JP7469776B2 - Communication device, communication method, and communication program for performing propagation path equalization and quality estimation - Google Patents

Description

The present invention relates to a communication device, a communication method, and a communication program that train an adaptive equalizer that performs propagation path equalization and estimates communication quality.

General communication devices are equipped with an adaptive equalizer that performs propagation path equalization. The adaptive equalizer needs to estimate the propagation path characteristics and calculate parameters for equalizing inter-symbol interference, etc. For this purpose, a known training signal is transmitted from the transmitter, and the receiver calculates parameters so that, for example, the error between the received training signal and a training signal stored in advance converges to a minimum, and sets the parameters in the adaptive equalizer.

In general wireless communications, training signals exist within the frame format as synchronization signals. In particular, in TDMA communications, training signals are often used to synchronize the timing of received signals.

On the other hand, communication systems need to manage communication quality, and communication devices have the function of estimating the communication quality of a propagation path. For example, a transmitter transmits a known signal called a Sync word signal, and a receiver estimates communication quality (e.g., SINR (Signal to Interference and Noise power Ratio)) based on the received Sync word signal. For example, a technology is used in communication systems to transmit an SINR measurement signal to measure communication quality (see, for example, Patent Document 1).

International Publication No. 2009/019759

However, in order to measure communication quality using an SINR estimation signal, training using the training signal must be completed, and the SINR estimation signal must be transmitted after the training signal.

The number of bits required for the adaptive equalizer to converge using the training signal varies depending on the situation (initial values, noise, etc.). If the bit sequence of the training signal is a fixed length, if the adaptive equalizer converges too early on the first half of the bit sequence, the remaining bit sequence will be wasted, reducing the available communication capacity.

On the other hand, a certain number of bits is required to estimate the SINR, and when noise is particularly high, it takes a large number of bits to complete the SINR measurement. Furthermore, since the SINR is displayed in dB, a small number of bits are required to display the dB value when the propagation path has a large amount of noise relative to the signal, but a large number of bits are required to display the dB value when the propagation path has a small amount of noise relative to the signal.

In addition, when the adaptive equalizer has sufficiently converged, it is necessary to perform SINR estimation with high accuracy, so the bit string of the SINR estimation signal becomes long. If the bit string of the SINR estimation signal is of fixed length, the bit string of the SINR estimation signal must be set according to an environment that requires a large number of bits.

The bit string of the training signal of the adaptive equalizer and the bit string of the SINR estimation signal are both known signals, but their importance is different. It is essential that the parameters of the adaptive equalizer converge in order to perform communication, but the number of bits required for convergence varies depending on the initial values and the transmission path environment. In addition, when measuring the SINR of the output signal of the adaptive equalizer, errors will be large unless the adaptive equalizer has converged to a certain extent. For this reason, when setting bit strings of the required length for each of the training signal and the SINR estimation signal as fixed-length fields, it is necessary to set bit strings that take into account various situations for both, which results in longer bit strings and less available communication capacity.

The present invention aims to provide a communication device, a communication method, and a communication program that perform propagation path equalization and quality estimation that can reduce the total number of bits of the known signal to be transmitted and make effective use of frequency resources by using a single known signal that combines the training signal of an adaptive equalizer and an SINR estimation signal.

The communication device according to the present invention is characterized in having an adaptive equalizer that equalizes distortion of the propagation path characteristics contained in a received signal using parameters calculated by estimating the propagation path characteristics; a parameter calculation unit that estimates the propagation path characteristics based on a known signal of a predetermined length received from a transmitter and calculates the parameters of the adaptive equalizer; a training end determination unit that determines the end of training of the adaptive equalizer based on a convergence status of the parameters calculated by the parameter calculation unit; and a communication quality estimation unit that , when the training end determination unit determines the end of training, estimates communication quality of the propagation path using a remaining portion of the known signal other than the portion used for training .

Furthermore, a communication method according to the present invention is characterized in that it executes an adaptive equalization process for equalizing distortion of propagation path characteristics contained in a received signal using parameters calculated by estimating propagation path characteristics; a parameter calculation process for estimating propagation path characteristics based on a known signal of a predetermined length received from a transmitter and calculating the parameters to be used in the adaptive equalization process; a training end determination process for determining end of training of the adaptive equalization process based on a convergence state of the parameters calculated in the parameter calculation process; and, if end of training is determined in the training end determination process , a communication quality estimation process for estimating communication quality of the propagation path using a remaining part of the known signal other than the part used for training .

The communication program according to the present invention is characterized in that it causes a computer to execute the processes performed in the communication method.

The communication device, communication method, and communication program for performing propagation path equalization and quality estimation according to the present invention use a single known signal that combines the training signal of the adaptive equalizer and the SINR estimation signal, thereby reducing the total number of bits of the known signal to be transmitted and making effective use of frequency resources.

FIG. 1 is a diagram illustrating an example of a configuration of a communication device according to an embodiment. FIG. 1 illustrates an example of a conventional training signal and an SINR estimation signal. 1 is a diagram illustrating an example of a known signal in a communication device according to an embodiment. FIG. 13 is a diagram illustrating an example of the relationship between convergence speed and signal length. FIG. 13 is a diagram showing an example of the required signal length and occurrence probability with respect to convergence speed in training and SINR estimation. FIG. 13 is a diagram illustrating an example of a combination of the convergence speed of training and the convergence speed of SINR estimation. 1 is a diagram illustrating an example of a known signal in a communication device according to an embodiment. FIG. 13 is a diagram illustrating another example of the configuration of a communication device.

Below, an embodiment of a communication device, a communication method, and a communication program for performing channel equalization and quality estimation according to the present invention will be described with reference to the drawings.

The communication device described in the following embodiments transmits a single known signal that combines a training signal and a communication quality estimation signal, and can continuously train the adaptive equalizer and estimate the communication quality. This reduces the length of the known signal to be transmitted, enabling effective use of frequency resources. Here, in the embodiment, the communication quality is obtained by estimating the SINR of the signal equalized by the adaptive equalizer.

Figure 1 shows an example of the configuration of a communication device 100 according to an embodiment. In Figure 1, the communication device 100 has a receiving unit 101, an adaptive equalizer 102, a decoding unit 103, a training end determination unit 104, and an SINR estimation unit 105. Here, the SINR estimation unit 105 corresponds to the communication quality estimation unit.

The receiving unit 101 receives the signal transmitted from the transmitter via the propagation path. For example, in the case of a wireless system, the receiving unit 101 receives the wireless signal transmitted from the transmitter via an antenna and performs processing to convert it into a received signal. Here, the received signal may be a known signal that combines a training signal and a communication quality estimation signal, a data signal for communicating user data, etc.

The adaptive equalizer 102 has an equalization filter 201 and a parameter calculation unit 202, and performs processing to equalize distortion in the propagation path characteristics (corresponding to adaptive equalization processing).

The equalization filter 201 is a filter that equalizes the distortion of the propagation path characteristics contained in the received signal using the parameters calculated by the parameter calculation unit 202 as coefficients, and is configured, for example, as a transversal type filter. Here, as described above, the transmitter transmits one known signal in which the training signal and the SINR estimation signal described below are concatenated.

The parameter calculation unit 202 performs a process of calculating the coefficients (parameters) of the equalization filter 201 using an algorithm (e.g., LMS (Least Mean Square)) that minimizes the error between the known signal after processing by the equalization filter 201 and a known signal previously stored inside (corresponding to the parameter calculation process). The parameter calculation unit 202 then outputs information indicating the convergence status of the parameters to the training end determination unit 104. Note that the information indicating the convergence status may be, for example, an error at the time of parameter calculation or a parameter value. Note that the equalization process may be performed in the frequency domain. In this case, the parameter calculation unit 202 determines the transfer function of the propagation path using the known signal output by the receiving unit 101, and multiplies the received signal by the transfer function of the inverse characteristic as a parameter in the equalization filter 201.

The decoding unit 103 performs processing to decode the received signal equalized by the adaptive equalizer 102 into received data using a predetermined decoding method.

The training end determination unit 104 performs a process of determining the end of training based on information indicating the convergence status received from the adaptive equalizer 102 (corresponding to the training end determination process). Then, when the training end determination unit 104 determines that the training has ended, it instructs the SINR estimation unit 105 to start the SINR estimation process. For example, if the convergence status is an error during parameter calculation, training is determined to have ended when the error converges to a predetermined threshold value or less. Alternatively, if the convergence status is a parameter value, training is determined to have ended when the fluctuation of the parameter value is within a predetermined threshold value range. Note that the trigger for starting SINR estimation is not limited to the above example, and other conditions may be used as long as they can determine the end of training.

The SINR estimation unit 105 performs a process of estimating the communication quality of the propagation path based on the SINR estimation signal received from the propagation path (corresponding to communication quality estimation process). Here, the estimation of the communication quality of the propagation path is performed, for example, by estimating the SINR of the received signal. As described above, the transmitter transmits a known signal in which the training signal and the SINR estimation signal are combined. The SINR estimation unit 105 starts the SINR estimation process in response to a start instruction from the training end determination unit 104, and estimates the SINR using the remaining part of the known signal at the timing when the adaptive equalizer 102 ends training. The SINR estimation result is used to control the communication device 100, such as switching the transmission method based on the communication quality. In the embodiment, a description of the application such as control using the SINR estimation result is omitted.

In this way, the communication device 100 according to the embodiment uses one known signal that combines the training signal of the adaptive equalizer 102 and the communication quality estimation signal of the SINR estimation unit 105. Then, the end of training is determined based on the convergence status of the adaptive equalizer 102, and the SINR estimation unit 105 starts processing at the point when the training ends. The SINR estimation unit 105 estimates the SINR based on the remaining signal after the training of the adaptive equalizer 102 is completed among the known signals. In other words, regardless of the speed of convergence of the training, the training and SINR estimation are performed consecutively. As a result, the length (total number of bits) of the known signal to be transmitted is reduced compared to the case where the training signal and the SINR estimation signal are transmitted separately, and frequency resources can be used effectively.

Here, in the embodiment, the length is described as the length of signals such as a training signal, an SINR estimation signal, and a known signal, but since each signal is composed of multiple predetermined bit strings, the length of the signal corresponds to the length or number of bits of the bit string.

Figure 2 shows an example of a conventional training signal and SINR estimation signal. In Figure 2, the horizontal axis indicates time. In Figure 2 (a), conventionally, a fixed-length training signal and a fixed-length SINR estimation signal are separately transmitted from the transmitter, and training of the adaptive equalizer 102 and SINR estimation of the SINR estimation unit 105 are separately performed on the receiving side. In the example of Figure 2 (a), the training signal is transmitted from time T1 to time T2 regardless of the convergence status of the adaptive equalizer 102. Similarly, the SINR estimation signal is transmitted from time T3 to time T4 separately from the training signal. Note that time T2 and time T3 may be the same, but the length of the training signal and the length of the SINR estimation signal are fixed.

Here, the time it takes for the parameters of the adaptive equalizer 102 to converge due to the training signal varies depending on various circumstances (initial values, noise, etc.). For this reason, in the past, the longest training signal expected was used so that training would be completed in various circumstances. However, if the training signal is of fixed length, if the training of the adaptive equalizer 102 converges early, the remaining training signal after convergence is not used effectively, resulting in a problem of wasting frequency resources. Similarly, for the SINR estimation signal, the longest SINR estimation signal expected was used so that SINR estimation would be completed in various circumstances. However, if the SINR estimation signal is of fixed length, if the SINR estimation converges early, the remaining SINR estimation signal after convergence is not used effectively, resulting in a problem of wasting frequency resources.

For example, in the case of FIG. 2(b), similar to FIG. 2(a), the training signal is transmitted with a fixed length from time T1 to time T2. However, since the parameters of the adaptive equalizer 102 converge at time T12, the training signal in the period from time T12 to time T2 becomes useless. Similarly, the SINR estimation signal is transmitted with a fixed length from time T3 to time T4. However, since the SINR estimation process of the SINR estimator 105 converges at time T34, the SINR estimation signal in the period from time T34 to time T4 becomes useless.

Therefore, the communication device 100 according to the embodiment uses one known signal that combines the training signal and the SINR estimation signal. Then, when the training is completed, the remaining known signal is used to perform SINR estimation. In particular, the communication device 100 according to the embodiment uses a known signal that is shorter than the total length of the longest required training signal and the longest required SINR estimation signal based on the occurrence probability of the convergence speed, thereby making it possible to reduce wasted frequency resources. Note that a method for shortening the known signal based on the occurrence probability will be described later.

In this way, compared to conventional methods that ensure surplus signals with a low occurrence probability in each of the training signal and the SINR estimation signal, the communication device 100 according to the embodiment is capable of reducing surplus signals probabilistically.

Figure 3 shows an example of a known signal in the communication device 100 according to the embodiment. In Figure 3, the horizontal axis indicates time. In (a) and (b) of Figure 3, the known signal has a fixed length from time T5 to time T7.

Figure 3 (a) shows an example in which the convergence of the adaptive equalizer 102 is slow, and the known signal from time T5 to time T6 is used to train the adaptive equalizer 102. Then, after the adaptive equalizer 102 converges and training ends at time T6, the SINR estimation unit 105 starts SINR estimation, and the known signal from time T6 to time T7 is used for SINR estimation. For example, in Figure 1, the training end determination unit 104 determines the end of training at time T6, and outputs an instruction to start SINR estimation to the SINR estimation unit 105. Then, the SINR estimation unit 105 performs SINR estimation using the known signal from time T6 onwards.

Figure 3 (b) shows an example in which the adaptive equalizer 102 converges quickly, and known signals from time T5 to time T6a are used to train the adaptive equalizer 102. Then, after the adaptive equalizer 102 converges and training ends at time T6a, the SINR estimator 105 starts SINR estimation, and known signals from time T6a to time T7 are used for SINR estimation. For example, in Figure 1, the training end determiner 104 determines the end of training at time T6a, and outputs an instruction to start SINR estimation to the SINR estimator 105. Then, the SINR estimator 105 performs SINR estimation using known signals from time T6a onwards.

In this way, the communication device 100 according to the embodiment uses one known signal that combines the training signal and the SINR estimation signal, and performs SINR estimation using the remaining known signal after training is completed. This eliminates the wasted period described in FIG. 2, and allows frequency resources to be used effectively. In particular, the communication device 100 according to the embodiment uses the first half of the known signal as a training signal for the adaptive equalizer 102, and performs training until it is determined that convergence has occurred, so that the training process essential to the communication device 100 can be performed sufficiently.

Here, since training of the adaptive equalizer 102 is a necessary process for the communication device 100 to operate normally, the training signal among the known signals may be set to the longest expected length as shown in FIG. 2(a), and the remaining signal may be used as the SINR estimation signal. Note that SINR estimation is possible for known signals, and SINR estimation can be performed in the latter half of the training signal. However, in order to perform SINR estimation with high accuracy, it is preferable that the signal points of the multi-level signal are widely distributed, for example, and therefore it is desirable to concatenate an SINR estimation signal suitable for SINR estimation to the latter half of the known signal.

Figure 4 shows an example of the relationship between convergence speed and signal length. Figure 4(a) shows an example of a training signal when convergence is fast, and convergence occurs with a training signal of length a. Figure 4(b) shows an example of a training signal when convergence is slow, and a training signal of length a+b, which is b longer than when convergence is fast, is required.

Similarly, FIG. 4(c) shows an example of an SINR estimation signal when convergence is fast, and converges to an SINR estimation signal of length c. FIG. 4(d) shows an example of a training signal when convergence is slow, and an SINR estimation signal of length c+d, which is d longer than when convergence is fast, is required.

Figure 5 shows an example of the required signal length and occurrence probability for the convergence speed in training and SINR estimation. Note that the convergence speed and required signal length are the same as those described in Figure 4.

Figure 5 (a) shows an example of training the adaptive equalizer 102, where if the convergence speed is fast, a signal of length a is required, and if the convergence speed is slow, a signal of length a+b is required. Here, since a typical communication system is operated assuming an environment where communication is possible, the occurrence probability is high when the training convergence speed is fast, and the occurrence probability is lower when the convergence speed is slow compared to when the convergence speed is fast.

Figure 5(b) shows an example of SINR estimation by the SINR estimator 105, where a signal of length c is required when the convergence speed is fast, and a signal of length c+d is required when the convergence speed is slow. For the same reasons as in Figure 5(a), the occurrence probability is high when the SINR estimation convergence speed is fast, and the occurrence probability is lower when the convergence speed is slow than when the convergence speed is fast.

Figure 6 shows an example of a combination of the convergence speed of training and the convergence speed of SINR estimation described in Figure 5. For example, in Figure 6, the total length of the training signal and the SINR estimation signal required when the convergence speed of training is fast and the convergence speed of SINR estimation is also fast is a + c. Furthermore, the total length of the training signal and the SINR estimation signal required when the convergence speed of training is fast and the convergence speed of SINR estimation is slow is a + b + c. Similarly, the total length of the training signal and the SINR estimation signal required when the convergence speed of training is slow and the convergence speed of SINR estimation is fast is a + c + d. Furthermore, the total length of the training signal and the SINR estimation signal required when the convergence speed of training is slow and the convergence speed of SINR estimation is also slow is a + b + c + d.

In Figure 6, the probability of occurrence is highest when the convergence speeds of both training and SINR estimation are fast, and the probability of occurrence is lowest when the convergence speeds of both training and SINR estimation are slow, which occurs rarely.

In this way, even though the occurrence probability of using the longest expected training signal and the longest SINR estimation signal separately is a rare condition, the longest signal must be transmitted, making it impossible to use frequency resources effectively. Therefore, the communication device 100 according to the embodiment uses a known signal with a length that takes the occurrence probability into account.

In the examples of Figures 5 and 6, the required signal length is divided into two cases, one where the convergence speed is fast and the other where it is slow, but the "required signal length" is not necessarily divided into two. Depending on the situation, there may be cases where there are two or more signal lengths. For simplicity's sake, the embodiment will be explained by dividing it into two.

Figure 7 shows an example of a known signal in the communication device 100 according to the embodiment.

Figure 7 (a) shows an example of a signal that combines a training signal and an SINR estimation signal when the occurrence probability is rare, as described in Figure 6. In this case, the total length of the longest training signal and the longest SINR estimation signal, assuming a case where convergence is slow, is required to be a + b + c + d.

(b) of FIG. 7 shows an example of a known signal of the communication device 100 according to the embodiment. The known signal in the embodiment is shortened to a length that is e times the difference in length (b+d) between the cases where convergence is slow and where convergence is fast. Here, e is 0<e<1.

In conventional methods, as shown in FIG. 7(a), the signals used for SINR estimation and training are set separately. Therefore, even if the occurrence probability is low, it is necessary to secure a signal of the corresponding length for each. In contrast, in the method of the communication device 100 according to the embodiment, a single known signal is used as a common signal for both SINR estimation and training, so even if the convergence of one is slow, it is possible to converge both values as long as the convergence of the other is fast.

In addition, the probability of both convergences being slow is low, so the probability of both occurring is rare. Therefore, by not ensuring the longest signal length (a+b+c+d) but instead ensuring a signal length shorter than that (a+c+(b+d)×e), it is possible to converge both the training and SINR estimation values in most cases.

In this way, the communication device 100 according to the embodiment uses a known signal that combines a training signal and an SINR estimation signal, and makes the length of the known signal shorter than the combined length of the longest training signal and the longest SINR estimation signal. This reduces wasted unused periods and enables effective use of frequency resources.

Figure 8 shows another example of the configuration of the communication device 100 according to the embodiment described in Figure 1. In Figure 8, the communication device 100a has a receiving unit 101, an adaptive equalizer 102, a decoding unit 103, a training end determination unit 104, an SINR estimation unit 105, and a data separation unit 106. In Figure 8, the receiving unit 101, the adaptive equalizer 102, the decoding unit 103, the training end determination unit 104, and the SINR estimation unit 105, which are the same as those in Figure 1, operate in the same manner as in Figure 1, so duplicated explanations will be omitted. The difference from Figure 1 is that the data separation unit 106 is arranged between the equalization filter 201 of the adaptive equalizer 102 and the decoding unit 103. Then, the equalized known signal is output from the data separation unit 106 to the SINR estimation unit 105, and the determination result of the training end determination unit 104 is output to the data separation unit 106.

The data separator 106 normally outputs the signal output from the equalization filter 201 of the adaptive equalizer 102 to the decoder 103, but if the judgment result output from the training end judgment unit 104 indicates the end of training, it outputs the signal output from the equalization filter 201 to the SINR estimation unit 105.

For example, in the case of FIG. 3(a) described above, the training end determination unit 104 outputs a determination result indicating the end of training at time T6 to the data separation unit 106. Then, the data separation unit 106 outputs the known signal from time T6 onwards to the SINR estimation unit 105, which then performs SINR estimation. Similarly, in the case of FIG. 3(b), the training end determination unit 104 outputs a determination result indicating the end of training at time T6a to the data separation unit 106. Then, the data separation unit 106 outputs the known signal from time T6a onwards to the SINR estimation unit 105, which then performs SINR estimation.

In this way, the communication device 100a according to the embodiment of FIG. 8 uses one known signal obtained by concatenating the training signal of the adaptive equalizer 102 and the communication quality estimation signal of the SINR estimation unit 105, similar to the communication device 100 of the previous embodiment. Then, in the communication device 100a, the data separation unit 106 outputs the remaining known signal after the end of training to the SINR estimation unit 105 at the timing when the training end determination unit 104 determines the end of training. In other words, regardless of the speed of convergence of the training, SINR estimation is performed consecutively to the training at the timing when the training ends. As a result, the length of the known signal to be transmitted is reduced compared to the case where training and SINR estimation are performed independently and separately, and the frequency resources can be used effectively. In particular, as described in the previous embodiment, the length of the known signal is set shorter than the total length of the longest training signal and the longest SINR estimation signal based on the occurrence probability. As a result, the unused and wasted period of the known signal is reduced, and the frequency resources can be used effectively.

As described above, the communication device, communication method, and communication program for performing propagation path equalization and quality estimation according to the present invention can reduce the total number of bits of the known signal to be transmitted and make effective use of frequency resources by using a single known signal that combines the training signal of the adaptive equalizer and the SINR estimation signal.

Here, a program corresponding to the processing executed by each block of the communication device 100 or communication device 100a described in FIG. 1 or FIG. 8 may be executed by a general-purpose computer or an integrated circuit such as an FPGA (Field Programmable Gate Array). The program may be provided by being recorded on a storage medium, or may be provided via a network.

100, 100a: communication device; 101: receiver; 102: adaptive equalizer; 103: decoder; 104: training end determination unit; 105: SINR estimation unit; 106: data partitioning unit; 201: equalization filter; 202: parameter calculation unit

Claims (7)

an adaptive equalizer that uses parameters calculated by estimating propagation path characteristics to equalize distortion of the propagation path characteristics contained in a received signal;
a parameter calculation unit that estimates the channel characteristics based on a known signal of a predetermined length received from a transmitter and calculates the parameters of the adaptive equalizer;
a training end determination unit that determines end of training of the adaptive equalizer based on a convergence state of the parameters calculated by the parameter calculation unit;
and a communication quality estimation unit that , when the training end determination unit determines that the training has been ended, estimates communication quality of a propagation path by using a remaining portion of the known signal other than a portion used for training . 2. The communication device according to claim 1,
the known signal is a signal obtained by coupling a training signal for calculating the parameters of the adaptive equalizer and a communication quality estimation signal for the communication quality estimation unit to estimate communication quality of a propagation path. 3. The communication device according to claim 2,
The communication device, wherein the known signal is shorter than a total length of the training signal when convergence is slowest and a length of the communication quality estimation signal when convergence is slowest. an adaptive equalization process for equalizing distortion of the propagation path characteristics included in the received signal using parameters calculated by estimating the propagation path characteristics;
a parameter calculation process for estimating the channel characteristics based on a known signal of a predetermined length received from a transmitter and calculating the parameters used in the adaptive equalization process;
a training end determination process for determining the end of training of the adaptive equalization process based on a convergence state of the parameters calculated in the parameter calculation process;
and when it is determined that the training has been terminated in the training termination determination process , a communication quality estimation process is executed to estimate a communication quality of a propagation path by using a remaining portion of the known signal other than the portion used for training . 5. The communication method according to claim 4,
the known signal is a signal obtained by concatenating a training signal for calculating the parameters used in the adaptive equalization process and a communication quality estimation signal for estimating communication quality of a propagation path in the communication quality estimation process. 6. The communication method according to claim 5,
A communication method, characterized in that the known signal is shorter than a total length of the training signal when convergence is slowest and the communication quality estimation signal when convergence is slowest. A communication program that causes a computer to execute the process performed by the communication method according to any one of claims 4 to 6.

Images