Optimized cell ordering for multiuser macrodiversity detection with the conditional metric merge algorithm

Abstract

We consider maximum likelihood (ML) multiuser detection (MUD) in microdiversity. Unlike microdiversity, where diversity antennas are collocated, microdiversity employs widely spaced antennas. The sets of users seen by different antennas are in general different, but may be overlapping. From a computational perspective, the microdiversity ML-MUD problem is poorly structured, and risks becoming exponentially, complex in the total number of users. The conditional metric merge (CMM), a recently developed algorithm, dramatically reduces the computation load by exploiting the partial overlaps of user sets, without sacrificing the ML optimality of decisions. However, the CMM computation load still depends on the order of processing the antennas. This paper therefore presents a meta-algorithm to determine a sequence of processing antennas in CMM that has the lowest, or almost lowest, computation load. Even in configurations of a few cells, the sequence in which we process the antennas has a great impact on the required calculation, and the improvement gained by the algorithm is important. As with the original CMM algorithm, the ordering algorithm is applicable to both wideband and narrowband systems.