The optimality of D-BLAST-ZF with antenna order feedback without or with antenna selection

In [1], the authors proposed the GRT-VB spatial multiplexing architecture that significantly improves upon the diversity-multiplexing (D-M) tradeoff of the V-BLAST architecture by feeding back a channel dependent permutation matrix (denoting antenna order) via a few bits (log M_t!) of feedback. In this paper, it is shown that with the same antenna order feedback, if coding across the resulting parallel channels is permitted even if only via the low complexity D-BLAST architecture with a zero-forcing successive interference cancellation (ZF-SIC) receiver (hence-forth this architecture will be referred to as simply the "ordered D-BLAST-ZF"), then the optimal D-M tradeoff of the MIMO link can be achieved, thereby bridging the gap between the D-M tradeoff of the GRT-VB scheme of [1] and the optimal tradeoff curve. Furthermore, this result can be generalized to the case of the MIMO link with a limited number of R.F chains at the transmitter and the receiver that requires antenna selection (AS) at both ends. In other words, by selecting the (ordered) transmit and receive antennas using the fast algorithm in [2], it is shown that the fundamental D-M tradeoff with antenna selection obtained in [2] can be achieved in spite of restricting the complexity via the ordered D-BLAST-ZF processing. Moreover, the outage performance of the proposed ordered D-BLAST-ZF schemes with or without AS can be further enhanced via spatial power allocation using only the statistical information about the ordered channel gains in the effective parallel channel model.