TWI401900B - Method for partitioning a period according to noise characteristics - Google Patents

Description

Method of dividing cycle according to noise characteristics

The present invention relates to a method for estimating noise in a communication system, and more particularly to a method for dividing a period of noise according to a noise characteristic.

The power line has been a widely distributed line resource many years ago, distributed in various institutions such as families, companies, schools, etc., and due to the advancement of communication technology in recent years, the use of power lines to transmit data has been developed. For Power Line Communication. The technology of power line communication mainly transforms the data to be transmitted into a high-frequency signal, and then couples the high-frequency signal with the low-frequency power signal into the power line to transmit data and power simultaneously.

In the current channel using power line transmission, since the existing alternating current frequency of 50~60 Hz is transmitted in the power line, the alternating current is a periodic sine wave. Therefore, the noise in the channel causing the power line transmission will also appear periodically, which is called cyclostationary noise.

However, in the current power line communication technology, although the noise in the transmission channel has been observed to occur periodically, this feature is not utilized, and multiple time segments are distinguished in the cycle to correspond to different impurities. The characteristics of the signal, and use different noise characteristics to adjust the bit rate of the transmission or the energy of the transmitted signal. Therefore, when the noise is too large in part of the time period, the transmitted signal will be subject to excessive noise interference and the original data cannot be restored.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a method for segmenting a period of cyclically stationary noise based on noise characteristics, and dividing each period into different time segments by detecting changes in noise characteristics.

The invention provides a method for dividing the period according to the characteristics of the noise, comprising: providing a period; when the transmission channel is in a state of no packet, counting a plurality of noise characteristic changes in the i-th cycle; in the i-th cycle, looking for A maximum value of the noise characteristic change value and a corresponding first time point are obtained; and the first time point is used to divide the period into the first time segment and the second time segment.

The invention divides a cycle into a plurality of time segments by detecting a change in the noise energy and a change in the spectrum of the noise, and allows each time segment to correspond to a characteristic of a noise, so that the transceiver can Know the characteristics of the noise at this time. Therefore, the transceiver can appropriately adjust, for example, the bit rate of the transmission or the energy of the transmission signal according to the noise characteristics at this time.

The above and other objects, features and advantages of the present invention will become more <RTIgt;

In the prior art, the characteristics that periodically occur in the transmission channel are not utilized, and multiple time segments are distinguished in the cycle to correspond to different noise characteristics, resulting in transmission when the noise is too large. Loss of information. Therefore, the present invention provides a method for dividing the period of the cyclostationary noise according to the noise characteristic, which can divide the period into a plurality of time segments according to the energy of the noise and the spectrum of the noise, so that the back-end circuit can be based on The time zone is used to properly adjust the bit rate of the transmission or the energy of the transmitted signal.

Hereinafter, a method for dividing the period of the cyclosynchronous noise according to the noise characteristic according to the embodiment of the present invention will be proposed. Before the present embodiment is explained, the transmission channel is assumed to be a power line transmission channel, but the present invention is not limited. In the power line, since there is already one cycle of alternating current, this cycle is called the alternating current line cycle (Mains Cycle or AC Line Cycle), thus causing the occurrence of noise in the power line transmission channel periodically, that is, in this implementation. In the example, the noise in the transmission channel is, for example, a cyclostationary noise. Next, it is assumed that in the present embodiment, the frequency of the alternating current is 50 Hz, and the period of the alternating current power line is 0.02 sec. Also, assume that each AC power line periodically transmits 200 symbols, and each symbol time is 0.0001 sec. In order to simplify the description of the embodiment, the following AC power line periods are simply referred to as periods, and the length thereof is equal to the period of the cyclic stationary noise.

In addition, it is assumed that in the power line transmission channel, the data is transmitted in the form of a packet. Therefore, when the channel is in a non-packet state, the signal received by the receiving end is the noise in the channel. Also, assume that this method is suitable for use in a transceiver that can detect if a packet is being transmitted in the channel. For convenience of description of the embodiment, this embodiment assumes that the channel is in a no-packet state. However, those skilled in the art should be able to infer that since the transceiver can continuously detect whether a packet is being transmitted in the channel, when the channel transmits the packet, the transceiver will temporarily stop the method and wait until there is no packet in the channel. Then continue to start this method.

In addition, it is assumed that the transceiver is applied to an Orthogonal Frequency-Division Multiplexing (OFDM) system. The transmitter includes a Fourier transform unit, and assuming that each OFDM symbol contains 1536 subcarriers, a fast Fourier transform of 1536 points can be performed by the Fourier transform unit to obtain a component of the received symbol at each subcarrier frequency. Next, the method according to the embodiment of the present invention for determining the segmentation period of the noise characteristic will be described. However, the above-mentioned assumed environment cannot be used to limit the present invention, and the present invention can be applied to other types. Communication environment.

FIG. 1 is a flow chart showing the steps of a method for dividing a cycle according to a noise characteristic according to an embodiment of the present invention. Referring to FIG. 1, the segmentation period according to the noise characteristic is started (step S100), and a symbol record table is provided (step S102). In this embodiment, the symbol record table is used to record the start and end time points of each time segment. And because each time segment will probably correspond to different noise patterns, the symbol record table also records the noise type index corresponding to each time segment, and different noise type indexes. Will have different noise patterns. At this time, since the segmentation period has not yet started, the entire AC power line cycle can be regarded as the same time zone, and the noise is regarded as the same noise type in this time zone.

Next, the transceiver receives the symbols in the transmission channel in the i-1th cycle (step S105), wherein the symbol is the sampled received signal, and i is a positive integer. In the present embodiment, the value of i is not limited, that is, the method of dividing the period of the embodiment can be started by any time period. Since the above assumption is made that the channel is in a state of no packet, the signals in the received symbols are all noise. Moreover, for convenience of description of the present embodiment, each symbol in the period is expressed as r ( s ), where s is a symbol index, and since each period contains 200 symbols, s is between 0 and 199.

Thereafter, the transceiver uses the Fourier transform unit to transfer the received symbols to the frequency domain (step S110). In this embodiment, it is assumed that each symbol r ( s ) is sampled and contains 1536 sampling points, and in step S110, 1536 sampling points are subjected to fast FFT conversion at 1536 points, and 1536 is obtained. Subcarrier frequency components. For convenience of description of the present embodiment, each symbol r ( s ) is sampled and fast Fourier transformed is represented as R _s ( k ), and R _s ( k )= FFT [ r ( s )], where k is a sub- Carrier index, and k is between 0 and 1535.

Next, using 1536 subcarrier frequency components, the noise characteristic change value of each symbol in i-1 cycles is counted (step S120), which is expressed as diffOfDist ( s ). In this embodiment, step S120 includes a plurality of sub-steps and is illustrated in FIG. FIG. 2 is a flow chart showing the sub-steps of step S120.

Referring to FIG. 2, first, the transmission channel is divided into a plurality of sub-bands (step S210), that is, the adjacent sub-carrier frequencies are divided into the same sub-band, and in this embodiment, every 64 consecutive sub-carriers It is regarded as the same sub-band, and therefore, 1536 sub-carriers are divided into 24 sub-bands. Thereafter, a memory is provided (step S215), and the memory can be divided into 24 blocks to correspond to 24 sub-bands, respectively. And storing a reference sub-band noise energy in each block, and the reference sub-band noise energy in each block respectively represents the average value of the energy of one sub-band, and the calculation method of the reference sub-band noise energy will be A detailed description is given in the following steps.

In the following steps, each symbol performs the same difference operation, so the following steps take the sth symbol as an example. Next, the calculation of the s-th symbol, the energy of each subband (step S220). Taking the b- th sub-band as an example, the energy of the sub-band is represented as bandEnergy ( b ), and b is an integer between 0 and 23, and the energy of the b- sub-band is calculated as: Formula by using R _s (k), the accumulation of the b sub-band components in the each sub-carrier, and corresponding to the b-th subband subcarrier b * 64 + x, and x is an integer, and between 0 and 63 of between.

After calculating the energy of the sub-band, the energy of each sub-band is subtracted from the reference sub-band noise energy corresponding thereto, and the energy difference value of each sub-band is obtained (step S225). Taking the b- th sub-band as an example, the calculation formula of the sub-band energy difference is, for example, bandEnergy ( b ) -noiseTemplate ( b ), where noiseTemplate ( b ) is the reference sub-band noise energy of the b - th sub-band.

Next, each sub-band energy difference value is accumulated as the s- th symbol band energy difference value (step S230). Taking the b- th sub-band as an example, the arithmetic expressions of the above steps S225 and S230 are as follows: Wherein, dist 2 Tl (s) denotes the s-th symbol band energy difference, the difference value calculation formula described above with reference to the sub-band noise energy per sub-band and stored in the memory are added, and the symbol band the physical meaning of energy difference dist 2 Tl (s) is the s-th symbol difference spectrum stored in the memory to the spectrum.

Thereafter, the calculated energy of the s-th symbol (step S235). The calculation method may be to accumulate the energy bandEnergy ( b ) of each sub-band, or to accumulate each component at the carrier frequency, for example, the operation formula is: Wherein, symbolEnergy (s) denotes the s-th symbol energy.

Subsequently, the s-th symbol energy subtracting (s - C) a symbol energy to obtain the s-th symbol energy difference values (step S240). Physical meaning symbol energy difference is the s-th symbol energy difference between the energy of the symbol one symbol C before its time, and the calculation formula, for example symbolEnergy (s) - symbolEnergy (s - C). Where C is a natural number, the value of which can be preset by the designer of the transceiver, and its physical meaning is the length of transient time required for the change of the noise characteristic.

Thereafter, the accumulation section (s - C) a symbol energy difference of the band to the s-th symbol energy difference divided by the band C, giving the average energy difference between the s-th frequency band (step S245), the calculation may be represented by the formula for . The physical meaning of the average band energy difference is the average of the spectral changes of each symbol from ( s - C ) to s symbol time.

Then, the s-th symbol energy difference being multiplied by a first proportionality constant A plus (1- a first proportionality constant A) s by a first band average energy difference, to obtain the s-th noise characteristic change value ( Step S250), expressed as diffOfDist ( s ), the expression of which is expressed, for example, as follows: Among them, A is a rational number, and its value is determined by the designer of the transceiver according to the change of the symbol energy and the proportion of the spectrum change of the symbol. The physical meaning of the characteristic change in the s-th noise value diffOfDist (s) is the spectral energy variation of the symbol, while when large diffOfDist (s), represents the energy spectrum of the noise changes greatly, and the noise environment Will change.

Next, the reference sub-band noise energy noiseTemplate ( b ) in the 24 blocks in the memory is updated by using the energy bandEnergy ( b ) of the sub-band calculated in step S220 (step S255). The updated method uses, for example, an Auto Regressive operation, taking the b - th sub-band as an example, and the reference sub-band noise energy noiseTemplste ( b ) is calculated as follows: noiseTemplate ( b )=α* bandEnergy +(1 -α)* noiseTemplate' ( b ) where noiseTemplate' ( b ) is the reference sub-band noise energy read in the memory, and noiseTemplate ( b ) is the reference sub-band noise energy updated after autoregressive operation . And α is an autoregressive coefficient, and its value can be adjusted by the designer to determine how fast the reference sub-band noise energy converges. It can be seen from the above mathematical formula that the reference sub-band noise energy needs an initial value before the method has begun yet, and this value can be set by the designer. In each symbol, the reference sub-band noise energy stored in each of the 24 blocks in the memory is updated once. Therefore, in the next step, the separately updated reference sub-band noise energy noiseTemplate (b) is separately stored in the corresponding block in the memory (step S260).

After updating the reference sub-band noise energy, a reference noise energy (step S260) is updated by using the symbol energy symbolEnergy ( s ) calculated in step S235, and is represented as noiseTemplatePower . This reference noise energy noiseTemplatePower is not described in the above embodiments, and its main use will be described later in this embodiment.

In the present embodiment, the reference noise energy is stored, for example, in a temporary memory. Each symbol s calculates a symbolEnergy ( s ) in step S235, and in step S260, the reference noise energy in the register is updated. The updated method is, for example, the same as step S255 described above, using an autoregressive operation to update the reference noise energy, and the calculation is as follows: Among them, noiseTemplatePower' is the reference noise energy read in the scratchpad, and noiseTemplatePower is the reference noise energy updated after the autoregressive operation.

Next, please refer back to Figure 1. After counting the noise variation value of each symbol in the i- 1th period, after the diffOfDist ( s ), find the noise characteristics of all symbols in the entire i-1 period. The maximum value of the change value, and the corresponding time point (step S130), and the maximum value is the global maximum of the plurality of noise characteristic change values. For example, in step S120, the symbol energy and the noise characteristic change value calculated by each symbol are as shown in FIG. 3.

Figure 3 (a) shows the symbol energy corresponding to each symbol. FIG. 3(b) illustrates the change in the noise characteristic corresponding to each symbol. Figure 3 (c) shows the time division and corresponding noise patterns in the period. The abscissas of (a), (b), and (c) of Figure 3 are all symbolic time indices, while the ordinate of Figure 3(a) is the symbol energy, and the ordinate of Figure 3(b) is the noise characteristic. The change value, the ordinate of Figure 3(c) is the noise type index, where the noise pattern index will be detailed in the following steps.

In Fig. 3(b), the global maximum value of the change value of the noise characteristic occurs at the time point of the symbol time s = 103, where the time point of s = 103 is expressed as T _G (i-1), that is, The time point corresponding to the global maximum value found in the i-1th period, therefore, in step S130, the time point T _G (i-1) at which the found global maximum value appears is s = 103. Since the period has not been divided at this time, the noise pattern index is a straight line in FIG. 3(c).

Next, referring back to FIG. 1, in the i-th cycle, the noise characteristic change value of each symbol is counted (step S135). The method of its statistics, as described in Figure 2, is not described in detail.

Then, after calculating the change value of the noise characteristic of each symbol, the local maximum value of the variation value of the noise characteristic is found within a range before and after the time point T _G (i-1), and Its corresponding time point (step S140). In the present embodiment, the above range is, for example, 20 symbol times, that is, step S140 will find the region maximum value of the noise characteristic change value between s = 83 and s = 123. Here, for convenience of explanation of the present embodiment, the time point corresponding to the maximum value of the region found in the i-th cycle is represented as T _L (i).

Then, it is judged whether the time point T _L (i) corresponding to the maximum value of the region falls within a preset time range before and after the time point T _G (i-1) (for example, 10 times before and after the time point T _G (i-1) In the symbol time, that is, s = 93 s = 113) (step S145).

If the time point T _L (i) corresponding to the maximum value of the region falls within a preset time range before and after the time point T _G (i-1), it means that after observing the i-1th and i th cycles, the noise is The changes are all within a preset time range before and after the time point T _G (i-1). Therefore, next, the period is divided into the first time period and the second time period by using the time point T _L (i) (step s150). On the other hand, if the time point T _L (i) corresponding to the maximum value of the region does not fall within the preset time range before and after the time point T _G (i-1), the cycle will not be cut (step S155), and Go to step S105.

For example, in step S135, the symbol energy and the noise characteristic change value calculated by each symbol are as shown in FIG. 4, and the ordinate of FIG. 4(a) is the symbol energy, and FIG. 4(b) The ordinate of the symmetry is the change value of the noise characteristic. The ordinate of Fig. 4(c) is the noise type index, and the abscissas of (a), (b) and (c) of Fig. 4 are the symbol time index.

It can be seen from Fig. 4(b) that there is a region maximum value (just at the time point of s = 103) within the preset time range of the time point T _G (i-1) s = 103, therefore, In step S150, the period is divided into a first time period and a second time period using a time point of s = 103. In FIG. 4(c), the first time segment is s =0~102, and the corresponding noise type index is 1. The second time zone is s = 103~199, and the corresponding noise type index is 2.

Referring back to FIG. 1, after step S150, the symbol record table will be updated (step S157). After updating the symbol record table, the symbol record table is shown in FIG. 5. In Figure 5, there are multiple column fields to store the data for each time segment. Each time segment uses multiple blocks to store its data, and the multiple blocks are the noise type index, start time, end time and next time segment. Taking the first time zone as an example, the data is stored in the first column of the symbol record table, wherein the time zone is 1, the noise type index is 1, and the start time is s =0. Point, the end time is the time point of s = 102, and the next time zone is the time zone of 2 (the second time zone).

It is worth mentioning that, in FIG. 5, a usage indicator is used to indicate the time segment starting in the cycle, and then a link list is used to connect to the next time segment, that is, the use symbol. A block in the next time zone in the meta-record table to indicate the next time zone. In FIG. 5, the indicator is used to point to a block with a time segment of 1, and the next time segment is 2. In addition, an idle indicator is also utilized in FIG. 5 to indicate unused time segments in the symbol record table, and likewise to link the next unused time segment using a link string.

After that, steps S105~S157 are repeated, and the cycle can be continued according to the change of the noise. The only difference is that there are two kinds of noise patterns in the cycle (that is, the noise type index is 1 and miscellaneous). The mode index is 2). Therefore, in the (i+1)th cycle, the reference sub-band noise energy noiseTemplate ( b ) used in step S120 is also divided into two kinds of reference sub-band noise energy and two kinds of reference noise energy due to the noise pattern. That is to say, the noise type index of 1 will correspond to the reference sub-band noise energy of 24 different sub-bands and one reference noise energy. Similarly, the noise type index of 2 will correspond to the reference sub-band noise energy of 24 different sub-bands and 1 reference noise energy.

Therefore, in sub-step S225 of step S120, the symbol record table is used to find the time segment in which the symbol time s is located at this time, and to find the noise type index of the time segment in which it is located, and then Using this noise type index, the 24 reference sub-band noise energy can be found, and the 24 reference sub-band noise energy is used as a reference value to calculate the energy difference of each sub-band. In the sub-step S255 of step S120, the symbol record table is also used to find the time segment where the symbol time s is located at this time, and the noise type index of the time segment in which the time zone is located can be found. The noise type index updates the corresponding 24 reference sub-band noise energy. In addition, in sub-step S260 of step S120, the symbol record table is similarly used to find the time zone in which the symbol time s is located at this time, and the noise type index of the time zone in which the time zone is located is found. The corresponding reference noise energy is updated according to the noise type index.

In addition, in the above step, although the global maximum value in the change value of the noise characteristic is found in the i-1th cycle, and the maximum value of the region in the change value of the noise characteristic is found in the i-th cycle, and The period determines whether or not to divide the period. However, those of ordinary skill in the art should be able to infer that, in accordance with the above steps, it is possible to simultaneously calculate and find the most global in each cycle. The large value and the maximum value of the region, so in each cycle, it is possible to decide whether to divide the cycle according to the time point at which the global maximum value and the regional maximum value appear.

In addition, if in the j-1th cycle after the ith cycle, the global maximum value of a noise characteristic change value is found, and the corresponding time point is expressed as T _G (j-1). And, in the jth cycle, the maximum value of the region within a predetermined time range before and after the time point T _G (j-1) is found, and the corresponding time point is expressed as T _L (j) (at this assumed time point) When T _L (j) is s = 136), the time point T _L (i) is separated from the time point T _L (i) into a third time zone. After the new third time segment is segmented, the symbol record table will be updated as shown in FIG. 6. It can be observed from FIG. 6 that the newly segmented third time segment is located in the third column field in the symbol record table, and has a new noise pattern corresponding to the noise type index 3.

It is worth mentioning that although in the above embodiments, a possible pattern has been drawn for the method of dividing the period according to the characteristics of the noise, those skilled in the art should know that the methods applied to various communication systems are not. Again, the application of the invention is therefore not limited to this possible type. In other words, as long as the maximum value of the noise characteristic change value is found in the period (for example, the region maximum value in the above embodiment), dividing the time segment by using the maximum value is already in accordance with the present invention. The spirit is there.

In the above embodiment, although the period is divided into the first, second, and third time segments, those skilled in the art can understand from the above embodiments that the spirit of the present invention is to divide the time according to the characteristics of the noise. This implementation The first, second and third time segments in the example do not have a certain priority in the actual transmission time, only to express each time segment corresponding to a noise pattern, and between each other The correspondence can be a many-to-one correspondence.

Next, since S105~S157 are repeated, the period will be cut into a plurality of time segments and a plurality of noise patterns. However, in practical applications, due to hardware limitations (such as the capacity of the memory, etc.), the period cannot be divided into excessive time segments and excessive noise patterns without limitation. Therefore, each time the symbol record table is updated, it is determined whether the number of divided time segments reaches a first set value or whether the number of noise patterns reaches a second set value (step S160). The first set value is a maximum value of the number of time segments divided, and the second set value is a maximum value of the number of noise patterns. The first set value and the second set value are determined, for example, by the designer of the transceiver end according to the actual hardware configuration. In this embodiment, the first set value is assumed to be 8 and the second set value is 5.

If the determination in step S160 is negative, the process returns to step S105 to continue the cutting cycle. If one of the determinations in step S160 is YES, a merge step is performed (step S170).

For example, in the m-1 period after the jth period, the change of the symbol energy and the noise characteristic calculated by each symbol is as shown in FIG. 7, and the symbol record table is updated as shown in FIG. Show. As can be seen from FIG. 7 and FIG. 8, the number of time segments is 6, and the number of noise patterns is five. Since the number of noise patterns has reached the second set value, step S170 will be performed. In this implementation In the example, step S170 includes a plurality of sub-steps and is illustrated in FIG. FIG. 9 is a flow chart showing the sub-steps of step S170.

Referring to FIG. 9, the time period is merged according to the noise pattern (step S900). First, the transceiver receives the symbols in the transmission channel at the mth cycle (step S905). In the next few steps (S910~S945), since the steps performed in each time segment are the same, the fourth time segment in FIG. 7 is taken as an example (the symbol time is between s =127). ~141), that is, the time zone in the symbol record table in Fig. 8 is a column of 4.

In step S910, the transient symbols starting from the fourth time segment are excluded, wherein the transient symbols are, for example, the first three symbols starting from each time segment, that is, the symbols excluding the s = 127, 128, 129 time points.

Next, in the fourth time zone, the average symbol energy in a specific segment is calculated (step S915), wherein the length of the specific segment is, for example, 4 symbol times, that is, using the steps of FIG. The calculation method of S235 is to calculate the symbol energy at time points of s = 130, 131, 132, 133, 134, and then add the four symbol energies and divide them by 4 to obtain the average symbol energy in a specific segment.

After calculating the average symbol energy of the fourth time zone, the noise pattern appearing before the fourth time zone is found according to the symbol record table (step S920). The start time of the fourth time zone is s = 127, and the time zone before s = 127 is the first to second time zone, and the corresponding noise type indexes are 1 and 2.

Next, using the found noise pattern, calculating a difference between each of the noise patterns of the fourth time segment (step S925), the difference is calculated by using the average value obtained in step S915. The meta energy is subtracted from the reference noise energy noiseTempl at ePower corresponding to the noise pattern found in step S920 to obtain a difference corresponding to each of the noise patterns. Then, the minimum value of the plurality of differences and its corresponding noise type index are found (step S930), and the noise type corresponding to the minimum difference is the target noise type.

The calculation method of the noise type index corresponding to the minimum value in the above step S925 and step S930 is as follows: Wherein, in the above-described first time period y, for example, the index g is the noise patterns occurring prior to the time corresponding to the sector y, a fourth time period for example, the value of g is 1 and 2, and noiseTemplatePower (g) g noise reference noise patterns corresponding to the energy index is, averageEnergy (y) of the mean energy of the symbol time segments of y. The above mathematical formula will find the noise type index g with the smallest difference as the target noise type index TargetTl.

The above y , g is a natural number, wherein the start time is s =0 due to the first time segment, so there is no time period before the first time segment, so the above y value is between 2 and 6 between.

Next, in the fourth time zone, waiting for a delay time after the specific zone ( s = 130~134) (step S935), and using the first symbol after the delay time to calculate each of its children The energy of the frequency band (step S940), and in the present embodiment, it is assumed that the delay time is one symbol time, so in the fourth time period, the first symbol after the delay time is s = 135, and this The calculation method of the energy of each sub-band in the symbol is as shown in step S220 in FIG.

Thereafter, in the fourth time zone, a combined distance is calculated using the energy of each of the first symbols (the symbols of s = 135) after the delay time (step S945). The calculation method of the merged distance is obtained by using the target noise type index TargetT1 found in step S930, and using a calculation method similar to step S230 in FIG. 2 to obtain the merged distance of the fourth time section. The merge distance mathematical formula is as follows: Where noiseTemplate ( b ) is the reference sub-band noise energy of the b -th sub-band corresponding to TargetTl, and mergeDist ( s ) is the combined distance calculated by the first symbol s after the delay time, and represents the symbol s The merging distance of the time segment, the physical meaning of which is the degree of similarity between the time segment and the target noise type index TargetTl, and the smaller the merge distance mergeDist ( s ), the more the spectrum of the time segment is represented. Similar to the target noise type index TargetTl.

Next, it is assumed that steps S910 to S945 are performed to obtain the merge distance corresponding to each time segment, and therefore, in the next step, the smallest merge distance mergeDist ( s ), and its corresponding time segment will be found (step S950). And setting the noise type index of the time segment having the smallest merge distance mergeDist ( s ) to the target noise type index TargetT1 (step S955), that is, the time having the smallest merge distance mergeDist ( s ) The noise type index corresponding to the segment is merged into the target noise type index TargetTl.

In this embodiment, it is assumed that the sixth time zone (that is, the time zone of the time zone of FIG. 8 is 6) has the smallest merge distance mergeDist ( s ), and its target noise type index TargetTl. For example, it is 3, therefore, the noise type index of the sixth time zone is set to 3.

In the last step, the symbol record table will be updated (step S960) and as shown in FIG. The difference between FIG. 10 and FIG. 8 is only that the noise type index corresponding to the sixth time segment (that is, the time segment is 6 columns) is changed from 5 to 3. After the noise pattern of the fifth time segment is merged into the target noise type index TargetTl, as shown in FIG. Since the noise characteristic change value is not calculated in the merging step S170, the blank in Fig. 11(b) is blank. It can be seen from FIG. 11(c) that the noise type index corresponding to the fifth time zone is set to 3.

Next, after updating the symbol record table, it will return to step S160 in FIG. 1 to continue to determine whether the number of divided time segments reaches a first set value or whether the number of noise patterns reaches a second setting. When the value is.

If the determination in step S160 is negative, then return to step S105 of FIG. 1 to continue the cutting cycle. If one of the determinations of step S965 is YES, then return to step S170 to continue to merge the noise pattern or time zone. segment.

In summary, the present invention divides a period into a plurality of time segments by detecting changes in noise energy and changes in noise in the spectrum, and causes each time segment to correspond to a noise pattern. So that the transceiver can know the type of noise at this time. Therefore, the transceiver can be based on the noise pattern at this time. The bit rate of the transmission or the amount of energy of the transmitted signal is appropriately adjusted.

While the present invention has been described in its preferred embodiments, the present invention is not intended to limit the invention, and the present invention may be modified and modified without departing from the spirit and scope of the invention. The scope of protection is subject to the definition of the scope of the patent application.

S100~S170‧‧‧ steps of the method according to the embodiment of the invention according to the noise pattern segmentation period

Substeps of S210~260‧‧‧Step S120

Sub-steps of S900~960‧‧‧Step S170

FIG. 1 is a flow chart showing the steps of a method according to a noise pattern segmentation cycle according to an embodiment of the present invention.

FIG. 2 is a flow chart showing the sub-steps of step S120.

Figure 3 (a) shows the symbol energy corresponding to each symbol.

FIG. 3(b) illustrates the change in the noise characteristic corresponding to each symbol.

Figure 3 (c) shows the time division in the period and the corresponding noise pattern index.

Figure 4 (a) shows the symbol energy corresponding to each symbol.

Figure 4(b) shows the change in the noise characteristic corresponding to each symbol.

Figure 4(c) shows the time division in the period and the corresponding noise pattern index.

Figure 5 is a symbol record table.

Figure 6 is a symbol record table.

Figure 7 (a) shows the symbol energy corresponding to each symbol.

FIG. 7(b) shows the variation value of the noise characteristic corresponding to each symbol.

Figure 7(c) shows the time division in the period and the corresponding noise pattern index.

Figure 8 is a symbol record table.

FIG. 9 is a flow chart showing the sub-steps of step S170.

Figure 10 is a symbol record table.

Figure 11 (a) shows the symbol energy corresponding to each symbol.

Figure 11 (b) shows the change in the noise characteristic corresponding to each symbol.

Figure 11 (c) shows the time division within the period and the corresponding noise pattern index.

S100~S170‧‧‧ steps of the method according to the embodiment of the invention according to the noise pattern segmentation period

Claims (13)

A method for dividing a period according to a noise characteristic includes the following steps: providing a period; when a transmission channel is in a non-packet state, counting a plurality of noise characteristic changes in the i-th cycle; in the i-th cycle, looking for The maximum value of the noise characteristic change values and corresponding time points are represented as time points T _L (i); and the time points T _L (i) are used to divide the period into a time segment A and a Time section B. The method according to claim 1, wherein the period is divided into M symbol times according to the transmitted symbols. The method according to claim 2, wherein the method according to the noise characteristic segmentation period further comprises the steps of: dividing the frequency band of the transmission channel into N sub-bands; and providing a memory, wherein the memory has N regions. Blocks, and corresponding to N sub-bands, respectively. The method according to claim 3, wherein the step of calculating the noise characteristic period in the i-th cycle comprises the following steps: storing N in N blocks Referring to the sub-band noise energy; defining a first proportional constant A and a second proportional constant B, wherein A, B are between 0 and 1; and at the s-th symbol time, performing a difference operation, The operation steps include: Receiving the energy of the bth sub-band; subtracting the energy of the b-th sub-band from the b-th reference sub-band noise energy to obtain the energy difference of the b-th sub-band; accumulating N energy difference values as the s-th symbol band Energy difference; accumulate energy of N sub-bands as the sth symbol energy; subtract the (s-C)th symbol energy from the sth symbol energy to obtain a s-th symbol energy difference, Where C is a natural number; accumulate the (s-C) symbol band energy difference ~ the sth symbol band energy difference is divided by C to obtain the sth average band energy difference; the sth sign The meta energy difference multiplied by the first proportional constant A plus (1 - the first proportional constant A) multiplied by the sth average band energy difference to obtain the sth noise characteristic change value; the bth subband Multiplying the second proportional constant B by (1 - the second proportional constant B) multiplied by the bth reference subband noise energy as the bth reference subband noise energy; and the bth The reference sub-band noise energy is stored to the b-th block. The method according to claim 1, wherein the dividing the period into the time segment A and the time segment B by using the time point T _L (i) comprises the following steps: In the i-1th cycle, the global maximum value of the noise characteristic change values and the corresponding time point are found, expressed as a time point T _G (i-1); and when the time point T _L (i) is During a predetermined time range before and after the time point T _G (i-1) of the i-th cycle, the period is divided into the time segment A and the time segment B by using the time point T _L (i). The method according to claim 5, wherein the maximum value of the noise characteristic change value refers specifically to a preset time range before and after the time point T _L (i). Find the maximum value of the area. The method according to claim 5, wherein the method according to the fifth aspect of the patent includes the following steps: setting the time segment A to a noise type A; and setting the time segment B to one. Noise type B. The method according to claim 7, wherein the step of dividing the period into the time segment A and the time segment B comprises using the first time point T _L (i); The following steps: providing a symbol record table; in the first column of the symbol record table, storing the start time and end time of the time segment A and the corresponding noise type index; The second column of the meta-record table stores the start time and end time of the time segment B and the corresponding noise pattern index. The method according to the seventh aspect of the patent application, according to the method for dividing the period of the noise characteristic, further includes the following steps: in the jth period after the ith period, finding the maximum value of the variation value of the noise characteristics and the corresponding Time point, the time point is represented as T _L (j); the time zone between the time point T _L (j) and the time point T _L (i) is taken as a time zone C; and the time zone is C is set to a noise type C. The method according to claim 9 is characterized in that: according to the method for dividing the period of the noise characteristic, the time point T _L (j) is between the end time of the period and the time point T _L (i), the time Section B is defined as the time point T _L (j) to the end time of the period. The method for dividing the cycle according to the noise characteristic according to claim 9 of the patent scope further includes: when the time point T _L (j) is between the start time of the cycle and the time point T _L (i), The time zone A is defined as the start time of the cycle to the time point T _L (j). The method according to claim 11, wherein the period is divided into M symbol times according to the transmitted symbols, and each transmitted symbol is divided into N sub-bands in the spectrum. a frequency band, and when the number of divided time segments is greater than a first specific value or the number of noise patterns is greater than a second specific value, the method according to the noise pattern segmentation period further includes the following steps: at j+1 Calculating, in the period, energy of each sub-band of a first specific symbol in the time segment C; calculating the time segment C and the noise pattern by using energy of each sub-band of the first specific symbol The combined distance of A; In the j+1th cycle, calculating the average symbol energy in the time segment B; comparing the difference between the average symbol energy in the time segment B and the reference noise energy corresponding to the noise pattern A, and The difference of the reference noise energy corresponding to the noise type C, and the noise type corresponding to the smaller difference is used as a target noise pattern; in the j+1th period, the time section is calculated The energy of each sub-band of a second specific symbol in B; calculating the combined distance of the time segment B and the target noise pattern by using the energy of each sub-band of the second specific symbol; comparing the time The combined distance between the segment C and its target noise pattern and the combined distance between the time segment B and the target noise pattern; when the combined distance of the time segment C and its target noise pattern is small, Setting the noise pattern of the time segment C to the target noise type; and when the combined distance of the time segment B and the target noise pattern is small, the time segment B is mixed. The mode is set to the target noise type. The method according to claim 9 for the segmentation period according to the noise characteristic further includes the following steps: providing a symbol record table, where each column of the symbol record table includes a start time block and an end time. The block and the next column index block; when the time point T _L (j) is between the time point T _L (i) and the end time of the period: in the first column of the symbol record table, The information of the time segment A is stored, including: storing the start time of the cycle in the start time block; storing the time point T _L (i) at the end time block; storing the time zone in the noise type block An index of a noise type A corresponding to the segment A; and a column in which the time zone C is stored in the next column index block; in the second column of the symbol record table, the time segment is stored The information of B includes: a storage time point T _L (j) at the start time block; an end time of the cycle at the end time block; and a corresponding one of the time segments B stored in the noise type block The index of the noise type B; and the third column of the symbol record table, storing the time Section C of information, comprising: a block of time in the starting point storage time T _L (i); at the end time point of the block storage time T _L (j); in the noise patterns for storing the time zone C block An index of the corresponding one of the noise patterns C; and storing the column of the time section B in the next column index block.