Energy-Efficiency Maximization of Hybrid Massive MIMO Precoding With Random-Resolution DACs via RF Selection

Abstract

Energy-efficiency (EE) is identified as a key 5G metric and will have a major impact on the hybrid beamforming system design. The most promising system designs include a reduced number of radio-frequency (RF) chains with digital-toanalog converters (DACs) of lower sampling resolution. However, naive reduction of beamformer components to reduce power consumption typically leads to significant loss of spectral-efficiency (SE). In this paper, we focus on the transmit beamforming (precoding) and we introduce an architecture with low-end components that maximizes the EE while minimizing the effects on SE. This is achieved by the novel design of the analog part of the precoder, where the number of the RF chains is not reduced a priori, but deactivated based on an optimization algorithm. Thus, the problem becomes a subset selection one, where only the RF chains with the optimal SE-EE performance are being activated. The selection algorithm not only determines the optimal number of RF chains to activate but also selects optimally between DACs of randomly-allocated resolution. Through simulations, we verify that the proposed architecture exhibits improved performance when compared with baseline precoding techniques which use a predefined number of RF chains with low-resolution DACs.

Energy Efficiency Maximization Precoding for Quantized Massive MIMO Systems

“…For the EE point of view, however, using the maximum transmit power may decrease the EE because it increases the SE with a logarithmic scale while it increases the power consumption with a (approximately) linear scale. For this reason, using the maximum power without adjusting a power level, the EE converges to zero eventually as shown in [24].…”

“…As shown in the prior work, Bussgang decomposition is useful to characterize a 1-bit quantizer, while its tractability is highly limited if a quantizer is more than 2 bits. For precoding design in more general bits quantizers, an additive quantization noise model (AQNM) was also used in several works [21]- [24], providing a linear approximation of a quantized signal. Employing the AQNM, [21] considered hybrid precoding architecture with low-resolution DACss in a point-to-point MIMO channel and performed extensive performance evaluation by using a conventional precoder.…”

“…Similar to [21], the AQNM was also used in [22]. In [23], [24], an algorithm that selects an active antenna set was proposed for a given precoder. Without linear modeling, the alternating direction method of multipliers (ADMM) was used to solve an inter-user interference minimization problem in [25].…”

“…For example, the EE performances of other methods degrade as the transmit power increases since the SE does not efficiently improve as the transmit power increases in a high signal-to-noise-ratio (SNR) regime. For this reason, using conventional methods, e.g., [24], the EE converges to zero eventually if the transmit power continues to increase. On the contrary, our algorithm offers the monotonically increasing EE by adjusting the actual transmit power in the algorithm.…”

Spectral-efficient hybrid precoding for multi-antenna multi-user mmWave massive MIMO systems with low complexity

“…Traditional full digital precoding reduces the interference between data streams and users and the complexity of receiver computation [9,10]. However, full digital precoding schemes require every antenna element to be driven by its own dedicated energy-intensive radio frequency (RF) chain, imposing extremely high costs and energy consumption [11]. In contrast, an antenna array of full analog precoding is connected to only one RF chain by a phase shifter [12,13].…”