Researches On Implementation Techniques Of Key Algorithms For Continuous Phase Modulation Systems

Unmanned Aerial Vehicles(UAV) play an increasingly important role in military and civilian fields. The UAV data link is a integrated communication system which transmit and exchange information between UAV and its ground control station. The performance of data link system directly affects the overall performance of the UAV. Continuous Phase Modulation(CPM) is a type of constant envelope digital modulation technique with high spectral efficiency and power efficiency. In this paper, we propose that CPM modulation can be employed in UAV data link. The key algorithms for serially concatenated CPM communication systems, including frequency offset estimation, demodulation and decoding algorithm, and their efficient VLSI implementation are investigated in this paper. This paper is expected to promote the CPM technology and provide some guidance to design the UAV data link communications systems. The major contributions can be summarized as follows.1. This paper proposes a serially concatenated TCM-CPM systems based on LDPC codes.TCM technology improves spectrum efficiency while the bandwidth does not need expand. We combine the TCM technology with CPM modulation and apply it in UAV data link communication systems, and employ LDPC codes as the outer code in order to improve system reliability. The proposed TCM-CPM-LDPC serially concatenated system try to achieve a better combination of efficiency and reliability. At the receiver,the soft information obtained by TCM-CPM demodulation is directly transfered to the LDPC decoder for channel decoding. The soft information does not need iterate between the demodulation and decoding blocks. Moreover, the proposed SC-CPM system model eliminates the interleave and deinterleave operations, significantly reduces the receiver hardware complexity and processing delay. Simulation and implementation results show that the proposed serially concatenated TCM-CPM system based on LDPC codes has a higher coding gain and can tolerate a wide range of frequency offset.2. To estimate the Doppler shift caused by UAV at high speed mobile environment, we use a data assisted method by a special pilot sequence, and transform the frequency offset estimation to the problem of estimation the frequency of single-tone complex sinusoid signal. We propose a simplified frequency estimation algorithm based on Fourier interpolation. The performance of the proposed algorithm close to the theoretical lower bound, and the operation mainly includes FFT calculation, the maximum search and two DFT calculations and other steps. During the VLSI implementation of frequency estimation, this paper starts from the FFT processor and proposed a low cost CORDIC-based configurable FFT processor. The CORDIC units are used to compute complex multiplication with no dedicated multiplier. The rotation angles of the CORDIC units are generated internally based on the transform of butterfly counter,which eliminates the need of ROM making it memory-efficient. Based on the expansion of basic CORDIC unit, the triple-mode CORDIC is proposed to effectively support the complex multiplication, complex absolute calculation and real division in the frequency estimation algorithm. And we design and implementation of a computationally efficient frequency estimator based on FPGA.3. An enhanced implementation of max*operator is proposed for the Log-MAP algorithm. The improved max*operator uses maximum operation and piecewise linear function approximation with a specific coefficients and constants. It requires only shift and addition operation to realize the max*operator efficiently. Simulation results show that the performance of the Log-MAP alogrithm with enhanced max*operator compare to the original Log-MAP algorithm performance is less than 0.05 d B. Based on the LogMAP algorithm framework, a modified branch metric calculation is introduced and a soft-decision noncoherent CPM demodulation algorithm is presented. The proposed noncoherent CPM demodulation algorithm is suitable for carrier recovery difficult situation because it can can tolerate large frequency offset and phase noise. In order to reduce storage overhead and latency, we apply the sliding window technique to CPM demodulation. Based on sliding window technique and enhanced max*operator, we design and implement a non-coherent demodulator for TCM-CPM system.4. This paper studies the encoding algorithms of the LDPC codes with quasi-cyclic dualdiagonal parity check matrix. We demonstrate that matrix decomposition encoding algorithm and cumulative recursion encoding algorithm are equivalent for implementation. The cumulative recursion encoding algorithm is straightforwardly facilitated to hardware implementation. We proposed a partly parallel encoding architecture for the QC-LDPC codes with dual-diagonal parity check matrix and design an area-efficient and high-throughput LDPC encoder compatible with IEEE 802.11 n standard. Besides,this paper investigate the LDPC decoding algorithms, such as BP algorithm, minimumsum algorithm and modified minimum-sum algorithm. We also analyze two different message passing mechanism, including flooding and layered scheduling. A generic decoder architecture for QC-LDPC codes based on layered minimum-sum algorithm is proposed. By utilizing parallelism 73 and 27 respectively, we design and implemen- tation of a(8176,7154) QC-LDPC decoder for CCSDS standard and a LDPC decoder for IEEE 802.11 n standard. FPGA implementation results show that the area overhead of the designed decoder is low while its throughput is high.