UBC Theses and Dissertations
Design, analysis, and implementation of a DSP-based modem for Code-Phase-Shift Keying Link, Robert G.
This thesis presents the design, analysis, and implementation of a radio-frequency modem which employs the Code-Phase-Shift Keying (CPSK) method of Direct-Sequence Spread-Spectrum (DS-SS) signaling. A correlation receiver is designed for a DSP based implementation. The received RF signal is brought to a digitally represented base-band signal by product demodulation at the nominal RF frequency, followed by low-pass filtering and dual-channel 8 bit sampling at 2 MSPS. Base-band signal processing is done on a quad TT TMS320C40 general purpose DSP. The SS spreading factor is user selectable from a range of 7 to 127; with corresponding data rates in the range 100 to 9.4 KBPS. The herein designed tracking and acquisition algorithms are adapted from those of conventional DS-SS systems; while the carrier-phase tracking problem is solved by a new method hereby called Phase-Invariant-Reception (PIR). Extensive bit-error-rate (BER) measurements have been made in Additive White Gaussian Noise (AWGN) and in the presence of single tone interference. All measured BER vs SNR or BER vs JSR curves are compared to those of an ideal optimal receiver. The implementation loss, with respect to the optimal receiver, for single-channel data flow in AWGN is approximately 1.8 dB in power efficiency when the stationary receiver and transmitter use local oscillators of the same nominal frequency. The additional BER power efficiency loss as a function of the difference in frequency of the latter two oscillators, is obtained exactly analytically and is confirmed experimentally. The main contributions to the suboptimal BER performance are due to approximating the optimal correlator with an analog filter plus digital correlator, to tracking the timing slippage, and to using PIR instead of optimal coherent reception. The performance of the analog filter/digital correlator combination is obtained approximately analytically, and found to be independent of the signal-to-noise ratio. The BER degradation incurred by the tracking is derived analytically and found to be negligible for the timing error induced by the typical present day TTL clock oscillators, unless the SNR is very high. The theoretical BER performance of PIR is obtained exactly analytically, and found to approach that of coherent reception as the SNR increases. In addition, measured data for the mean time-to-acquire and the mean time-to-loselock in AWGN is presented; along with an outline on how these curves could be obtained analytically, and their implications for packet transmission.