EXIT-Chart Aided Hybrid Multiuser Detector Design for Frequency-Domain-Spread Chip-Interleaved MC-CDMA

With the advent of extrinsic information transfer (EXIT) charts, we are capable of analyzing, predicting and visually comparing the convergence behaviours of different turbo multi-user detectector (MUD)s. The different MUDs have diverse EXIT characteristics and hence their superposition allows us to create a combined EXIT curve, which closely matches that of the channel decoder. Hence a near-capacity operation is facilitated by combining the benefits of different MUDs and therefore to create a superior MUD. Thus in this contribution, we propose a novel hybrid MUD combining scheme, which combines the advantages of a high performance and low complexity in form of an advanced hybrid MUD solution. The transmitted bits are unknown at the receiver, hence it is not feasible to directly evaluate the mutual information gain of the iterative MUD in consecutive iterations, hence we propose a realistic algorithm for estimating this mutual information gain, which is then used for activating the most appropriate constituent MUD as and when it is necessary. The constituent MUDs are the matched filter (MF) based soft interference cancellation (SoIC) and the optimum Bayesian MUDs, which are invoked in the scenario of frequency-domain-spread chip- interleaved (FDSCI) multiple carrier code division multiple access (MC-CDMA). The resultant hybrid MUD is capable of outperforming both the MF-SoIC and Bayesian turbo MUDs in the terms of the attainable complexity and bit-error-rate (BER) performance.