Rate-robustness tradeoffs in multicarrier wireless communications

Emerging wireless communication systems exploit various resources to increase their robustness and data rate. Since these resources are limited, there is a tradeoff between the need for robust communication and the desire for high throughput. The aim of this dissertation is to study and optimally balance this tradeoff for a few important cases in multicarrier communications. First, multi-code code division multiple access (CDMA) techniques tradeoff the number of supportable subscribers with the per subscriber data rate. However, the interference scales linearly with the data rate of each user since they use multiple codes. To resolve this interference problem, a novel multi-code multicarrier CDMA system is proposed, and this system clearly outperforms previous systems in terms of bit error probability and user capacity. This shows that flexible data rates can be successfully balanced with robustness in a multiuser multi-rate CDMA system by carefully choosing the data rates number of subcarriers. Second, in multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM), pilots are used to estimate the channel, but in addition to consuming bandwidth, they reduce the transmitted energy for data symbols under a fixed transmit power constraint. This suggests a tradeoff between the power allowed to data symbols and the accuracy of channel estimation. The optimal pilot-to-data power ratio (PDPR) for maximizing a capacity lower bound is formulated and derived for four likely pilot patterns and two different channel conditions. The optimal PDPR shows about 10%∼30% higher capacity lower bound than equal power allocation. Third, and closely related to the second contribution, adaptive M-QAM, spectral efficiency, and symbol error rate (SER) are considered since these are respectively the dominant modulation type and quality metrics in emerging standards. The effect of the system structure on the PDPR is analytically shown, and the optimal PDPR for minimizing the average SER and maximizing the spectral efficiency is derived for two well-known linear receivers; zero-forcing and minimum meansquare error. The distributions of the SNR (including channel estimation error) for these receivers are derived and used to find the optimal PDPR. Exact guidelines are provided for the power allocation between data and pilot symbols for these cases.