Energy coding gain

For practice the question about expediency of application of error-control codesin telecommunications is important. This question is decided with the taking into account the following. introduction of redundancy by encoding changes not only expenses of a frequency band for transmitting of coded signals, but also demands the account of a redundancy by energy calculations. Really, according to the formula (8.3) for probability of error registration of channel signals (code symbols) is defined by their energy E _s which with taking into account redundancy of a code appears a little bit less energy E _b spent for transfer of one information symbol (bit). It is followed from equality kE _b = nE _s, i.e. E _s = E _b R _code. Therefore in all power calculations of systems with coding using as a rule the value of the ratio of signal energy spent for transmitting of one information binary symbol (bit) to noise power spectral density E _b/ N ₀. The probability of erroneous decoding of the block is defined by formulas (8.1) and (8.3) which in argument of function Q (z) include value E _s – the energy of a signal spent for transmitting through the channel of one binary signal (a code symbol). Really, according to formula (8.3) probability of an error registration of channel signals (code symbols) is defined by their energy E _s which taking into account redundancy of a code, appears a little bit less energy E _b spent for transmitting of one information symbol (bit). Then used in power calculations of systems with coding the relation of energy E _b to noise power spectral density N ₀ can be designated as . Taking into account relation of signal energy E _s and bit energy E _b (8.3) the value entering into settlement formula will be . Then taking into account expenses of energy for transmitting of additional symbols of a redundancy code (8.3) it is possible to present the formula as follows:

, (8.7)

and bit error probability by expression (8.2). If necessary to define probability of an error in channel without coding it is enough to take advantage of the formula (8.3), having put R _code = 1:

. (8.8)

Exercise 8.1 decoding noise immunity of block code.

Let's take advantage of the formula (8.8) for calculations of an error probability with optimum receiving of signals BPSK in channel without coding. Results of calculations are resulted in table 8.1. Initial parameter for calculations is the relation a signal/noise on demodulator input . The used in practice value is defined by formula . In table 8.1 settlement data by definition of error probability by optimum receiving of signals BPSK (formula (8.8)) including argument z of the function Q (z).

The dependence curve p = f (, dB) constructed on these data (BPSK) is resulted on figure 8.1.

Under formulas (8.2), (8.7) we will define of an bit error probability by decoding in the channel with BPSK words of cyclic code average length (31, 26) with parameters R _code =0,84, d _min = 3, q _cor = 1. The code is chosen from table А.1.

Results of calculations are presented on figure 8.1 (curve «Code (31,26)»).

table 8.1 – Calculation of signals BPSK noise immunity

, dB	h b	R code	z	p
	1,12		1,59	5,8∙10–2
	1,26		1,18	3,9∙10–2
	1,41		2,00	2,4∙10–2
	1,59		2,24	1,3∙10–2
	1,78		2,52	6,1∙10–3
	2,00		2,82	2,4∙10–3
	2,51		3,17	7,9∙10–4
	2,82		3,99	2,0∙10–4
	3,16		4,47	4,2∙10–6

table 8.2 – Calculation of a decoding noise immunity of the cyclic code

modulation method BPSK, cyclic code (31, 26)
, dB	h b	R code	z		p d
	1,122	0,84	1,454		0,26
	1,259	0,84	1,632		0,13
	1,413	0,84	1,831		5,4∙10–2
	1,585	0,84	2,054		1,9∙10–2
	1,778	0,84	3,305		5,1∙10–3
	1,995	0,84	2,586		1,0∙10–3
	2,239	0,84	2,902		1,5∙10–4
	2,512	0,84	3,562		1,4∙10–5
	2,818	0,84	3,653		7,2∙10–7
	3,162	0,84	4,099		1,9∙10–8

In all energy calculations of systems with coding it is used as a rule value of the relation of energy of the signal spent for transmitting of one information binary symbol (bit) to power spectral density of noise which is considered as auniform criterion of power expenses for an information transfer through channel with coding and without it. Size change shows efficiency of application of error-control code. The effect of errors decrease on decoder exit can be used on a miscellaneous.

Error-control coding provides reduction of the error probability in the received messages. It is well visible from comparison of curves p = f (E _b/ N ₀) on figure 8.1 for cases of an information transfer by method uncoded BPSK and with using of a cyclic code (31, 26). It is visible, that by using of an error-correcting code it is possible to admit certain decreasein a channel signal/noise ratio and to receive accordingly a energy coding gain g (dB). The energy coding gain from error-control coding g is equal to a difference of values E _b/ N ₀ necessary for maintenance of bit error probability in transmitted data by both absence and using encoding. The value of gain can be defined at various levels of bit error probability p on demodulator and decoder outputs. Told it is illustrated by the curves of a noise immunity presented on figure 8.1.

In particular, for the data resulted on figure 8.1 value gain is g = 1,55 dB at p = 10^‑5.

gain values g is widely used for a choice of codes by designing of transmission systems. Values g received at use of cyclic codes in channels with BPSK are resulted in table А.2 of Attachment А.1.

Example 8.1 Optimisation of a cyclic code parameters.

Let's consider the optimisation procedure of cyclic codes parameters used in the binary symmetric channel with signals BPSK for purpose of a maximum energy coding gain from error-control coding provided by factor of signal spectrum band expanding will not exceed K _{D F} = 2 (double expansion of a signal frequency band in the channel). Preliminary, under table А.2 of Attachment А.1 we will make selection of cyclic codes which can meet requirements on band expansion factor (K _{D F} < 2,
R _code > 0,5). Results of such selection are shown in table 8.3. In table columns values of code rate are specified. in cells in the lines the gain values (in dB) for various lengths of code word n are presented. Under table А.1 of Attachment А.1 we select the cyclic codes with block length n = 255 with rate which is closed to optimum rate R _code = 0,8. It is visible, that the greatest value of gain g = 4,0 dB is reached at using enough long cyclic codes with word length n = 255. In table 8.4 parameters of the optimum cyclic code are shown.

table 8.3 – Parameters of a cyclic codes meeting requirements on a code rate

Word length n	Code rate R code
0,5	0,6	0,7	0,8
	2,7	2,8	2,7	2,1
	3,4	3,5	3,3	2,8
	3,9	4,0	3,8	3,3

Table 8.4 – Characteristics of an optimum cyclic code

n	k	q corr	Code rate R code	Gain g, dB
			0,811	4,0

The selected code (255, 207) provides a power gain 4,0 dB at rate R _code = 0,811. factor band expansion K _{D F} = 1,23 not exceeding preset value max K _{D F} = 2.