Multirate Codes and Multicarrier Modulation for Future Communication Systems

Doctoral thesis, 1999

In this thesis, new detection and modulation methods for multicarrier systems are presented. The proposed multicarrier detectors are based on feedback of previously detected symbols and the idea is to exploit the correlation, both in time between consecutive symbols, and in frequency between neighboring sub-carriers. This enables us to calculate a reliable channel estimate, to be used for coherent detection. The proposed detectors have significantly lower irreducible bit error probabilities compared to a conventional differential detector. Furthermore, a Parallel Combinatory - Orthogonal Frequency Division Multiplexing (PC-OFDM) system is proposed and analyzed. Here, some of the information bits select, in each symbol interval, a subset of the available sub-carriers. The selected sub-carriers are then modulated by points from an M-PSK signal constellation. PC-OFDM systems can be designed to have lower bit error probability on Gaussian channels, lower Peak-to-Average Power Ratio (PAPR) and higher bandwidth efficiency compared to conventional OFDM systems. The thesis is further focused on convolutional channel codes of multiple rates. New high-rate encoders obtained by puncturing, and low-rate encoders obtained by nesting are presented. These new codes are rate-compatible, they have higher constraint lengths and a wider range of code rates than what has previously been presented. The new codes are applied to a multicode direct-sequence code-division multiple-access (DS-CDMA) system and are shown to provide good performance and rate-matching capabilities. Further presented in this thesis is a family of very low-rate convolutional codes which all have maximum free distance. This low-rate family of codes are suited for bandwidth expansion instead of conventional direct-sequence spreading in CDMA systems. The performance of a CDMA system employing these very low-rate maximum free distance codes for combined coding and spreading is analyzed when successive or parallel interference cancellation is applied in the detector. Results show that the code spread system outperforms the conventionally spread system. Without interference cancellation in a multiuser system the single-user bound is never reached (except for one user). With two stages of parallel interference cancellation, a code-spread system with a load only slightly less than 1 bit/chip obtains a bit error rate very close to that of a single user system even for rather low signal-to-noise ratios. The thesis also derives a new metric for Turbo decoding on Rayleigh fading channels with noisy channel estimates. By including the error variance of the channel estimator in the decoder metric derivation, we calculate the correct channel reliability factor for this case. The gain by using this new metric, compared to what we obtain if we ignore the uncertainty of the channel estimator, may be as large as 1 dB at a bit error probability of 0.001.