You are on page 1of 4

Worksheet 2

The safety of the data transfer using a communication channel affected by noise.

The coding efficiency

Problem:

For increasing the safety of the data transfer on a channel that is affected by noise we apply
the coding technique. Thus, we concatenate control bits to the data bits obtaining an
extended word of = + bits.

The control code allows the correction of one bit from the transmitted word.

We denote by the error probability when a bit is transmitted through the communication
channel.

Considering that for each sequence of bits the errors that may appear represent independent
events, we have to determine the probability of errorless transfer of a word.

Evaluate the coding efficiency .

Solution:

First we analyze the data transfer without coding. In this case the transfer probability without
error of one word is = 1− .

If we use the code we have:

= 1− + ∙ ∙ 1− = 1− + + ∙ ∙ 1−

SEC_SED analysis method

For a Hamming coding scheme type SEC_SED (single error correcting – single error
detecting) if = 2 than = + 1. For example, for transmission of an octet 4 control bits
are necessary.

The coding efficiency is evaluated using the ratio between the probability of transmission of
an erroneous word in non-redundant system and the probability of transmitting a word in the
redundant version of the system.

1−
=
1−

The numerical computation for = 10 and = 10 and = 8, 16, 32 bits when


" _" $ method is used is given in the tables 1 and 2. For coding case the overloaded of the
communication channel is % = / ∙ 100 (%)
Table 1. Error probability: = 10 ; error detection method: " _" $
M %= / ∙ 100 %
8 0.9920 0.(9)43 121.6 4 50
16 0.9841 0.9997 76.55 5 31.25
32 0.9685 0.9995 45.91 6 18.75

Table 2. Error probability: = 10 ;error detection method: " _" $


M
8 0.9992 0.(9)63 1212.5
16 0.9984 0.(9)57 762.3
32 0.9968 0.(9)52 455.579

Observations:

- The lower values from the first column demonstrate the necessity for introducing coding.
For example the probability of a successful transfer of 1 % is = 0.992 +, = 0.00026
for = 10 and = 0.9992 +, = 0.44 for = 10 .

- The redundancy efficiency increases with the reduction of noise influence. The coding
efficiency is greater when is lower which means that the quality of the communication
channel is better.

- The coding efficiency is higher for low values of . (for words shorter in length)

- The above observation is justified by the fact that the weight of control bits in the extended
word of length + decreases as long as increases. Thus, the error tolerance is lower.

SEC_DED analysis scheme

We analyze a coding scheme type SEC_DED (single error corrected – double error detected)
that allows the detection of two errors in the extended word.

Using a Hamming coding scheme, if = 2 then = + 2.

Table 3. Error probability: = 10 ; error detection method: " _$ $


m %= / ∙ 100 %
8 0.9920 0.(9)42 102.9 5 62.5
16 0.9841 0.99977 69.7 6 37.5
32 0.9685 0.99927 43.6 7 21.875

Table 4. Error probability: = 10 ; error detection method: " _$ $


m
8 0.9992 0.(9)62 1026
16 0.9984 0.(9)57 693
32 0.9968 0.(9)52 432
The reliability of the communication channel decreases slightly over the SEC_SED method.
On the other hand increases the security of the data transfer by identification the words that
include two uncorrectable errors. Thus, the probability of a word transfer without error, with
one or two correctable errors is:

. , ,
= 1− + 1− + 1− = 1− + + 1−
,
, ,
+ ,
1−

.
1−
= .
1−
. .
Table 5. and Efficiency for = 10 ; and the efficiency for = 10

M . M .
1− 1−
.
/ 1− / 1− .
8 0.(9)67 28084 8 0.(9)97 2798323
16 0.(9)58 10460 16 0.(9)88 1.03∙ 10/
32 0.(9)51 3542 32 0.(9)80 3.5∙ 100
By the retransmission of the erroneous words the reliability of the communication channel can
be improved near the level.

The load of the communication channel increases when by retransmission of the words with
errors.

Multiple bit transmission

Another technique for increasing the safety of the data transfer on the communication
channels affected by noise is the multiple transmissions of the same bit and implementation of
the decoding logic at the receiver side based on majority wins.

For example when each bit is transmitted = 3 times the probability of the errorless transfer
of a bit is:

1 = 2 + 32, 1 − 2 = 32, − 22 where 2 = 1 − .

The probability of transmitting one word of bits without error is


,/
= 1 = 32, − 22

The efficiency for two values of are presented in the table:


,/ ,/
Table 6. and efficiency for = 10 ; and efficiency for = 10

M ,/ M ,/

8 0.(9)47 332.3 8 0.(9)67 3332


16 0.(9)45 331.1 16 0.(9)65 3331
32 0.(9)40 328.4 32 0.(9)60 3328
Observations:

The channel reliability is higher than the obtained reliability with the SEC_SED method, but
with the price of significantly increasing the channel overloading. The reliability dependency
with the word length is almost negligible.

Homework

For = 5 ∙ 10 determine the transfer probability without error of a data bloc of 1024 bits
without coding, using SEC_SED coding method and using voting 2 out of 3. The dimension
of the data blocks (of words) must take into account the quality of the communication
channel. If the data blocks are too large, no coding method can eliminate the noise influence
that affects the transfer.

For = 5 ∙ 10 determine the transfer probability without error of a data block of 1024 bits
without coding and with voting 2 out 3.

The probability of errorless transfer without coding is:

= 1− = 0.5992

The probability of transfer without error with SEC_SED coding:

= 1− + + 1− = 0.9044

= 1024; =2 ; = + 1 implies = 10 and = 11

The probability of errorless transfer with voting 2 out of 3 is:


,/ ,
= [3 1 − −2 1− ] = 0.9992

You might also like