Wireless communication systems require high data transfer rate is committed to research and development makes it more advanced and efficient coding, modulation techniques to improve the efficiency of radio frequency, but also on the RF front-end to a more demanding requirements. Power amplifier efficiency and linearity is a set of contradictory indicators, the two must be coordinated so that the communication system performance optimization. LINC (Linear Amplification Using Nonlinear Components) technique in relation to this issue is put forward that can at the same time meet the high efficiency and high linearity transmitter technology. LINC transmitter technology framework in 1935 was Chirex first proposed, but the analog circuit was used to achieve the trigonometric functions and more complicated, such as prescribing circuit, resulting in LINC technology has not been enough attention in North America in 1974 until the first time Cox to make in the name of the LINC. LINC AM technology is a phase modulation signal decomposition into two Constant envelope phase modulation signal, and then through the two properties were the same amplifier, the final synthesis of the two signals change the initial output signal envelope. LINC technology uses a highly efficient non-linear power amplifier to enlarge heng envelope phase modulation signal, can be avoided by the non-linear power amplifier caused by the adverse effects, which not only guarantees the high linearity of the transmitter and ensure the efficient transmitter. The efficiency of power amplifier can theoretically reach 100%, which for non-Constant Modulus Modulation of critical importance. Input signal is separated into one of two Constant envelope signal SCS (Signal Component Separate) of the transmitter plays a key role in performance. This article will make the SCS algorithm, and through analysis of the realization of different circuits, a resource-less method. At the same time, given the LINC system simulation.

　　1 SCS algorithm and the realization of circuit analysis

　　Principle 1.1 LINC

　　As shown in Figure 1, after the separation of the signal s1 (t) and s2 (t) for Constant Modulus Signals, so that they can be through a highly efficient but sensitive to the envelope E (F) after the synthesis of type of amplifier output, achieve high efficiency linear amplification. LINC transmitter system but an inherent disadvantage is that the imbalance in the two slip roads to enlarge, resulting in synthesis signal output compared to the original signal has distortion, bringing band-band distortion and interference in the LINC system often necessary to add a feedback slip through some algorithms, such as adaptive filtering to compensate for this distortion caused by the imbalance.

　　1.2 SCS Algorithm and Implementation

　　Separation of vector signals as shown in Figure 2, available from the map e (t) of the expression as follows:

　　Set the input signal s (t) that the Cartesian form of s (t) = si (t) + j * sq (t), the type (5) can be written as follows:

　　Introduced from the above is known that if e (t) values, respectively, with the original input signal plus / minus signal separation can be completed here with DSP realization of SCS. At present, e (t) the main method using look-up table, but the method has some shortcomings, such as: the error signal e (t) the real and imaginary parts of the value stored in the two-dimensional look-up table, the use of the input signal is Department of si (t) and imaginary parts of sq (t) addressing. The main disadvantage of this method requires a large number of storage units to store all the corresponding input signal e (t) values. If the input signal of the real and imaginary parts are quantified for the 12 b, the output e (t) as well as 12 b, the total storage capacity: 2 * 12 * (212) 2 = 403 Mb; another method as follows: to continue to simplify type (2), type (3) available:

　　Therefore can be an inverse cosine function table (by the type (4), can be θ (t)) and cosine are a group of look-up table can be separated on the signal. If the input signal of the real and imaginary parts have been quantified to 12 b, the three look-up table to complete a total of 3 * 12 * 212 = 147.456 Mb storage capacity.

　　SCS method to achieve the above two are occupied by a large number of memory cells, even if the store 1 / 4 sine table, still need to at least 36.75 Mb storage capacity, if the median increase in bit quantization, the storage of the index will be greater than 2 times increase in the actual realization of the system is not desirable, because of all these methods, we have adopted a lower storage capacity of the method, using one-dimensional structure of look-up table storage orthogonal error signal, and then use the input signal component and the Cartesian The multiply-add operation can be obtained after the error signal. In this method, with the input signal power (r2 (t)) the index address, to find the best examples of a few do not need to mean that all possible values of the input signal power, the value of each table by a group of input power range to index stored value by the group in the maximum input power of the scope of the decision, thus significantly reducing the occupied resources.

　　Trials using 210 = 1 024 of the storage table, then the maximum value of storage: use of fixed-point processing DSP, the signal of the simulation after the separation as shown in Figure 3 (s1 in the example, s2 the same), when taking into account the fixed-point storage in the table of the value of an integer part of (for example: a value of 31.98, only 31 store), the introduction of round-off errors, resulting in separation s1, s2 is not a strict 2 Road Constant Modulus (Table of the number of Constant Modulus influence the situation), thereby affecting the final output synthesizer ACI (Adjacent Channel Interference) seriously, so the expansion of the use of stored numerical precision. Such as the extension of a time, namely: that the five will be extended to 31 with 6 b that 63.96 = 31.98 * 2 of the integer part, the actual reach that 63 / 2 = 31.5, reduces the error. Extension of the median precision accuracy can be set. The article has been to find the best examples of the optimal expansion of the number and accuracy of research, the results shown in Figure 4.

　　As can be seen from the graph, the table entry for the 210 = 1 024, the accuracy in the expansion of 8 b for the case, ACI reached -70 dBc. Resources at this time occupied by the number of logic: 13 * 210 = 13.3 kb. Occupation of resources compared to the first two-dimensional look-up table method to reduce the 104-fold.

　　2 SCS Simulation

　　Chosen as a simulation of 16QAM modulation signal, spread spectrum, modulation and after roll-off factor of 0.22 the root raised cosine filter, four times after interpolation into the 15.36 MHz frequency for the DSP for signal separation, the separation of choice to achieve the above-mentioned article two kinds of methods, after separation of the signal spectrum as shown in Figure 5 (s1 in the example).

　　After the separation of the signal amplifier and synthetic F-type output, which can achieve high linearity high efficiency amplification purposes. LINC system but the two branches of the circuit does not completely match, mainly by the two branches of the circuit amplifiers, quadrature modulators, such as synthetic imbalance will cause the output signal can not be completely offset by the introduction of pre-separation error signal e (t), so that spectrum extended to the adjacent channel, resulting in temporary channel interference, so the use of adaptive filtering to compensate for this imbalance, LINC system, the final output signal waveform with the original input signal power spectral density as shown in Figure 6, while the input signal directly also carried out into the amplifier simulation, the simulation results shown in figure 7.

　　From Figure 6, Figure 7, we can see that the use of technology than the direct LINC amplifier output power spectrum has improved about 10 dB.

　　3 Conclusion

　　In this paper, the use of the LINC power amplifier linearization SCS has done a brief analysis of algorithms, and occupied one of the fewer resources are available, the algorithm practical simulation results show that the use of the authentication method than the direct use of F-type amplifier input signal, LINC technology can improve the 10 dB of ACI, the realization of a linear amplification.

　　Authors: Zhang, He Songbai (Electronic Science and Technology Institute of Electronic Engineering, Sichuan University, Chengdu 610054)