastic network model [3–6]. The stochastic network model allows us to reduce the number of computations under the work content determination in comparison with the Markov model. Therefore, we prefer to use the stochastic network model. As we noted previously, the network graph model of complex DSP algorithm is the initial condition to evaluate the work content. There is a need that this graph would not contain the cycle paths and the graph tops must be numbered in such a way that the graph top number, to which the transition is carried out, will be higher than any graph top number, from which such transition is possible. Moreover, the graph end top must have the maximal number k. An example of such graph satisfying all the previously mentioned requirements is depicted in Figure 7.8.

Design Principles of Complex Algorithm Computational Process in Radar Systems

251

a0 a1 N1

P 12

P 13

a2

 

N2

 

P

 

 

24

 

 

a4 N4

a3

P

34

a5

 

 

 

N3 P35 N5

P

 

 

26

 

 

P46

a6

as

P56

N6

 

 

 

FIGURE 7.8  Example of network graph for complex DSP algorithm.

The average number of addresses to elementary DSP algorithms or the graph tops under a single realization of complex DSP algorithm is denoted by n1, n2, …, nk−1. Since the graph does not contain the cycles, then at a single realization of the complex DSP algorithm, the average number of hits of the computational process to the graph top with the number i is defined as

k −1

 

ni = nj Pji , i = 1, 2,…, k.

(7.26)

j = 0

Under established order of the top numbering at the instant of computation ni, the magnitudes of all previous n1, n2,…, ni−1 are known. The average number of operations under a single realization of the complex DSP algorithm is determined as

k

 

M = ni Ni ,

(7.27)

i=1

where Ni is the average number of the elementary DSP algorithm operations corresponding to the ith top of the complex DSP algorithm graph.

Example: There is a need to define the average work content of the complex DSP algorithm presented by the network graph depicted in Figure 7.8 at the following initial conditions: N1 = 100;

N2 = 30; N3 = 150; N4 = 20; N5 = 200; N6 = 30; P12 = 0.25; P13 = 0.75; P24 = 0.3; P26 = 0.7; P34 = 0.2; P35 = 0.8; P46 = P56 = 1.

1.

Applying (7.26), we obtain

 

 

 

n1 = 1, n2 = n1P12 = 0.25, n3 = n1P13

= 0.75, n4 = n2P24 + n3P34 = 0.225,

 

 

 

 

n5 = n3P35 = 0.6, n6

= n2P26 + n46P46 = 1.

(7.28)

 

 

 

 

2.

Applying (7.27), we obtain

 

 

 

 

 

= 100 + 30 × 0.25 + 150 × 0.75 + 20 × 0.225 + 200 × 0.6 + 30 = 374.5.

(7.29)

 

M

Thus, the average work content of the complex DSP algorithm presented by the graph depicted in Figure 7.8 is 374.5 reduced arithmetical operations.

If the graph of complex DSP algorithm possesses the cycle paths, we cannot apply the considered

procedure to determine the work content M. Owing to this reason, at first, there is a need to exclude the graph cycle paths, that is, to change the graph cycle paths by operator with the equivalent work content. The general procedure to transform the graphs of complex DSP algorithms with the purpose of excluding the cycle paths is discussed in Ref. [7]. As an example, we consider a transformation of the graph depicted in Figure 7.9a. This graph has several cycles different by rank (the order). The cycles that do not contain in the interior none cycle are covered by the rank 1. The number of iterations for this cycle is denoted by n(1). The cycles consisting of one or several cycles with the

252

Signal Processing in Radar Systems

rank 1 are covered by the rank 2. The number of iterations for this cycle is denoted by n(2), and so on. The graph transform is reduced to changes in cycle paths by a single operator. These changes for the graph shown in Figure 7.9a are carried out in accordance with the following formula:

=

{

 

 

+

[

 

 

 

 

 

 

 

 

]

 

 

}

(7.30)

 

 

 

 

 

 

 

N2

N2

 

N3 + (N4 + N5 )n(1) + N6 n(2)

+ N7 r(3).

The resulting graph is shown in Figure 7.9b. Thus, the network model of complex DSP algorithm allows us to define, in principle, the average work content. If we know a realization time of a single reduced operation, then we can compute the average realization time of a complex DSP algorithm. Inversely, if a limitation on the average realization time of complex DSP algorithm is given, we are able to determine the required work content of microprocessor subsystems to realize the given complex DSP algorithm. Sometimes, to solve the problems of computational resource analysis we need to know information about the work content variance. The procedure to define the work content variance is very cumbersome and we do not discuss it in this section.

7.3.3  Evaluation of Complex Digital Signal Reprocessing

Algorithm Work Content in Radar System

We continue consideration and discussion of the example on analysis of the digital signal reprocessing algorithm in CRSs started in Section 7.1. The initial conditions for analysis are given by the graph flowchart of this algorithm depicted in Figure 7.5. In our analysis, we use the following data:

The average number of tracking targets Ntrgate = 80

The average number of target tracks under detection NDtrue = 10

The average number of started target tracks Nbeg = 5

The average number of true target pips assigned as the initial points of new target tracks

Ninitialtrue = 5

Thus, the average number of target pips subjected to processing is equal to NΣ = 100. False target pips and miss of true target pips are not taken into consideration in this example. In accordance with the initial conditions, we can compute the following probabilities:

Identification of new target pip among the target tracking trajectories, Ptr = 0.8.

Identification of new target pip among the detected target track, PD = 0.1.

Identification of new target pip among the begun target tracks, Pbeg = 0.05.

The new target pip is assigned as the initial point of new target track, Pnew = 0.05

N1

N2

N3

N4

N5

N6

N7

1

2

3

4

5

6

7

 

 

 

 

n(1)

n(2)

 

 

 

 

 

 

n(3)

(a)

 

 

 

 

 

 

 

 

 

 

 

 

 

N1

N2΄

 

 

 

 

0

1

S

 

 

 

(b)

 

 

 

 

 

FIGURE 7.9  Example of graph transformation: (a) the graph of cycles different by rank, and (b) the resulting graph.

Соседние файлы в папке Diss