Abstract:
An optimal ADC Bit-Allocation (BA) Algorithm based on the maximization of the cost \mathrm{K}_{f} was derived in [1], [2]. The optimal BA ensures Minimum Mean Squared E...Show MoreMetadata
Abstract:
An optimal ADC Bit-Allocation (BA) Algorithm based on the maximization of the cost \mathrm{K}_{f} was derived in [1], [2]. The optimal BA ensures Minimum Mean Squared Error performance of the massive Multiple-Input Multiple-Output receiver. However, this Algorithm has an additive complexity of of O(N_{b}^{N_{s}}), where N_{b} is the ADC bit range and N_{s} the number of RF paths. In this paper, we propose a modified dynamic programming algorithm that significantly reduces the additive complexity to O(N_{b}^{2}N_{s}N_{e}^{\prime}). Typically, dynamic programming is used to solve optimization problems with linear constraints, given the cost function satisfies the principle of optimality. Here, we modify the cost \mathrm{K}_{f} as a multi-valued function and show that it satisfies the principle of optimality under a power constraint, which is non-linear. This results in maintaining multiple winning paths (survivors) at each node in the trellis to arrive at the optimal BA solution. We also derive the minimum number of survivors N_{e}^{\prime} required to do so. Using Monte Carlo simulations, we compare the MSE performance of the proposed algorithm with that of the algorithm in [1].
Published in: 2019 IEEE International Conference on Advanced Networks and Telecommunications Systems (ANTS)
Date of Conference: 16-19 December 2019
Date Added to IEEE Xplore: 16 June 2020
ISBN Information: