A novel framework for scalable equalization

Abstract

In this paper, we propose a novel framework for the design of equalization techniques that provide an efficient superior performance and exhibit a flexible scalable performance-complexity trade-off. The 3GPP time-duplexing high speed packet access (TD-HSPA) wireless communication system is chosen for application due to its time-relevance. The proposed framework utilizes a low-complexity pre-processing stage that implements linear filter-assisted progressive group detection (PGD), a technique which we have proposed in previous works. PGD is a near-maximum likelihood (ML) detection technique that spans and intelligently prunes the set of possible transmit-symbol combinations, and provides a set of the most probable combinations as interim hypotheses for symbol-decisions. Afterwards, the intermin hypotheses are utilized by an equalizer such as a constrained-Viterbi algorithm or an adapted decision-feedback equalizer, as a reduced set of candidates for transmit-symbol combinations. Hence, the equalizer stage decides on the best candidate according to the equalizer metric. Numerical simulations show that the proposed receiver outperforms traditional receivers found in literature, and provides substantial performance gains that scale with complexity. The proposed receiver architecture is able to approach the performance of the optimal equalizer with significant complexity savings.