Spiridon Zettas
Start date: 1st October 2011
Email: Spiridon.Zettas@brunel.ac.uk
Supervisor: Prof. John Cosmas
PhD Research Title: Modeling and Performance Analysis of DVB-T2/NGH Channel Estimation and Equalizer
Channel Estimator and Equalizer Based on scattered pilot patterns for DVB-T2
The DVB-T2 standard arranges the information data into blocks and modulates them with Quadrature Amplitude Modulation (QAM) before using Inverse Fast Fourier Transform (IFFT) to obtain the Coded Orthogonal Frequency Division Multiplexing (COFDM). Because of broadcast channel imperfections, errors occur, reducing the system’s throughput. The use of a channel estimator and equalizer counteracts the channel induced distortions and thus improves the Bit Error Rate (BER). In our method, a very simple to implement Channel Estimator is proposed that can be used in real world receivers with limited computation power and with reduced memory (RAM) sizes.
The proposed estimator can be used in slowly varying channels which can be considered in practice as stable. In the frequency domain the channel is considered to be frequency selective. The proposed method is based on simple calculations of the channel’s impulse response (H) in the frequencies of the subcarriers modulated by known pilot symbols. In each OFDM symbol reception the channel’s impulse response is updated with information of the successive OFDM symbol and thus the system is able to follow any slow variation in time. Then, the channel’s response for all subcarriers is calculated with well-known one dimensional interpolation techniques, from simple to more computational intensive, specifically: Nearest neighbor, Linear, Cubic spline and Piecewise cubic Hermite interpolation. In figures 1 and 2 the proposed method is depicted. In Fig.1 the k-th carrier C(k,l) of the l-th OFDM symbol is the average of the cells included in the last Time Averaging Area TAA(k,l) which includes a multiple number (called depth) of a Time Averaging Element TAE(k,l), each TAE include just on pilot. The impulse response in the k-th cell C(k,l+1) of the next l+1 OFDM symbol is updated as shown in Fig.2 using the TAA(k,l+1) value. After the C(k,l+1) calculation, the TAA(k,l+1) is updated and equalized to the C(k,l+1) value. The performance of the estimator for SISO transmission is depicted with graphs of BER vs. SNR for different combinations of QAM order, FFT sizes and the eight different scattered pilot patterns supported by the DVB-T2 standard. The simulations clearly show that the proposed estimator–equalizer performs acceptably well for most of the above mentioned combinations.
Fig1. Reception of the l-th OFDM symbol
Fig2. Reception of the l-th+1 OFDM symbol
A further improvement will be achieved by introducing more complex methods for time interpolation namely: linear interpolation, least square estimator LSE, minimum mean square error MMSE and in conjunction with the average estimator the more sophisticated Kalman filters will be used to compensate the distortion of fast time varying channels.
Furthermore, the proposed algorithms will be implemented into MISO and MIMO transmission systems.
