JP2009267930A - Information transmission system having non-uniformity error property - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information transmission system having higher reliability and performance as a whole, than in the case where error protection is performed uniformly, by not applying uniform coding to all information items but applying error protection in accordance with importance of information. <P>SOLUTION: A non-uniform error protection system includes: a coding modulator which integrally processes coding and modulation, through trellis coding modulation, for transmitting a signal to a communication path; an adaptive equalizer which reduces phasing waveform distortion of a signal received via the communication path; and a demodulating decoder which integrally processes demodulation and decoding, while using Viterbi decoding, on a signal output from the adaptive equalizer. It is preferable for the non-uniform error protection system to connect in parallel a plurality of coding modulators having different levels of error protection, to switch the coding modulators in accordance with importance of information, to connect in parallel a plurality of demodulating decoders having different levels of error protection, and to switch the demodulating decoders in accordance with an importance of information. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

The present invention relates to an information communication system and an information transmission method, and more particularly to a non-uniform error protection method applicable also to fading communication paths such as wireless communication.

  The main purpose of communication is to send information from the information source to the receiving side how efficiently and accurately. From the viewpoint of transmission, information bits from information sources tend to be handled uniformly, but the information bits may have different properties depending on the information source. In this case, it may be desirable to perform transmission taking into account the nature of the information. Specifically, sufficient error correction is performed for information with high importance, and simple error correction is performed for information with low importance, compared to the case where uniform coding is applied to any information. It is efficient as a whole, and high reliability and performance can be expected as a whole system. For example, error correction is performed even in conversations between people, and even if there are words that cannot be heard due to noise, the content of the conversation can usually be understood from the context of the conversation. In other words, since errors in information with high importance are reduced, even if information with low importance is lost to some extent, communication between humans is not hindered. In many communication applications, the bits to be transmitted require different degrees of importance. An example is an ATM cell as an information transfer unit in ATM (asynchronous transfer mode). Of the ATM cells having a fixed length of 53 bytes, 5 bytes are allocated to the ATM header. Because the ATM header is a control tower of ATM cells, it is desirable to protect it stronger than the other 48 bytes. The present invention is a proposal of an information communication system apparatus having such non-uniform error protection performance.

In order to create non-uniform error protection, it is required to provide non-uniform error rate characteristics for information bits having different importance. A multilevel code is used as the method. A multilevel encoder is shown in Non-Patent Document 1, for example, and its general configuration is shown in FIG. The information source encoder divides the information sequence into M parallel bit sequences in which the priorities are arranged in descending order. The channel encoder creates non-uniform error characteristics by comprising _M different error correction codes C ₁ ,..., C _{M in} which the inter-code distances are arranged in descending order. The problem with this multilevel encoder is that each encoding is simple, but the overall decoding is complicated. Therefore, enormous processing is required for multilevel decoding. In addition, since the multilevel encoder requires an extra code bit in order to achieve the encoding gain, the efficiency of the bandwidth is not achieved. On the other hand, for the purpose of improving the bandwidth efficiency, there is a system using uniform signal point arrangement, non-uniform signal point arrangement, and time division multiplexing of coded modulation. This method is shown in Non-Patent Document 2, for example, and FIG. 8 shows an example of producing non-uniform error characteristics using coded modulation.

A stronger code (encoder 1) is used for the higher importance group, and a lower code (encoder 2) is used for the less important group. In the configuration of FIG. 8, since the output from the encoder 1 and encoder 2 on the information source side is used for mapping to the signal point arrangement, equation (1) must be satisfied.

Here, m ₁ and m ₂ represent highly important information bits and less important information bits, respectively, and r ₁ and r ₂ represent redundant bits added by the encoder 1 and the encoder 2, respectively.

  On the receiver side, a decoder corresponding to each encoder is prepared and decoded. In these conventional methods, the importance is assumed to be two levels, and information with high importance and information with low importance are often switched for each bit or switched with a regular data string.

H. Imai and S. Hirakawa, "A newmultilevel coding method using error correcting codes" ", IEEE Trans. Inform. Theory, pp. 371-377, May. 1977. L. -F. Wei, "Coded modulation with unequal error protection", IEEE Trans. Commun, vol. 41, pp. 1439-1449, Oct. 1993.

  Incidentally, in recent years, there has been an increasing demand for wireless communication using digital technology such as mobile phones. However, many conventional non-uniform error protection systems have a system configuration that is premised on use in an AWGN (white noise) communication channel. On the other hand, there have been almost no evaluations on fading channels, and application and introduction of non-uniform error protection systems in fading channels have been desired.

  An object of the present invention is to provide a non-uniform error protection system that can be applied to fading communication paths such as wireless communication and has high reliability and performance as a whole system.

  The non-uniform error protection system according to claim 1, which is made to achieve the above object, integrally processes a code and a modulation by trellis coded modulation, which transmits a signal to a communication path. Code modulator, adaptive equalizer that reduces fading waveform distortion of the signal received via the communication channel, and demodulation that integrally processes demodulation and decoding of the signal output from the adaptive equalizer by viterbi decoding And a decoder.

  The non-uniform error protection method according to claim 2, wherein the input signal is trellis-coded and modulated, and the code and the modulation are integrated and transmitted to the communication path, and the fading distortion of the signal received through the communication path is adaptively equalized. After that, demodulation and decoding are integrally performed by Viterbi decoding.

  A non-uniform error protection system according to claim 3 is the non-uniform error protection system according to claim 1, comprising a plurality of the code modulators having different error protection levels connected in parallel, and the plurality of code modulators. A transmission switching unit is provided to switch the input / output to a code modulator with a high level of error protection for high importance information and to a code modulator with a low level of error protection for low importance information. A plurality of demodulation decoders having different levels are connected in parallel, and the inputs and outputs of the plurality of demodulation decoders are switched to a demodulation decoder having a high level of error protection for highly important information. The low information includes a reception switching unit for switching to a demodulator / decoder having a low level of error protection.

  A non-uniform error protection system according to claim 4 is the non-uniform error protection system according to claim 3, comprising two code modulators having different levels of the error protection, and a double QPSK having different energy is arranged. Using the signal point arrangement, highly important information is arranged in a high energy QPSK arrangement.

  A non-uniform error protection system according to claim 5 is the non-uniform error protection system according to claim 3, comprising three code modulators having different levels of error protection, and three QPSKs having different energy levels are arranged. Using the signal point arrangement, information with higher importance is arranged in order from the QPSK arrangement side with higher energy.

  A non-uniform error protection system according to claim 6 is the non-uniform error protection system according to claim 3, comprising a plurality of the code modulators having different levels of error protection, and each set of four signal points on the same RING. In this case, information with higher importance is arranged in order from a set of signal points having a larger amount of phase change, using signal point arrangements arranged with different amounts of phase change.

  The non-uniform error protection system according to claim 7 is the non-uniform error protection system according to claim 6, wherein the code modulator is designed so that the Rice parameter K = 5 or more or the speed of the moving body is 3 km / h or less. It is characterized by.

  The non-uniform error protection system according to claim 8 is the non-uniform error protection system according to claim 3, wherein the demodulator / decoder determines the importance of the information for every N bits of transmission data.

  The main effects of the present invention are listed.

  In the conventional method, paying attention to the importance of each bit, different error correction is performed for each bit using a block code. They usually perform error correction in two stages, high and low, and the bits with high importance and the bits with low importance are regularly switched, or they are often premised on parallel data strings. The proposed method, on the other hand, considers information sequence N bits that contain various importance parts in one information sequence, and encodes them using encoders with different error correction capabilities according to their importance. To correct the error according to the importance.

  With regard to importance, the conventional method presupposes use at two levels (high importance, low importance), but in the present invention, implementation at three levels (high importance, medium importance, low importance) is implemented. It is possible.

  In the conventional system shown in FIG. 8, modulation and coding are performed separately, but in the proposed system, modulation and coding are integrated by trellis coded modulation to simplify the system. By using trellis coded modulation, the number of transmission bits is increased by coding. However, the modulation multi-level number is increased correspondingly, and the reliability in demodulation is increased while keeping the bandwidth unchanged.

  Since each encoded signal is modulated, there is no restriction on the signal point arrangement as shown in Equation (1) in FIG.

  Since it is equipped with an adaptive equalizer as a countermeasure against fading, it can also be applied to fading channels such as wireless communications, and its usage is much wider than the conventional system based on AWGN (white noise).

  Thus, fading communication paths such as wireless communication can be applied, and information transmission can be performed with high reliability and performance as the entire system.

  The invented system configuration is shown in (Fig. 1). In the configuration for transmitting the information before the communication path, the encoder and the modulator are integrated and M code modulators that process the code and the modulation integrally are arranged in parallel (in the case of M = 3 in FIG. 1). It is the structure which was made. The M code modulators have different levels of error protection. For the input and output of these M code modulators, switch to a code modulator with a high level of error protection for information of high importance, and switch to a code modulator with a low level of error protection for information with low importance. A transmission switching unit (upper selector switch in FIG. 1) is connected. The transmission switching unit functions as a switch.

  After the communication path, an adaptive equalizer that reduces fading waveform distortion of the signal received via the communication path is arranged, and the signal output from the adaptive equalizer is demodulated by Viterbi decoding and the demodulation is integrally processed. A decoder is connected.

  The demodulator / decoder is connected in parallel with M demodulator / decoders with different levels of error protection. A reception switching unit (lower switch in FIG. 1) for switching to a decoder and switching to a demodulator with a low level of error protection is connected to information of low importance. The reception switching unit functions as a changeover switch.

  When information having two importance levels is used, two code modulators are arranged in parallel and switched. The input signal is a random bit string of 0 and 1, and the importance of the information of the input signal changes every N bits. This study does not consider what information is important or low. That is, when evaluating the communication system, a certain bit string N is interpreted as having high importance. In addition, when the information on the respective importance levels is biased, the information on the importance levels is generated with the same probability because the information approaches the error-corrected information. In the encoder, encoding is performed using a convolutional code. The coding rate is set lower as the importance of information is higher, so that errors are less likely to occur. In the system configuration proposed in the invention, modulation and coding (or demodulation and decoding) are integrally performed by trellis coding modulation to simplify the system. By using trellis coded modulation, the number of transmission bits is increased by coding. However, the modulation multi-level number is increased correspondingly, and the reliability in demodulation is increased while keeping the bandwidth unchanged. In addition, since each encoded signal is modulated, there is no restriction on signal point arrangement as shown in Equation (1).

Although there are no particular restrictions on the communication path in the proposed invention method, a fading communication path is considered assuming mobile communication. In a multipath environment such as wireless communication, intersymbol interference occurs due to fading. Intersymbol interference is a phenomenon that occurs due to the attenuation of high-frequency components in the communication path, and is distorted in amplitude and phase, causing a loss in communication efficiency and reliability. In order to reduce intersymbol interference, an adaptive equalizer is inserted on the receiving side. Thereby, the waveform distortion received by fading is corrected. In the proposed invention system, the training mode (FIG. 9) and the tracking mode (FIG. 10) are switched by a switch. The desired signal in the adaptive equalizer is a transmission signal to be decoded based on the principle of the equalizer. Therefore, if a transmission signal can be used on the receiving side, an adaptive equalizer can be basically realized. Therefore, a method of generating the same signal as a transmission signal called a training signal on the normal receiving side is adopted. The training signal can be easily generated if the transmission side and the reception side can be synchronized. In this way, the state where the training signal is generated on the receiving side is a state called a training mode in the adaptive equalizer. In this training mode, the coefficient of the adaptive equalizer is adjusted. However, substantial information transmission is not performed in this training mode. In order to perform substantial information transmission, after a certain period of time, the training mode is switched to a state called a tracking mode (also referred to as a determination direct mode). In the tracking mode, the training signal is not used and the sign determination value of the output of the adaptive equalizer is set as a desired signal. The reason why such a desired signal is used in the tracking mode is that if the coefficient adjustment is sufficiently performed in the training mode, the sign determination value of the output of the adaptive equalizer is almost equal to the transmission signal. In fact, it is known that the adaptive equalizer works stably in the tracking mode when the channel characteristics are time-invariant and the convergence characteristics in the training mode are good. It is also known that even if the channel characteristics are time-varying, the characteristics are not so severe, and if the error probability of the determiner is ½, it works effectively.

  On the receiving side, signals transmitted from the same communication path are received and processed in parallel by each decoder and demodulator. Also, encoder estimation processing is performed in parallel with this flow. An algorithm for the encoder estimation process is shown in FIG.

In the encoder estimation processing algorithm shown in FIG. 6, first, in step 1, distances d _ij (j: 0 to 3) from the i-th transmission signal (or received signal) to each signal point number (0 to 8). ).
In step 2, the minimum distance (minimum value) of the distance _dij between the transmission signal (or reception signal) measured in step 1 and the importance signal (0 to 7) is calculated. In step 3, the number of the importance signal forming the minimum distance with the transmission signal is determined. In Step 3, Steps 1 and 2 are performed for each N bits that are data switching speeds, and these are added to determine the importance signal number _Hmin . Similarly, in step 4, the number of the importance signal forming the minimum distance with the received signal is determined. In step 4, even for signal points with low importance, step 1 and step 2 are performed for each N bits which is the data switching speed, and these are added to determine the importance signal number L _min . In step 5, H _min and L _min obtained in step 3 are compared, and if H _min <L _min, it is determined as a high importance level, and if H _min > L _min, it is determined as a low importance level. . In this case, if the number H _min of the importance signal forming the minimum distance from the transmission signal obtained in step 3 is 4 or less, it is estimated as the transmission signal for the inner RING (low importance), and if it is 5 or more, the outer RING (important) Estimated to be a transmitted signal for (degree high).

  As a result, the most appropriate encoder output is adopted for every N bits.

  In the present invention, FIGS. 2 to 4 are devised as signal point arrangements in order to give importance to information.

  (Fig. 2) shows the signal point arrangement (2RING) in which double QPSK with different energies are arranged. Therefore, there are two levels of importance, high or low. A signal having a low importance is associated with the inside, and a signal having a high importance is associated with the outside. In addition, the phase of each QPSK is shifted by π / 4 in order to increase the distance between the codes.

  (Fig. 3) is a 2RING type with one more importance added (3RING). That is, the signal point arrangement can be considered as triple QPSK with different energy. Therefore, there are three levels of importance: high, medium, and low. As in the 2RING type, the phase of each QPSK is shifted by π / 4.

(FIG. 4) shows a case (TRAP) in which four signal points having different phase variations are arranged on the same RING. The relationship d _L : d _H = 1: 2 is satisfied. Since there are two types of phase differences, there are two levels of importance, high or low.

  The importance of the input signal changes every N bits, and encoding and modulation are performed according to the importance of three levels. In order to perform this coding and modulation as one body, a trellis coded modulation system is used. In a multipath environment of wireless communication, intersymbol interference occurs due to fading. Intersymbol interference is a phenomenon that occurs due to attenuation of high-frequency components in the communication path, and is distorted in amplitude and phase, thereby causing a deterioration in communication efficiency and reliability. In order to reduce intersymbol interference, an adaptive equalizer is provided on the receiving side. Thereby, the distortion received by fading is corrected. In the invented system, the training mode and the tracking mode are switched by a switch. The output signal from the adaptive equalizer is estimated by which encoder is encoded, and is demodulated and decoded to become an output signal.

  The effectiveness of the computer simulation will be described below as an embodiment of the present invention. The assumed environment is shown in (Fig. 13). This shows a Rician fading channel model in which a single-frequency carrier wave is transmitted from a transmitting station, and a direct wave and one elementary wave arrive at a moving receiving station (mobile body). The power ratio between the direct wave power A2 / 2 and the average wave power σs2 is called the Rice parameter, and is defined by the following equation.

  A simulation model with 2 levels of importance is shown in Fig. 12.

  As shown in FIG. 12, the simulation is performed with a configuration in which a transmission sequence that transmits a signal to a communication channel, a communication channel sequence that artificially adds noise and fading in communication, and a reception sequence are connected in this order. did. The transmission sequence is code-modulated at two levels. The importance level determination switch switches at which level the code modulation is performed. When the importance is high, code modulation is performed by a convolutional encoder, a matrix interleaver, and a high-level modulator. When the importance is low, code modulation is performed only with a low-level modulator.

  The transmission sequence will be described in detail. The input data series in FIG. 12 is binary data of {0, 1}. In importance determination, there are multiple importance levels by using a switch (importance level determination switch) that switches to each importance level with the same probability every N bits without adding a bit string representing the importance level in the input data series. Assume that Information with high importance is encoded using a convolutional code. Further, when a fading channel is assumed, a local error occurs, so that the effect of the error correction code may not be sufficiently expected. Therefore, the bit error rate is improved by using a matrix interleaver. If the matrix used in the matrix interleaver is larger than the encoder switching speed N, the delay on the receiving side will be very large, so the minimum size that represents the data string that represents one importance is the matrix size. . That is, the encoder switching speed N becomes the matrix size. It is assumed that the information with the lowest importance is not encoded. This is because information with high importance has as few errors as possible, and information with low importance is considered to have no problem even if some errors occur. The transmission rate is improved by not encoding information of low importance. Since the transmission signal sent to the encoder and the modulator used for each importance uses the same communication path, the importance determination switch is also used for a signal that has been subjected to high-level and low-level modulation.

  Next, the communication path sequence will be described. The channel series assumes an AWGN channel and a fading channel. For the noise added in the AWGN channel, consider the noise ratio to the average power of signal points used for two or three importance levels, and for Gaussian noise, consider the SNR per information bit. That is, if information bits included in a bit string encoded at a coding rate of 1/2 are 1 / T, information bits included in a bit string that is not encoded are 2 / T. By making the SNR per information bit equal, both are evaluated fairly. In addition, Rayleigh fading is considered for the fading channel. In addition, the power of noise applied on the communication path is different depending on the RING in the RING type, and therefore the problem is how much noise power is applied. In this study, the average value of signal power at the transmission point is used as noise power.

  A reception sequence will be described. The received sequence was composed of an adaptive equalizer and a demodulation / decoding unit. The demodulation / decoding unit performs demodulation and decoding at two levels. The importance level estimation switch switches at which level the code modulation is performed. If the degree of importance is high, the signal is demodulated and decoded by a high-level demodulator, matrix deinterleaver, and Viterbi decoder. When the importance level is low, only the low level demodulator demodulates and decodes.

Viterbi decoding is used for decoding. In the decoder, a hard decision is made and the traceback is 5 times the constraint length. Here, in the Viterbi decoding method, a method called traceback is used to determine the final decoding result. Survival path memory (survival path memory)
Memory). Traceback is an operation that reviews the contents of this memory at a certain time t. Although this time is related to error correction ability, it is empirically known that it is sufficient to take 5 times the constraint length. In other words, it is considered that the signal up to the time five times before the constraint length affects the current signal, and the survivor path is searched for error correction.

  Under the above conditions, error rate characteristics at each encoder, encoder estimated error rate characteristics with respect to SNR, and encoder estimated error rate characteristics with respect to encoder switching speed are obtained.

  For highly important information, encoding is performed in order to protect it from errors. On the other hand, it is assumed that information with low importance does not need to be encoded because it does not need to be protected from errors so much. On the transmission side, encoding is performed using a convolutional code, and a bit error rate is improved using a matrix interval. The training mode and tracking mode in the adaptive equalizer are switched using a digital timer. On the receiving side, decoding is performed by viterbi decoding and a matrix deinterleaver is used. For the noise added in the AWGN channel, consider the noise ratio to the average power of signal points used for 2 to 3 importance levels, and Gaussian noise considers the SNR per information bit. That is, if information bits included in a bit string encoded at a coding rate of 1/2 are 1 / T, information bits included in a bit string that is not encoded are 2 / T. By making the SNR per information bit equal, both are evaluated fairly. In addition, the noise power in the communication path is different depending on the RING in the RING type, so the average value of the signal power at the transmission point is used. Under the above conditions, error generation in each encoder and error generation by encoder determination are assumed to be independent, and each characteristic is acquired and evaluated. The simulation specifications are shown in Fig. 13. For the proposed 3 signal point arrangements (2RING, 3RING, TRAP), the three speeds of 0.1 km / h, 10 km / h, and 50 km / h are considered as the speed of the receiving station (mobile unit).

  First, characteristic evaluation in each encoder will be described. The bit error rate in each encoder in 2RING-type, 3RING-type, and TRAP-type is shown in FIGS. In the figure, ◯ indicates high importance, □ indicates medium importance, Δ indicates low importance, and a simple line without a symbol indicates a case where uniform encoding is performed by a high importance encoder.

  The bit error evaluation in RING-type is described. In the case of 2RING-type (Fig. 14) and 3RING-type (Fig. 15), regardless of the speed of the moving object, information with lower importance shows worse characteristics than when uniform, but it is more important. The information shows better characteristics than when any information is uniformly encoded. As a result, by using the non-uniform error protection code, the efficiency as a whole is good and high reliability and performance can be expected as the whole system.

The bit error evaluation in TRAP-type is described. In the case of the TRAP type (Fig. 16), when the speed is 0.1 km / h, a non-uniform error protection code can be realized as with the RING type, but the speed is 10, 50 km / h, and an error floor has occurred. . The cause seems to be related to the fact that the distance d _L between the points of low importance is 1.035. FIG. 20 shows the result of calculating each signal point distance up to the third decimal point from the geometric relationship, assuming that the radius of each circle in FIGS. From FIG. 20, among d _C , d _L and d _H related to the bit error rate of each encoder (d _L in TRAP type) is the minimum value as the distance between code points. The TRAP type is also considered to be affected by the fact that all points are on the same circle and non-uniform signal point arrangement. In general, when the value of the rice parameter K is increased, the direct wave becomes dominant as compared to the Rayleigh wave (elementary wave), so the amplitude and phase distortion imparted to the signal point by the Rayleigh wave is reduced and the characteristics are improved. . Therefore, when the value of K was raised and evaluated, it was found that the error floor can be avoided if the TRAP type is designed with K = 5 or more. The case of K = 5 is shown in FIG. However, since we want to confirm the characteristic change when K is changed, only the data with high and low importance are shown. In addition, if the speed of the receiving station is low, it is less susceptible to fading and the characteristics are improved. However, it was found that an error floor can be avoided by designing at 3 km / h or less.

Next, characteristic evaluation of encoder estimation will be described. Encoder estimation is performed based on the importance estimation algorithm shown in FIG. Here, a comparison was made between the RING type and the TRAP type, which have a maximum speed of 0.1 km / h and a two-level importance. Fig. 18 shows the encoder estimation error rate when the encoder switching speed (significant high bit string N bit) is kept constant and the information bits are increased. Also, the encoder estimation error rate when the SNR per information bit is kept constant and the encoder switching speed is increased is shown in (FIG. 19). The encoder estimation error evaluation for SNR is described. The results of evaluation at the encoder switching speed N = 100 are shown in FIG. The TRAP type shows better characteristics than the RING type. Regarding the RING type, it can be seen that the encoder estimation error is hardly improved even if the SNR per information bit increases. This is presumed that the signal placed in the outer RING enters the inner RING, making encoder estimation difficult. Regarding the TRAP type, the characteristics are improved as the SNR increases. This is presumed that since all the points in the TRAP type are on the same RING, even if the amplitude drops due to fading, the signal points for different importance levels are not so much affected. The encoder estimation error evaluation for the encoder estimation speed is described. The results of evaluation with an SNR per information bit of 10 dB are shown in (Fig. 19). The TRAP type shows better characteristics than the RING type. Regarding the TRAP type, the encoder error is improved when the encoder switching speed N = 100 or more. As for the RING type, good characteristics are exhibited when N is sufficiently large. In the proposed unequal error protection system, the characteristics depend on the signal point distances d _C , d _L and d _H. The signal point distance d _C is related to the encoder estimation, and the signal point distances d _L and d _H are related to the bit error rate of high importance and low. Therefore, they are in a trade-off relationship.

The main conclusions of the proposed invention will be described. A system using trellis coded modulation and an adaptive equalizer was proposed for the purpose of realizing non-uniform error protection codes in fading channels, assuming use in mobile communication environments. Computer simulations using RING and TRAP signal point arrangements confirmed the effectiveness of the system. The main conclusions are as follows.
(1) It was shown that the proposed non-uniform error protection system can receive error protection according to importance in fading channels. It has been shown that the higher importance part receives higher error protection than the case of uniform error protection.
(2) In the TRAP type, it was found that the error floor can be avoided and the intended non-uniform error protection system can be realized if the rice parameter K = 5 or higher or the speed of the moving body is designed to be 3 km / h or lower.
(3) The arrangement like the RING type is effective in reducing errors in each encoder, and the arrangement like the TRAP type is effective in reducing errors in encoder estimation. all right. This is a trade-off in signal point distance, and further study is necessary in the future.

It is a block diagram which shows the nonuniform error protection system of this invention. It is a 2RING type signal point arrangement diagram to which the present invention is applied. It is a 3RING type signal point arrangement diagram to which the present invention is applied. It is a Trap type signal point arrangement diagram to which the present invention is applied. 2RING type constellation to which the present invention is applied (used to explain encoder estimation) 4 is an encoder estimation algorithm to which the unequal error protection method of the present invention is applied. It is a block diagram which shows a prior art (example using a multilevel encoder). It is a block diagram which shows the prior art (example using time division encoding modulation). It is a block diagram which shows the training mode of the adaptive equalizer used for this invention. It is a block diagram which shows the tracking mode of the adaptive equalizer used for this invention. It is an assumption environment (Rice fading model) figure. It is a block diagram which shows a simulation model. It is a table | surface which shows a simulation specification. It is a simulation result graph of 2RING type bit error rate (K = 0). It is a simulation result graph of 3RING type bit error rate (K = 0). It is a simulation result graph of TRAP type bit error rate (K = 0). It is a simulation result graph of TRAP type bit error rate (K = 5). It is a simulation result graph of the encoder estimation error rate with respect to SNR. It is a simulation result graph of an encoder estimation error rate with respect to an encoder switching speed. It is a table | surface which shows the result of having calculated the signal point distance.

Claims (8)

  A signal modulator for transmitting a signal to a communication path, a code modulator for integrally processing codes and modulation by trellis-coded modulation, an adaptive equalizer for reducing fading waveform distortion of a signal received via the communication path, and A non-uniform error protection system comprising: a demodulation decoder that integrally processes demodulation and decoding of a signal output from an adaptive equalizer by Viterbi decoding.   The input signal is trellis-coded and modulated, and the code and modulation are combined and transmitted to the communication path. Fading distortion of the signal received via the communication path is reduced by adaptive equalization, and then demodulated and decoded by Viterbi decoding. A non-uniform error protection method characterized in that it is performed as a single unit.   A plurality of code modulators having different levels of error protection are connected in parallel, and the input / output of the plurality of code modulators is switched to a code modulator having a high level of error protection for highly important information, A transmission switching unit that switches to a code modulator having a low level of error protection for information of low importance, and a plurality of the demodulation decoders having different levels of error protection are connected in parallel, and the plurality of demodulation decoders A reception switching unit that switches to a demodulator / decoder with a high level of error protection for information of high importance and switches to a demodulator / decoder with a low level of error protection for information of low importance The non-uniform error protection system according to claim 1.   Arranging high-importance information in a high-energy QPSK arrangement using a signal point arrangement in which two code modulators with different levels of error protection are provided and double QPSK with different energies are arranged The non-uniform error protection system according to claim 3.   Uses three code modulators with different levels of error protection, and uses signal point arrangements in which triple QPSKs with different energies are arranged to support information with higher importance in order from the higher QPSK arrangement side The non-uniform error protection system according to claim 3, wherein the non-uniform error protection system is arranged.   A plurality of the code modulators having different error protection levels, and using a signal point arrangement in which a phase change amount is different for each set of four signal points on the same RING, the phase change amount 4. The non-uniform error protection system according to claim 3, wherein information having a high importance is arranged in order from a set of signal points having a large.   The non-uniform error protection system according to claim 6, wherein the code modulator is designed with a Rice parameter K = 5 or more or a moving body speed of 3 km / h or less.   4. The non-uniform error protection system according to claim 3, wherein the demodulator / decoder determines the importance of information for each N bits of transmission data.