METHOD FOR CONVERTING BINARY NUMBER TO RANDOM NUMBER OR CONVERTING RANDOM NUMBER TO BINARY NUMBER, AND APPARATUS THEREOF

In hereinafter, reference binary random number conversion method and device - the present invention according to attached drawing on, method and device for converting binary random number - herein disclosed.

Figure 6 shows a binary random number conversion device - also the present invention according to one example of a drawing and output structure shown, Figure 7 the present invention according to one embodiment - binary random number conversion device shown in the example configuration are disclosed.

6 and 7 may also reference the, - binary random number conversion device (600) is n (n is a natural number larger than 1) is controlled binary bit, w (a natural number larger than 1) of bits v (a natural number larger than 1) outputs a group consisting of a random number. In hereinafter, heat provided by construction to each other.

- Binary random number conversion device (600) includes producing a coding department which is a distribution (700), group shuffle unit (710) and bit shuffle unit (720) having a predetermined wavelength.

Uniform coding department which is a distribution (700) encodes a plurality of groups including the same bit number input value. The, uniform coding department which is a distribution (710) having respective group 1 (b) or to the number of weights such as encodes portions. For example, the number of second group of 1 g i_i As the, g_i Group g chamber is installed to enclose_i Most small group involves the following relationship is under or over.

Wherein, weight (b) example embodiment is easily the pre-set value. (B) input value when a weight is also 8 is 3 237 is shown a result event transmitter stuck and, (b) input value 1 is stuck when a result event transmitter 72 is also 9 weight is shown disclosed. In the case of Figure 8, it is ten weights when considered 36 bit group 1100 1 3 (=2³ * 3 * 1 + 2² * 3 * 1 + 2 * 3 * 0 + 1 * 3 * 0) representing, group 3 of bit stream 1011 is 33 representing other. I.e., also such as 8, 36, 36, 33, 33, 33, 33, 33 a 237 input values each representing a plurality extracting a group may encode disclosed. For the encoding method described in Figure 9 and also 8 again stuck to each other.

Group shuffle unit (710) is coding department which is a distribution uniformity (700) has a plurality of group optionally group ordering produced by mixing. The embodiment examples LFSR (730) shown in a sequence of a plurality of group using for example a real mixing composite which are not limited to, a sequence of a plurality of group using a variety of method thereof can bear. Small area and low power circuits group shuffle unit (710) in Figure 10 for the described one example of 2000.

Bit shuffle unit (720) is spray-bit sequence of bits has arbitrarily within each group. The embodiment examples LFSR (732) order of bits within each group using for example a real-limited to mixing shown has the composite, each group of bit order using a variety of method thereof can bear. Small area and low power circuits of possible bit shuffle unit (720) in Figure 14 one example of for the described substrate.

For example, 14 also shown in bit shuffle unit (720) optionally mixing using bit order when, weight (b) according to the initial bits set (1400) unlike, uniform coding department which is a distribution (700) input of input values to require an adjustment disclosed. For example, such as also 8, 3 weight (b) using a group 7 stuck when encoding, 258 input values it uniform coding department which is a distribution (700) rather than entering input value (=weight (b)* group number=258 - 3 * 7=237 - input values) adjusted equalizing coding department which is a distribution (700) to inputs.

The present invention according to one example of surface 8 and Figure 9 shows a stuck encoding method also shown also are disclosed.

8 also reference surface, uniform coding department which is a distribution (700) the same number of bits input value 237 encodes a plurality of groups. I.e., since 237=36 + 36 + 33 + 33 + 33 + 33 + 33, 3 weight (b) uniform coding department which is a distribution (810) 237 using a 1100 (representing 36) 1011 and a bit array of group 2 (representing 33) encodes a bit array of group 5.

The number of bits for encoding input value with the number of stuck group within each group can be set to a value various weights a motive. Figure 8 (b) when the weight or 3, 2 or 1 weight (b) is wrong, the number of the number of bits for representing 237 group in the group can be changed. In addition, the weight of Figure 8 (b) is allowed to stand in a 3, the number of bits can be differently setting the group with the number of processors within each group. For example, the number of bits within each group according to the two 5 a, both filled with a most significant bit 0 disapproval.

Uniform coding department which is a distribution (700) has a minimum value and a maximum number of 1 within each group are used to encode reference point such as (α) so weight is constructed to be hereinafter also 8 (850) can efficiently encoding for using hereinafter.

First, uniform coding department which is a distribution (700) comprises a plurality of group arranged longitudinally in a matrix (800) in, reference point (α) (850) based on the row above (802) is filled with all 1 or 0, reference point (850) same row and to the left (804) is 1, 0 right is filled with substrate. The reference point (850) under the rows are reference point (850) is where the gate is filled with 0, reference point (850) including right (806) 1 is filled with substrate. Where 0 is non-matrix of Figure 8 by a goniophotometer.

Reference point (850) 0 or 1 is then matrix caving hereinafter for through each interrupt disclosed. For example, matrix filling both when first rows 1, 3 weight when considered, first row sum is 169 (2³ * 3 * 7) and is, the high value less than 237. The first row 1 both filled with substrate. Both 1 and filled with the second row, a second row sum is 84 (2² * 3 * 7) is under or over. Adding the first row to a second row 84 169, 246 237 is greater than the input value. The second rows filled with both capable of free 1. Reference point (850) moved from left to right while a second row sequentially from the most left 1, 1 [...] side caving reference point is represented each group through the total 237 (850) thereof can locating.

The embodiment reference point (850) encoding method using the present invention is not necessarily limited to one yale stuck provided has the, within each group a number of difference weight hereinafter 1 so that various method may encode can be present.

Reference point (850) when a transmission-, reference point (850) and using of saturation (saturation) digit (820), digit index (830), group index (840) value such as recognize the substrate. Saturated digit (820) a reference folding (850) 0 or 1 to indicate an upper matrix and filled, digit index (830) on group index (840) each reference point (850) exhibits matrix of position. Embodiment of Figure 8, (saturation digit) (group index) - - (digit index) consisting of 1 - 10 - 010 (or 100 - 10 - 010) is stuck code value (ED Code) are disclosed. Group index (840) is then group such as shuffle unit (710) and the like are used.

The reference also 9, a weight 1 76 input values, two group number 15, stuck bit angularly within each group 4 is shown a result event transmitter in 2000. (Α) searches the input value for the reference point 76 (or 010 - 01 - 0001) ED Code 01-01-0001 are disclosed.

Figure 10 shows a group of the present invention according to one example of the method also relates to a method for mixing a plurality group ordering surface are disclosed.

10 also reference surface, group shuffle unit (710) is LFSR (730) and the selection signal generating section (1000) address optionally has a plurality of groups using spray.

Select signal generating unit (1000) is LFSR (730) in Figure 9 and also a value of 8 in his group index (840) receives, can select a plurality of any one select signal (G) occurs. A plurality of group, such as 8 and 9 also reference point (α) also can be region about two subframes. I.e. reference point (α) is located in the front and rear group including reference point (α) of group can be divided into. Select signal generating unit (1000) is about two subframes region reference point (α) unit is either a selection signal generating (G) can be, in one example, steppers (1000) value of the front and rear group 1, 0 can be back rear group. For selecting the number of front group 1 group index (840) determined according.

(B) of Figure 10 is the example of Figure 8 earth group, group index is '010' since, two group (group 0, 1) is located in the front of the reference point (α) for selecting a 1 signal (1010, 1012) any of 7 outputs a clock position twice.

Each group of Figure 10 Figure 11 shows a spray comprises interrogating the group using a shuffle of Figure 8 shown also are disclosed.

The reference also 11, 10 (b) is also a group of selection signal (G) since 0010100, group 1 - 2 - 3 - 4 - 0 - group to the group selection signal according to the shuffling group group group group 5 - 6 sequentially selects group group. Order bit group is blended shuffle unit (720) is outputted to.

Figure 12 shows a logic circuit implemented by the present invention according to group shuffle unit shown one example surface also are disclosed.

The reference also 12, group shuffle unit output of digit index, saturated digit, determined by group selection signal can be know. The highest order bits of groups (A3) outputs the output value of a same logic circuit is of Figure 12 (b) such as can be implemented. In addition the bit output the output value of the logic circuit can be implemented using logic.

Figure 13 shows a sampling of the present invention according to 14 also shown in the drawing and also bit shuffle, one example of the method according to Figure 14 the present invention relates to a method for mixing surface bit within each group are disclosed.

First, with reference to the 13 also, residual symbol definition, (a) input signal and output signal M N X, Y output circuit, when the output signal X 1 is S, Y input signal M, N is equal to, when X is 0 S output signal, the input signal M Y, N the out side is outputted. (B) the same symbol represented more simple such as disclosed. If S 0 and 1 the same nth output is 50% probability, two input signal M, N selected in any sequence.

Figure 14 shows a of Figure 13 (b) defined symbol represented using bit shuffle unit (720) one example of are disclosed. The reference also 14, bit shuffle unit (720) (b) a weight of each bit in the group number equivalent probability set considering prevention (Equal Probability, EP set) (1410, 1420, 1430) is defined. For example, weights and (b) is 2, when group 3 bits, 2 most significant bits² A EP set angularly garment number * 2=8 (1430) creates, then 2 bits¹ A EP set angularly garment number * 2=4 (1420) creates, angularly garment number 2 last bits 160,170 EP set (1410) is defined. I.e. the most significant bit value of the back 1, 1 is the EP set angularly garment number 8 (1410) is generated. In addition bit shuffle unit (720) is an initial EP set corresponding weight (1400) comprises. In the case of Figure 14, since the weighting 2, initial EP set (1400) consists of 2 is of value.

Bit shuffle unit (720) are each other EP set bits of one-to-one mapping to each other. One-to-one mapping to the most significant bit EP set method (1430) EP set of bits of the least significant bit (1420, 1410) and initial EP set (1400) and a plurality of and one-to-one mapping, second bit EP set (1420) then EP set of bits of the least significant bit (1410) and initial EP set (1400) of each bit and one-to-one mapping to each other. The method can be one-to-one mapping of the other face of the EP set bits, using symbols in fig. 14 of Figure 13 the same diagrams such as disclosed.

Bit shuffle unit (720) has a one-to-one mapping bits are output sequence has arbitrarily spray. I.e., look in Figure 13 as one-to-one mapped bits LFSR (732) 50% probability by a bit output value calculator determines whether the order of correction and whether it as it is. The bits are order of each mapping optionally blended substrate.

EP set bits in each group (1410, 1420, 1430) in addition to the initial EP set (1400) exist, initial EP set (1400) bit shuffle of the output bit stream by the number of processors increases, uniform coding department which is a distribution (700) adjusting an input value input needs to be disclosed. In the embodiment, weighted 2 initiated EP set (1400) and two bits, if two group number m which, total 2m further bit is outputted.

The uniform coding department which is a distribution (700) initial EP set input an input value (1400) in view of the need to adjust, representing the same expressions as follows.

Wherein, L input value, b is weight, number of group v, additionally includes a signal line S represents the value of a substrate.

E.g., such as 8 also, weight 3, group number 7 stuck when angularly encoding, input value input value back 258 group number of product weight (21=3 * 7) is essentially larger than than, uniform coding department which is a distribution (700) 258 258 - 21=237 which is input value input value by using a thickness of 3 nm to 150.

The present invention according to the embodiment of Figure 15 shows a binary random number conversion method - also shown in example flow are disclosed.

The reference also 15, random number conversion device - binary input value (S1500) encodes a plurality of bits of a plurality of groups. The maximum number of each group is contained in a binary random number conversion device - 1 encodes difference so weight hereinafter. The reference point can be also shown in 8 and 9 also encoding using.

- Binary random number conversion device (S1510) has respective group optionally mixing group ordering. The device 10 also includes a random number conversion such as binary - LFSR (730) group for using the index to each group optionally obtained in Figure 8 and a selection signal thereof can mixing group order.

After mixing group ordering, random number conversion device includes a binary bit sequence optionally mixing (S1520) has - within each group. More specifically, binary random number conversion device within each group is - bit for a number EP set to become useful to produce, after different EP set values of one-to-one by mapping, mapping bits are output sequence outputs arbitrarily determining (S1530).

Figure 16 shows a binary random number conversion method - also the present invention according to one example of result accuracy of experiments show surface are disclosed.

The reference also 16, 0 input values. 6 2 10 bit LFSR for use⁵ -2¹⁰ Bit accuracy at the time of generating a random number is shown disclosed. (A) conventional method is a random number of invitation a minimum error is yet, the present invention according to the binary random number conversion method can be relatively small error - personnel to know.

Figure 17 shows a binary random number conversion device - it is also implemented by the present invention according to one example of the experiment result shown in a circuit are disclosed.

The reference also 17, Conv. Conventional method, conventional method unique to the LFSR Shared method, Prop. The present invention according to method is by a goniophotometer. The present invention according to binary random number conversion device - circuit area, power consumption or the like can be know better than conventional.

Figure 18 shows a binary conversion device shown in the drawing and also to determine the one example of the present invention according to random number -, - details of one example of a binary conversion device configuration shown in Figure 19 the present invention according to the random number are disclosed.

Binary conversion device in random number - 1 is the number of cells constituting a bit string serves a plurality of hierarchies. For example, when distilled 11011110 bits, binary conversion device 6 - 1 is the number of cells by random number among other information. The bit stream among the gross bit number 8, value of 6/8 can be achieved.

Probability can be effectuated calculated approximate exponential method outputs, the number of cells organized as existing counter 1 does not accurately, the present invention according to binary conversion device includes a random number - approximate exponential method have diameters less than 2000. In particular the embodiment example circuit area, such as etched structure for power consumption efficiency in number.

The reference also 18 and 19 also, random number - (Approximate Unit, AU) binary conversion device is greatly approximate unit (1800) and a parallel counter (paralle conuter, PC) (1810) consists of to. A layer (layer) is formed on the gate unit 12 AND OR consists of alternately arranged at least include one or more of the. (B) of Figure 19 includes a layer SDA, (c) and (d) when each layer exhibits two and three. Layer increases and increased whenever AND OR gate hierarchies. (C) OR AND gate error on the mutually complementary, (d), (e) positioning the offset error such as AND and OR gate alternately equal to or less than. A parallel counter (1810) is the output value of 1 for AU beam to a subject substrate.

Figure 20 shows a circuit implemented by the present invention according to binary conversion device random number - also illustrates one example of result when performance of a surface are disclosed.

20 also reference surface, the present invention according to the complementary binary random number - (Prop) conversion device, circuit delay time (critical path delay) and power consumption and the like can be know treatment for conventional (Conv) from the parallel.

Figure 21 shows a binary conversion method the present invention according to one example of the experiment result shown in accuracy of a random number - also are disclosed.

The reference also 21, 1024 bit row 1 (x axis) has a smaller number of error (y axis) is shown disclosed. (A) the reference of Figure 21, and the average error 0 for each entity, a standard deviation of about maximum 12 can be known. Of Figure 21 (b), (c) the reference, 1 when the number of error probability distribution to each 512, 128 is shown, each standard deviation 11. 32 and 9. 40 are disclosed. The 95% confidence interval of 2 in error. 2% and, in errors in 70% confidence interval is 1. Since 1%, the value of counting can be relatively accurate than that of the present invention according to method 1.

The present invention refers to in addition to computer-readable recording medium storing a computer-readable codes process from J4. The computer-readable recording medium which can be read by the computer system where the data is stored for all kinds of recording device without using a tool. Examples of computer-readable recording medium ROM, RAM, CD BD-ROM, magnetic tape, floppy disk, optical data storage device etc.. In addition the computer system connected to the computer-readable recording medium storing a computer-readable code is stored and executed in a distributed manner dispersed can be.

The number into the tank by the present invention to the preferred embodiment the transformed for flaws. The present invention is in the field of the present invention is provided essentially from deviating from a person with skill in the art of the present invention is embodied in the form of modified inputted properties may be understand it will rain. The definitive aspect as well as the descriptive disclosure in the embodiment are contemplated aspect should. The aforementioned range of the present invention description and claim rather than as shown, and the present invention is in a range equal to all differences may be carried on an will be interpreted.