Digital predistortion is an efficient and robust technique for improving the linearity of power amplifiers. In this paper, we propose a digital predistorter based on frequency domain for OFDM signals, which exhibits superior accuracy against the predistortion in time domain. Simulation results show that, the desired MER can be achieved with fractional delay error as large as 6% of sampling time. It is also shown that since in frequency domain a fractional delay translates to phase rotation, phase rotation estimation and compensation can be performed by a low complexity scheme.

No abstract

No abstract

No abstract

quantization and the minimal hardware required for Abstract-In this paper, different structures for an elliptic filter implementation, both at the same time. with fixed point arithmetic are implemented and compared. TheIn this paper different structures of a digital filter is filter must be quantized for hardware implementation. This investigated and the optimal structure with the minimal quantization is done in two steps. First the coefficients of filter g hardware required for implementation and the minimum are quantized and then the accuracy of internal nodes are . .limited. According to the simulation results, lattice and DF2-round-offnoiseisitroduced.parallel structures have minimal sensitivity to coefficient The rest of paper is organized as follows. In Section II, quantization. Also, the area (gate count) that each of the different structures for implementation of a digital IIR filter structures occupy on the chip are computed. We show that are reviewed and our experimental setup is given. In Section overall, the DF1-parallel structure is the optimal structure for III hardware implementation process is explained. In the next hardware implementation that requires minimal chip area at a section, bit-true modeling is described. HDL modeling is reasonable precision.Keywords: round-off noise, bit-true modeling digital filter explained in Section V. Before concluding the article in implementation.Section VII, our simulation results and the optimal structure for fixed-point implementation are presented in Section VI.DIFFERENT FILTER TYPES AND STRUCTURES Aparticular linear time-invariant discrete-time system can 1DIFRNFLTRYPSADTUCRE be implemented by a variety of computational structure.The direct forms (DF) are the simplest structure for One motivation for considering alternatives to the simple implementation of digital filters [6]. DF2 can save up to 50% direct form structures is that different structures that are in the number of required delay elements. theoretically equivalent may behave differently when Cascade and parallel forms consist of second order direct form implemented with finite numerical precision.sections. In both cases, each pair of complex conjugate poles We are almost always interested in implementations that is realized independently of all other poles. A DF transposed require the least amount of hardware or software complexity. (DFt) structure is obtained by changing the direction of all However, we cannot find the optimal structure on this branches and the input and output signals position, starting criterion alone, since some of the minimal hardware structures from a DF structure. In cascade and parallel forms, one can are very sensitive to quantization noise (the effect of finite-use DFt second order sections (a.k.a. biquad blocks).In this precision computations is modeled with this noise).paper, an IIR low-pass filter used for voice filtering is studied. Much work has been devoted to estimation of quantization In this application, only the amplitude characteristic of the noise and its re...