Adaptive modulation for maximizing practicable sum capacity in MU-MISO downlink

This paper addresses adaptive modulation and power allocation for maximizing the practicable sum capacity (sum of the uncoded throughputs of the users) of a multiuser multiple-input single-output MU-MISO) system. Since the optimal solution, dirty paper encoding (DPC), is complicated to implement, we use the simpler linear precoding based on signal- to-leakage-plus-noise ratio (SLNR). For different choices of constellation sizes, the maximum practicable sum capacities and their corresponding optimal power allocations are obtained via individual non-convex optimization. The best constellation sizes for each user are then identified through a search for the highest maximized practicable sum capacity. Simulations demonstrate the significant performance improvement from this approach compared to existing power allocation schemes. Finally, a selection procedure between different constellation sets is presented to obtain the highest practicable sum capacity while maintaining the instantaneous BER of each user below a target value. This approach allows management of the trade-off between the capacity and error performances.