Patent Application
20090104882
This application is based upon and claims the benefit of priority from the prior Japanese Patent Application No. 2007-271270, filed on Oct. 18, 2007, and the prior Japanese Patent Application No. 2008-119473, filed on May 1, 2008,the entire contents of which are incorporated herein by reference.
1. Field of the Invention
The present invention relates to a radio transmission apparatus performing distortion compensation processing using a distortion compensation coefficient, and more particularly radio transmission apparatus performing distortion compensation processing to a transmission signal in the time division duplex (TDD).
2. Description of the Related Art
As a low noise amplifier for use in a power amplifier for amplifying a linearly modulated signal and a receiver unit for a linearly modulated signal, an amplifier having high linearity is required to suppress deterioration in a spectral characteristic and a transmission characteristic caused by signal distortion.
In particular, when a multilevel amplitude modulation scheme is applied to radio communication, a technique is required on the transmission side to suppress nonlinear distortion and reduce adjacent channel leakage power by linearizing the amplification characteristic of the power amplifier. Further, although high power efficiency is generally required for the amplifier, the linearity and efficiency in the amplifier are in general contrary characteristics. When it is intended to improve power efficiency using an amplifier having inferiority in linearity, it is an essential technique to compensate the distortion produced thereby.
As one of such the distortion compensation scheme, a predistortion scheme is known. The principle of the predistortion scheme is that to obtain a desired signal having no distortion in an amplifier output by adding beforehand an inverse characteristic of the amplifier distortion characteristic to an amplifier input signal. As to the predistortion scheme, for example, detailed descriptions have been given in the following patent documents 1, 2, 3 and 4.
A transmission signal before distortion compensation is compared with a demodulated feedback signal, and using the error thereof, a distortion compensation coefficient is calculated and updated. The distortion compensation coefficient is stored into a memory of which address is determined by the amplitude or the power of the transmission signal, or the function thereof. Then, using the updated distortion compensation coefficient, predistortion is processed to a transmission signal which is to be transmitted next, and the process result is output. By repeating the above process, finally, the result converges on an optimum distortion compensation coefficient. The distortion in the transmission power amplifier is compensated accordingly.
Meanwhile, in recent years, in addition to the 3G scheme, which is the mainstream at present to realize simultaneous transmission and reception by the FDD (Frequency Division Duplex), the TDD (Time Division Duplex) scheme such as WiMAX is being put into practical use for radio communication.
When the conventional predistortion is processed on a transmission signal in the TDD radio communication system, transmission and reception is switched at high speed. Accordingly, abrupt bias current flows in a power amplifier immediately after the start of a transmission period. This produces fluctuation in the amplitude and phase component immediately after the rise of the transmission signal. As a result, a relatively large error becomes included in an updated distortion compensation coefficient. This makes it impossible to calculate a correct distortion compensation coefficient, which causes a problem that a sufficient distortion compensation effect is not obtained.
Further, in the WiMAX system using the TDD, in case of transmitting through a full subcarrier, the transmission level of the top symbol (preamble) of the transmission signal becomes approximately 3 dB greater than the transmission level of a data symbol period. This produces a narrower linear region. As a result, a lager error in the distortion compensation coefficient is produced, and thereby correct distortion compensation operation is impeded.
Also, in case of the TDD scheme, the transmission period and the reception period are switched alternately. The distortion compensation coefficient is not updated during the reception period. If the distortion compensation coefficient is updated only during the transmission period, a longer time is required for the convergence of the distortion compensation coefficient, as compared to the case of the FDD scheme.
Further, in the case of the TDD scheme, the number of transmission subcarriers of the transmission signals of the WiMAX system varies symbol-by-symbol. This causes a fluctuated transmission power level in each symbol period. In some cases, the transmission level is varied after the distortion compensation coefficient at a certain transmission level is calculated. This makes it difficult to obtain an optimal distortion compensation coefficient at all times. Therefore, it is necessary to shorten the convergence time of the distortion compensation coefficient, and to obtain a faster processing speed.
Accordingly, it is an object of the present invention to provide a radio transmission apparatus, capable of reducing a distortion compensation coefficient error and shortening the convergence time of the distortion compensation coefficient.
In the time division duplex (TDD), a correct distortion compensation coefficient cannot be calculated because of large fluctuations in the amplitude and phase components immediately after the start timing of the transmission period. Accordingly, the radio communication apparatus does not update the distortion compensation coefficient for a certain period, which is shorter than the transmission period, after the transmission period start timing. Instead, update operation of the distortion compensation coefficient is carried out after the lapse of a certain period from the start timing. Additionally, distortion compensation operation on the transmission signal is performed during a certain period from the transmission period start timing in which no update operation is carried out.
Further, in order to shorten the convergence time of the distortion compensation coefficient, the radio transmission apparatus performs interpolation processing of the distortion compensation coefficient. The distortion compensation coefficient is given to a two-dimensional coordinate (address): one is the power level of an input transmission signal and the other is the difference of power between the input transmission signal and a feedback signal. The distortion compensation coefficient at a predetermined coordinate is updated to an interpolated value, which is obtained through interpolation calculation using the distortion compensation coefficient at a coordinate adjacent to the predetermined coordinate.
Further, in the transmission period, the radio transmission apparatus is made to vary predetermined timing for starting the update processing of the distortion compensation coefficient on the basis of each transmission period.
As such, update processing of the distortion compensation coefficient is not performed for a certain period after the transmission period start timing in the time division duplex (TDD). Instead, the update operation of the distortion compensation coefficient is performed after the lapse of a certain time. Accordingly, it is possible to avoid the update of the distortion compensation coefficient in a period having large fluctuations in the amplitude and phase of the transmission signal. Thus, an increased distortion compensation coefficient error is prevented, and normal distortion compensation operation can be maintained accordingly.
Further, an interpolation calculation is carried out on a predetermined distortion compensation coefficient to be stored into a memory, using a distortion compensation coefficient which is adjacent in the memory to the distortion compensation coefficient of interest. An interpolated value of the predetermined distortion compensation coefficient is obtained, so as to be updated to the above interpolated value. Thus, it is possible to shorten the convergence time of the distortion compensation coefficient.
Further scopes and features of the present invention will become more apparent by the following description of the embodiments with the accompanied drawings.
Hereafter, the preferred embodiment of the present invention will be described with reference to the charts and drawings. However, the described embodiment does not limit the technical scope of the present invention.
A timing generator 3 generates timing to start updating a distortion compensation coefficient at the timing after a lapse of a certain period from transmission period start timing, and also generates timing to complete the update of the distortion compensation coefficient at transmission period end timing. More specifically, a signal information detector 31 in timing generator 3 detects map information (information indicative of a data position) of the transmission signal from the parallel data string. From the map information and switchover timing between a transmission period and a reception period stored in advance, a preamble detector 32 obtains the transmission period of a top symbol (preamble) in the transmission signal. Then, a distortion compensation timing controller 33 outputs a distortion compensation update timing signal to a distortion compensator 4 (making a distortion compensation update timing signal ON) at preamble transmission end timing. Also, distortion compensation timing controller 33 makes the distortion compensation update timing signal OFF at the transmission period end timing to complete the update of the distortion compensation. With regard to distortion compensation operation, distortion compensation timing controller 33 outputs to distortion compensator 4 a timing signal to start the distortion compensation operation at the transmission period start timing (making a distortion compensation operation timing signal ON). Also, distortion compensation timing controller 33 makes the distortion compensation operation timing signal OFF at the transmission period end timing. Namely, although the distortion compensation operation is performed in the period from the transmission period start timing to the preamble transmission end timing, the distortion compensation is not updated because the period has large fluctuations in the amplitude and phase of the transmission signal. Here, the period in which the distortion compensation is not updated is not limited to the transmission period start timing to the preamble transmission end timing, but can properly be modified depending on the state of fluctuations in the amplitude and phase of the transmission signal. In a WiMAX system using the TDD, the transmission level in the top symbol (preamble) of the transmission signal is approximately 3 dB greater than the transmission level in a data symbol period, by which a conspicuous error may occur easily. Therefore, as one example of the period in which the distortion compensation is not updated, the period from the transmission period start timing to the preamble transmission end timing is exemplified above.
Referring back to
The distortion-compensated transmission signal being output from distortion compensator 4 is input into an orthogonal modulator 6 through a D/A converter 5. Orthogonal modulator 6 outputs by performing orthogonal modulation by adding the results of multiplying the input transmission signal and a 90° phase-shifted transmission signal respectively by reference carrier waves from a reference carrier wave generator 7.
A frequency converter 8 converts the transmission signal into a radio frequency signal by mixing the orthogonally modulated signal with a local oscillator signal. After being amplified in a power amplifier 9, the radio frequency signal is transmitted using time division duplex. Also, the transmission signal is looped back inside the radio transmission apparatus through a directional coupler. After being converted into an IF signal by means of a frequency converter 11 and an orthogonal detector 12, the above signal is passed through an A/D converter 13, and fed back to distortion compensator 4 as a feedback signal.
Distortion compensation update processing to be performed in distortion compensator 4 includes calculation processing based on the transmission signal and the feedback signal, calculation of the distortion compensation coefficient, and periodical update processing. Since the transmission period and the reception period are alternately switched over in the TDD scheme, a time for use in updating the distortion compensation coefficient is short. As a result, a large amount of time is consumed until the coefficient is converged. To cope with the above problem, interpolation calculation processing of the distortion compensation coefficient is performed in interpolation processor 43 of distortion compensator 4, so as to make the convergence time faster and stabilize the coefficient. The interpolation calculation processing of the distortion compensation coefficient will be described in the following.
In the interpolation processing of the one-dimensional direction, interpolation calculation is performed for a coordinate to be interpolated (for example, B in
where
B′: distortion compensation coefficient at the coordinate B after interpolation calculation (which is also referred to as interpolated value)
B: distortion compensation coefficient at the coordinate B before interpolation calculation
A: distortion compensation coefficient at the coordinate A adjacent to the coordinate B in the one-dimensional direction
C: distortion compensation coefficient at the coordinate C adjacent to the coordinate B in the one-dimensional direction
M: interpolation coefficient
For example, in case the interpolation coefficient M=3,
B′=(4B+A+C)/6
is obtained. Namely, the distortion compensation coefficient at the coordinate B is obtained through the interpolation calculation of the ratio 2/3 of the distortion compensation coefficient at the coordinate B before the interpolation calculation, the ratio 1/6 of the distortion compensation coefficient at the coordinate A, and the ratio 1/6 of the distortion compensation coefficient at the coordinate C.
Subsequent to the coordinate B, with regard to the coordinate C, interpolation calculation in the one-dimensional direction using the above expression (1) is performed, using the distortion compensation coefficients at the both adjacent coordinates B and D, so as to obtain the interpolated value of the coordinate C. Subsequently, interpolation calculations are successively performed with regard to the coordinates adjacent to the right.
Additionally, when the distortion compensation coefficient at an adjacent coordinate is unchanged from the initial value, instead of the above coefficient, a distortion compensation coefficient at a coordinate not having the initial distortion compensation coefficient value in the one-dimensional direction is used. For example, in
The interpolation calculations in the one-dimensional direction are carried out successively from the left end side (excluding the coordinates on both ends) of a row having a value in the two-dimensional direction of 0 (gradient 0). More specifically, by proceeding to the right direction from the coordinate B, the interpolation calculations are completed up to the last coordinate (G) excluding the coordinate (H) on the right end. Then, with regard to the row lower by one than the row having the gradient 0, the interpolation calculations are carried out from the left end side toward the right end side. Subsequently, after being shifted to the row upper by one than the row having the gradient 0, the interpolation calculations are continued. As such, the interpolation calculations are alternately performed for the rows in the plus direction and the minus direction, with the row having the gradient 0 as a center.
In the two-dimensional direction also, the interpolation processing similar to the interpolation processing in the one-dimensional direction is performed. In
where
P′: distortion compensation coefficient at the coordinate B after interpolation calculation (which is also referred to as interpolated value)
P: distortion compensation coefficient at the coordinate B before interpolation calculation
A: distortion compensation coefficient at the coordinate A adjacent to the coordinate B in the one-dimensional direction
Q: distortion compensation coefficient at the coordinate C adjacent to the coordinate B in the one-dimensional direction
N: interpolation coefficient
For example, in case the interpolation coefficient N=3,
P′=(4P+A+Q)/6
is obtained. Namely, the distortion compensation coefficient at the coordinate P is obtained through the interpolation calculation of the ratio 2/3 of the distortion compensation coefficient at the coordinate P before the interpolation calculation, the ratio 1/6 of the distortion compensation coefficient at the coordinate A, and the ratio 1/6 of the distortion compensation coefficient at the coordinate Q.
Subsequent to the coordinate P, interpolation calculation with regard to the coordinate Q in the two-dimensional direction is performed by use of the above expression (2). In the above calculation, the distortion compensation coefficients at the both adjacent coordinates P and R are used. Thus, the interpolated value of the coordinate Q is obtained. Thereafter, interpolation calculations are successively performed with regard to the coordinates adjacent beneath.
Additionally, similar to the interpolation processing in the one-dimensional direction, when the distortion compensation coefficient at the adjacent coordinate is unchanged from the initial value, the above coefficient is not used. Instead, a distortion compensation coefficient at a coordinate not having the initial distortion compensation coefficient value in the two-dimensional direction is used. For example, in
The interpolation calculations in the two-dimensional direction are carried out, for example, from the column on the left end side to the lower end direction from the coordinate P on the lower left end side, with the row of gradient 0 as a center. On completion of the interpolation calculations up to the last coordinate S excluding the coordinate T on the lowermost end, the processing proceeds from a coordinate V on the upper left side to the upper end direction. Thus, the interpolation calculations are performed to the last coordinate Y excluding the uppermost end coordinate Z. Subsequently, after being shifted successively to the columns on the right side, the interpolation calculations are carried out alternately on the rows of the lower end side and the rows of the upper end side, with the row having the gradient 0 as a center.
Preferably, the aforementioned interpolation calculations in the one-dimensional direction and the interpolation calculations in the two-dimensional direction are carried out in one-time interpolation processing. However, it may also be possible to perform either one thereof. Additionally, when only the interpolation calculations of the two-dimensional direction are performed, first, the interpolation calculations in the one-dimensional direction are performed for the row having the gradient 0. Thereafter, the interpolation calculations in the two-dimensional direction are performed. The reason is that the distortion compensation coefficient in the row having the gradient 0 functions as a criterion.
Since distortion compensation coefficients are closely correlative with coefficients adjacent thereto, a distortion compensation coefficient (for example, a reference symbol ‘a’ shown in
When the interpolation calculations with regard to the distortion compensation coefficients for the entire coordinates in memory (LUT) 42 are not completed in one reception period, the calculations for the remaining portion are carried out in the next reception period.
Further, the interpolation processing is not carried out repetitively at any time. When one-time interpolation processing, that is, interpolation calculations for the distortion compensation coefficients in the entire coordinates of memory (LUT) 42 is completed once, thereafter for a certain period, the interpolation processing is not performed. The reason is that if the calculations are performed repetitively at all times, the distortion compensation coefficient becomes too smooth, which undesirably causes a deteriorated distortion characteristic. For example, only one-time interpolation processing is carried out in a first reception period after the start of communication (or a reception period in the second time or thereafter in case of necessity). Alternatively, the interpolation processing is executed intermittently. As described earlier, the first-time interpolation processing is carried out in the first reception period after the start of communication (or a reception period in the second period or thereafter in case of necessity). When second-time interpolation processing is to be performed, for example, the processing is performed after the distortion compensation coefficient becomes a converged state. Thereafter, the processing is performed at intervals of a certain period. By performing after the convergence, it is possible to improve the stability of the distortion compensation coefficient further.
The aforementioned interpolation processing of the distortion compensation coefficient according to the present embodiment is not limited to radio transmission apparatus in the TDD scheme. The above processing is also applicable to a scheme in which simultaneous transmission and reception are performed continuously, such as the FDD scheme enabling simultaneous transmission and reception by use of different transmission and reception frequencies.
In
When acquiring a transmission signal for updating the coefficient continuously at the same timing after the lapse of a certain time from the transmission period start timing (after the completion of preamble transmission), if the transmission signal is a pattern data (such as synchronization pattern, test data and dummy data), acquisition of similar data having substantially no fluctuations in the amplitude and phase continues. If the update processing of the distortion compensation coefficient is repeated by substantially the same data, an error component is accumulated in the distortion compensation coefficient of a particular range. This produces the unnecessary spurious and a deteriorated distortion compensation characteristic. Further, because of the above characteristic deterioration, the phenomenon of overall noise floor rise is produced, which remarkably appears in a portion corresponding to low power in the LUT. Using transmission signals having a variety of amplitudes and phases for distortion compensation update processing, the distortion compensation coefficient is updated randomly in an average manner. Accordingly, it is possible to suppress the generation of the unnecessary spurious.
Additionally, by performing the aforementioned interpolation processing to the distortion compensation coefficient to suppress an error component, the distortion compensation coefficient can be smoothed also, and the unnecessary spurious and the overall noise floor rise can be suppressed by the smoothed amount. Accordingly, by using both the interpolation processing and the timing adjustment processing described above, it becomes possible to suppress error components more effectively.
As described above, if the interpolation processing is carried out excessively, even a distortion compensation coefficient originally corresponding to the peak point is smoothed. This undesirably may cause a deteriorated high-power transmission function as a transmission amplifier. Therefore, it is particularly effective to apply the aforementioned timing adjustment processing in case that the unnecessary spurious caused by a pattern data cannot be sufficiently removed only by the interpolation processing.
In the exemplary third configuration, by means of the timing adjustment processing by timing adjuster 14, the output timing of the distortion compensation coefficient update timing signal is shifted by a certain interval width in each transmission period. Thus, data acquisition timing for use in updating the distortion compensation coefficient is shifted. By this, it is possible to acquire transmission signals having a variety of (moderately varying) amplitudes and phases, update the distortion compensation coefficient randomly in average, suppress the growth of an error component, and suppress the generation of the unnecessary spurious.
In contrast, in
The suspension time is made to vary for each transmission period (including a case of no suspension time). When letting a unit time of the suspension time to be α, as shown in the figure, there are set no suspension time for a transmission period 1, a suspension time α for a transmission period 2, a suspension time 2α for a transmission period 3, and a suspension time 3α for a transmission period 4. By this, the start timing of the distortion compensation coefficient update processing is shifted by a time α in each transmission period. Also, the data acquisition timing in the transmission period is shifted in each transmission period. In a transmission period 5, the suspension time is returned to zero. A possible number of times of shifts is obtained in advance from the length of the transmission period and the suspension time. The start timing is shifted in each transmission period. On reaching the possible number of times of shifts, the process is returned to no suspension time. Then, the shift of the suspension time on the basis of a unit time α is repeated. The length of the time α can properly be set to the extent that the amplitude variation in the acquired data disperses randomly.
If the start timing of the distortion compensation coefficient update processing is delayed due to the insertion of the suspension time, for example, as in the transmission period 3, the entire processing of the distortion compensation coefficient update processing may not be completed before the transmission period end timing. In such a case, the remaining processing (processing C) is processed in the subsequent transmission period (transmission period 4). Also, as shown by the transmission period 3, as to the processing which is in progress at the transmission period end timing (the coefficient update processing), the processing is continued in the subsequent reception period.
When timing adjuster 14 acquires preamble transmission end timing information with regard to each transmission period from preamble detector 32 (S108), timing adjuster 14 decides whether the count CNT1=0 (S110). In case of Yes, the process proceeds to the processing of the step S120 and after, which will be described later. In case of No, timing adjuster 14 measures the lapse of the unit time α (S112), and increments the real count CNT2 by one each time the unit time α elapses (S114). Timing adjuster 14 then compares the count CNT1 with the real count CNT2 (S116). If the incremented CNT2 is smaller than the set CNT1, timing adjuster 14 measures the lapse of the unit time α again (S112), and repeats the above measurement of the unit time a until the real count CNT2 reaches the count CNT1. The start timing of the distortion compensation coefficient update processing is shifted by the amount of the measured lapse of the unit time α.
When the real count CNT2 reaches the count CNT1, an update processing start signal is transmitted to distortion compensation timing controller 33 (S118). On receiving the update processing start signal, distortion compensation timing controller 33 outputs a distortion compensation coefficient update timing signal. Thereby the distortion compensation coefficient update processing is started. Thereafter, on acquiring transmission period end timing information from preamble detector 32 (S120), distortion compensation timing controller 33 increments the count CNT1 by one (S122). By incrementing the count CNT1, the start timing of the distortion compensation coefficient update processing is further delayed by the unit time α.
If the count CNT1 after being incremented is not greater than the possible number (n) of insertion times (S128), in the subsequent transmission period, the process returns to step S106. After the real count CNT2 is initialized and preamble transmission end timing information is acquired (S108), the operation from steps S110 through S128 is repeated. If the count CNT1 after being incremented exceeds the possible number (n) of insertion times (S128), in the subsequent transmission period, both the count CNT1 and the real count CNT2 are initialized (S104, S106). After the preamble transmission end timing information is acquired (S108), the operation from steps S110 through S128 is repeated.
In addition, the processing in steps S124 and S126 acquires transmission (Tx) symbol information in each transmission period (S124), and confirms the transmission period length. When the transmission period length is varied (S126), the process is returned to step S102. Then, the possible number (n) of insertion times is recalculated, and the process is transferred to the processing of step S104 and thereafter. However, usually, the transmission period length is not varied halfway.
Similar to
The radio transmission apparatus according to the present invention is applicable to a base station or a mobile station in the radio communication system.
The foregoing description of the embodiments is not intended to limit the invention to the particular details of the examples illustrated. Any suitable modification and equivalents may be resorted to the scope of the invention. All features and advantages of the invention which fall within the scope of the invention are covered by the appended claims.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2007-271270 | Oct 2007 | JP | national |
| 2008-119473 | May 2008 | JP | national |
