07.01.2013 Views

Lecture Notes in Computer Science 3472

Lecture Notes in Computer Science 3472

Lecture Notes in Computer Science 3472

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

si<br />

1<br />

2<br />

w1 t i<br />

w 2<br />

3<br />

a: <strong>in</strong> MS<br />

si<br />

1 w 1<br />

4 Conformance Test<strong>in</strong>g 105<br />

t i<br />

2<br />

b: <strong>in</strong> MI<br />

Fig. 4.5. Us<strong>in</strong>g two separat<strong>in</strong>g sequences to identify the state<br />

As usual, we first check that MI is similar to MS . We display for each state si<br />

the responses to all the separat<strong>in</strong>g sequences <strong>in</strong> a characteriz<strong>in</strong>g set W (Def<strong>in</strong>ition<br />

4.4). Suppose that W has two separat<strong>in</strong>g sequences w1 and w2. Wewantto<br />

apply the steps shown (<strong>in</strong> square boxes) <strong>in</strong> Figure 4.5 (a) : take MI to si, apply<br />

w1 (step 1), take the mach<strong>in</strong>e back aga<strong>in</strong> to si (step 2) and then apply w2 (step<br />

3). If we observe the right output, we can say that the mach<strong>in</strong>e MI has a state qi<br />

similar to si. We can start from i = 1 and proceed to verify all the states without<br />

us<strong>in</strong>g neither reset nor a dist<strong>in</strong>guish<strong>in</strong>g sequence. The problem is that we do<br />

not know how to br<strong>in</strong>g the mach<strong>in</strong>e MI back to si <strong>in</strong> a verifiable way, because<br />

<strong>in</strong> a faulty mach<strong>in</strong>e, as shown <strong>in</strong> Figure 4.5 (b), the transfer sequence τ(ti, si)<br />

(step 2) may take the mach<strong>in</strong>e to another state s ′ i wherewecouldobservethe<br />

expected output apply<strong>in</strong>g the w2 sequence, without be<strong>in</strong>g able to verify that s ′ i<br />

is si and without able to apply aga<strong>in</strong> w1. We use now the Assumption 6 on page<br />

90, namely that MI has only n states. Let x be an <strong>in</strong>put sequence and n be an<br />

<strong>in</strong>teger, x n is the concatenation n times of x .<br />

Theorem 4.8. Let s be a state of MI , x be an <strong>in</strong>put sequence, o the expected<br />

output sequence produced apply<strong>in</strong>g x to s, i.e. o = λ(s, x ), τ a transfer sequence<br />

from t = δ(s, x ) back to s, and o ′ the expected output produced apply<strong>in</strong>g τ to t.<br />

By apply<strong>in</strong>g the <strong>in</strong>put sequence (x τ) n to state s <strong>in</strong> MI , if we observe the output<br />

sequence (oo ′ ) n , then the mach<strong>in</strong>e ends <strong>in</strong> a state where apply<strong>in</strong>g aga<strong>in</strong> x we<br />

observe the same output o.<br />

s<br />

In MS<br />

x/o<br />

τ/o ′<br />

t<br />

s<br />

xτ/oo ′<br />

xτ/oo<br />

q1<br />

′<br />

In MI<br />

Fig. 4.6. Apply<strong>in</strong>g n times x and τ<br />

s’<br />

i<br />

w 2<br />

3<br />

xτ/oo ′ x/ o<br />

qn<br />

Proof. The scenario described <strong>in</strong> theorem is shown <strong>in</strong> Figure 4.6. Suppose that<br />

MI is <strong>in</strong>itially <strong>in</strong> state s. Apply<strong>in</strong>g x τ the mach<strong>in</strong>e should come back to s.<br />

However, due to some faults, the mach<strong>in</strong>e MI maygotoanotherstateq1 even if

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!