Noise-balanced qam detection

The present invention relates to the principles of phase modulation coding and decoding that can be used in various types of communication systems. In particular, it relates to a technology applicable to a wireless communication system.

When transmitting data over a wireless channel, a commonly used principle for overcoming the limits of the signal speed at which a binary sequence signal is sent is to utilize four or more unique symbols. This allows it to exceed the maximum signal speed (bits per second), which corresponds to a bit rate that is twice the passband (Hz) given by the Nyquist theorem.

4-Phase Phase Modulation (QPSK), also known as Quadrature Quadrature Amplitude Modulation (4QAM), encodes a 2-bit word into four discrete symbols. These symbols can be represented as signal vectors on the complex plane with constant amplitude and four different phase values with respect to the reference signal. Detection is performed by confirming which quadrant on the complex plane the received signal belongs to.

The bit rate can be further increased by using higher order modulation. However, on the complex plane, the individual symbols get closer and more difficult to distinguish, which imposes higher demands on the detection stage. Also, the number of symbols that can be used is limited due to signal degradation when transmitted over a given medium.

Higher-order modulations are commonly referred to as M'ary QAMs. Where M = 2^NIs the number of discrete symbols available, and N bits can be transmitted per symbol. M-value QAM is also called M-value APK (quadrature phase modulation) because the amplitude and phase change for individual symbols.

Fig. 1 shows a conventional transmitter, and Fig. 2 shows a conventional receiver.

The transmitter includes a data buffer 1, a mapper 2, a baseband filter 3, an intermediate frequency (IF) oscillator 6, a phase divider 5, an adder 7, and an adder 4 to which a radio frequency (RF) signal is transmitted. There is.

The data temporarily stored in the buffer 1 is transmitted to the mapper 2 according to the transmission speed of the data on the wireless interface. The data (which can be regarded as a binary bit serial string) is separated by Mapper 2 into symbols having I and Q components on the complex plane described above.

On the other hand, the receiver decodes the I and Q components by multiplying the input signal (RF) by a signal generated by the signal oscillator IF12 and shifted by 90 degrees by the divider 11. The signal of IF12 is usually coherent by a carrier regeneration PLL (Phase Locked Loop) with a carrier signal from IF6, so that the RF signal is filtered by filters 9 and 10 respectively and then decoded and on the complex plane. Returned to. The error signal 16 corresponding to the deviation between the expected symbol value and the detected symbol value is input to the PLL loopback filter 13 adjusting the IF generator 12.

Figures 3 and 4 show traditional schemes for transmitting data. The frame alignment word F1 consisting of symbols of a predetermined sequence has a function for referencing traffic data B1, B2 ... Bn-1 which are subsequent frames. For example, the frame word may be 8-bit long. After transmitting the frame for a period of time, the frame alignment word is repeated. The demodulator is then able to identify individual frame positions in each frame through the frame aligner 15 where a given sequence is restored.

As shown in FIG. 4'(FIG. 4A), the frame alignment term may include a single pilot signal P identifiable from the remaining traffic carrying symbol T.

The error signal vector E corresponding to the deviation of the detected symbol value D from the expected reference symbol value R is detected by the demapper 14 and input to the PLL loopback filter 13 that shifts the control value E which means the shifted error signal. .. For example, the angle φ between the vectors with respect to positions D and R is calculated and can be used as the error control signal E'.

The latter signal is used to tune the IF generator 12, so that the phase of the signal from IF12 is coherent with that of IF6.

Additive noise, which mainly consists of thermal noise of the received signal, is usually transmitted to the phase detector output. Since the noise portion is independent of a given symbol, the noise portion of the received signal constitutes a certain area around the transmitted constellation symbol.

All QAM schemes greater than 4 have constellations with different envelopes for individual symbols. Therefore, in a QAM scheme greater than 4, when the error signal E is used directly unprocessed, the noise transfer from the symbol with the smaller envelope (G) is much larger than the symbol with the large envelope (H). This relationship is shown in Fig. 5. This figure shows the detection symbols in a 16QAM constellation affected by thermal noise.

Therefore, it becomes necessary to compensate for the noise contribution.

The best thing about noise transfer is to "equalize" the phase detector with respect to the envelope, so multiply the detected phase error with the envelope of the signal as in the following equation: I: E'= φ · | D | where D is the detection signal and φ is the angle between the detection signal and the determined symbol reference R (center of the square). See Figure 6.

However, the above operation requires many program instructions and is not suitable for some applications.

On the other hand, the deviated signal is more easily calculated by the following equation: II: E'= D_Q · R_ID_I · R_Q D is the detection signal and R is the determined symbol.

In the latter case, the noise depends on the shape of the "square" in relation to the envelope of the signal. In FIG. 7, given values of the control signal E'in the above related equation II are shown by various lines.

In order to equalize the errors detected in relation to the noise, the errors should be separated from the actual envelope. Therefore, the following formula is used: III: E'= (D_Q R_ID_I R_Q) / ((D_Q))²+ (D_I)²However, the above two noise balancing methods also require a relatively complicated algorithm and require a large processing capacity in the receiving stage.

In the conventional technical document (Patent Document 1), the phase error value of the received data is derived by subtracting the determined I-channel data and multiplying the difference sign by the difference itself, and further, the weighting function is used as the phase error value. The phase error detection method to be applied is shown. This signal is used for phase correction of received data. In addition, a weighting function is applied to reduce false positives of phase errors caused by decision errors caused by adjacent errors between symbols.<patcit num="1"><text>U.S. Pat. No. 5,967,786</text></patcit>

<p> A first object of the present invention is to clarify a method of providing noise-resistant high bit rate data transfer, having high frequency efficiency, and reducing hardware requirements.</p><p> This purpose is claimed<u style="single">1</u>Achieved by the content specified in.</p><p> A further objective is to clarify how to provide noise balance in relation to the envelope of the QAM signal.</p><p> This object is achieved by the subject matter specified in claim 1.</p><p> A further objective is to clarify how to provide noise suppression associated with QAM determination thresholds.</p><p> This object is achieved by claim 4.</p><p> Further effects will be apparent from the following detailed description of preferred embodiments of the present invention.</p>

According to the first preferred embodiment of the present invention, noise equalization is achieved by directly using the orthogonal error components E_Q and E_I, which are independent of the signal envelope.

According to the first embodiment shown in FIG. 8, the constellation region is divided into four regions, and the shifted error control signal E'is determined as follows depending on the detected symbol vector D: IV: Region A | D_Q | | D_I |: E' E_I V: Region B | D_Q | &lt;| D_I |: E' E_Q

Looking at another aspect of the first embodiment, as shown in FIG. 9, the four lines passing through the origin divide the region into three regions A, B, and C. These three regions, and the offset error control signal E'in each region, are defined as follows: VI: Region A | D_Q | 2 · | D_I |: E' E_I VII: Region B | D_Q | &lt;1/2 · | D_I |: E' E_Q VIII: Region C (other than A, B): E' (E_I + E_Q) / 2 As you can see, it surrounds the I and Q axes. In regions A and B, when it is detected that the detected error symbol D belongs to any region, one of the orthogonal error components E_I or E_Q is used directly unprocessed. In region C, which is a corner region, it is used as an error control signal E'in which the average values of the orthogonal error components E_I and E_Q are deviated.

The error control value E'is derived, for example, by the PLL loopback filter 13 of the receiver shown in FIG.

It is clear that this can be easily realized by the above-mentioned program instructions having extremely few comparison and calculation steps.

Another source of noise is related to the threshold of the QAM determination device (ie, symbol detection performed by Demapper 14). Noise contributions can cause errors if the detection symbol approaches, for example, the boundaries of a square of size T shown in Figures 5 and 6. If the noise component is greater than the distance to the symbol boundary, the error signal may get a false sign that greatly interferes with PLL tracking.

According to the present invention, this effect can be reduced by applying the above weighting function that suppresses the detector output at a position close to the boundary.

Weighting can be achieved in many ways. However, according to the present invention, it is a basic object to reduce a sudden change in the detector output when the detection signal D is close to the determination boundary. Known demodulators have a typical "saw tooth" shape for detection vs. phase error response. Suppression of the reaction outside the detection boundary is a further objective.

According to a second preferred embodiment of the invention, the following relational expression is used to weight the symbolic error that occurs: IX: WE = E'(1-1 / 2T E' 1 / 2T). 2W / T) Here, as shown in FIG. 12, T corresponds to the boundary size, E'corresponds to the deviated control error, and W = Max {abs (E_I); abs (E_Q)}.

As a general rule, if the symbol boundaries to which E_Q and E_I belong are detected, the value of E'will not occur outside the above ± 1 / 2T range. Therefore, as an example, it can be a zero weighting function outside the above interval. X: E'&lt;-1 / 2T 1 / 2T &lt;E', WE = 0 Normally, when the error signal (E) approaches zero, the weighted error signal (WE) also approaches zero. In addition, the weight function is a positive value for a positive value near zero and a negative number for a negative number near zero. Furthermore, the weighting function approaches zero when the error signal vector approaches the symbol boundary of the detected symbol.

FIG. 10 shows an embodiment for implementing the above method, in which the weighting filter 17 performs weighting of the derived error control signal E'and outputs the feedback weighted control value WE to the IF generator 12. To do. As is clear from FIG. 10, the option signal 18 from the demapper 14 allows or excludes weighting of specific symbols.

Due to the above weighting, the shape of the detector vs. phase error response results in a parabolic shape. Error weighting is illustrated in line 26 of FIG.

As is clear from FIG. 11, the false phase error that would have occurred if the weighting function had not been used has been substantially reduced. Therefore, the possibility of misalignment of the PLL loop is also reduced.

Figure 12 shows other weighting functions that use discrete values. As is clear from the figure, the shape of the weight function is similar to the shape of FIG. But for the sake of simplicity, we use a large number of discrete values.

As explained above, the purpose is to suppress the response near the decision boundary. However, according to a third embodiment of the present invention, this is only applicable to the central symbol of the I / Q plane.

Figure 13 shows a typical example of the 128QAM method. The eight outer signal positions 32 in the constellation define what comes from the boundary to the outer region 34. If the signal is detected in the outer corner region 34, the missed error control signal is not weighted (WE = E').<u style="single">.. This</u>By doing so, the burst error performance can be improved.

In this way, according to the present invention, it is possible to realize an improved demodulator in which errors are reduced and tracking capability is enhanced even though noise is generated in the demodulator.

<figref num="1">It is a figure which shows the known transmitter.</figref><figref num="2">It is a figure which shows the known receiver.</figref><figref num="3">It is a known typical frame timing diagram.</figref><figref num="4">It is a figure which shows the detail of FIG.</figref><figref num="4A">It is another known frame diagram.</figref><figref num="5">Shows thermal noise in a 16QAM modulation constellation.</figref><figref num="6">A detailed diagram related to the error vector E is shown.</figref><figref num="7">It is a figure which shows the curve of various control error values E'derived according to the known method in 64QAM method.</figref><figref num="8">A typical quadrant error correction method according to the first embodiment of the present invention is shown.</figref><figref num="9">A typical octa-circle error correction method according to an alternative example of the first embodiment of the present invention is shown.</figref><figref num="10">A typical receiver for parabolic error weighting according to a second embodiment of the present invention is shown.</figref><figref num="11">A typical first parabolic error weighting function according to a second embodiment of the present invention is shown.</figref><figref num="12">A typical second parabolic error weighting function according to a second embodiment of the present invention is shown.</figref><figref num="13">A typical 128QAM scheme according to a third embodiment of the present invention is shown.</figref>