On the Performance of Cell-Free Massive MIMO With Low-Resolution ADCs

Abstract

This paper considers an uplink cell-free massive multi-input multi-output (mMIMO) system with multi-antenna access points (APs) and users, assuming low-resolution analog-to-digital converter (ADC) architecture is employed at the APs. Leveraging on the additive quantization noise model (AQNM), we derive a tight approximate expression for uplink spectral efficiency (SE). This trackable finding provides us with a tool for easily quantifying the impacts of the number of antenna arrays and the number of quantization bit of low-resolution ADCs. We find 5-bit is required in a cell-free mMIMO with low-resolution ADCs to achieve the same SE as a cell-free mMIMO with full-precision ADCs. Besides, when the number of antennas of the user is small, deploying more antennas at the users can boost the sum SE. Then, to further highlight the potential of low-resolution ADCs architecture, we also investigate the tradeoff between the SE and energy efficiency (EE) with design issues surrounding the quantization bit of low-resolution ADCs and the number of antenna arrays. The resulting observations reveal that by choosing a proper quantization bit, the cell-free mMIMO with low-resolution ADCs has the capability to enjoy a better SE-EE tradeoff compared to the perfect ADCs counterpart.INDEX TERMS Cell-free mMIMO, low-resolution ADC, AQNM, spectral efficiency, energy efficiency.

Fronthaul-Constrained Cell-Free Massive MIMO With Low Resolution ADCs

“…In particular, α θ is an increasing function of b θ and, for the non-uniform quantizer case, which will be considered in the next sections, can be approximately expressed as [35], and the corresponding values for b θ = 1, 2, 3, 4 and 5 are summarized in Table 1 (derived from [34,Table I]). This linear model for the quantization process (or its equivalent AQNM) has been extensively used in the massive MIMO and cell-free massive MIMO literature (see, among many others, [23], [25], [36]- [45] and references therein).…”

“…In [24], the authors solely focus on the downlink (DL) and assume the use of a specific CPU-AP functional split in which the control layer, in charge of delivering channel estimates to the CPU, is not subject to any capacity constraint, whereas the data layer, in charge of delivering user's precoded signals to the APs, is subject to capacity constrained links. Zhang et al in [25] only evaluate the UL and, although they assume the use of low-resolution analog-to-digital converters (ADCs), do not consider the effects that the use of capacity constrained fronthaul links may have on the spectral efficiency of the network. Masoumi et al in [26], analyze the performance provided by three different transmission strategies at the APs but, unfortunately, they do not assume the use of conventional low-resolution ADCs but the use of sophisticated compressors that, only by assuming the compression of very large signal sequences, can be modeled based on the rate-distortion theory.…”

Cell-Free Massive MIMO : Scalability, Signal Processing and Power Control

“…Uplink processing MMSE [59,124,145,146] ZF [147] Power allocation and load balancing [148] Max-min power control [149] Pilot power control [150] Sequential linear MMSE [120] mmWave Limited fronthaul capacity [151] Power control [152] Channel modeling [153] Energy efficiency [154] Energy Efficiency [116,132,[155][156][157] Imperfect Fronthaul [158][159][160][161][162][163][164] Low Resolution ADCs [163,[165][166][167] Hardware Impairments [164,[168][169][170] Non-orthogonal [171][172][173][174][175] Multiple Access…”

Sum‐rate maximization in uplink cell‐free massive multiinput multioutput system with jamming

“…Explicitly, cell‐free massive MIMO refers to the system enables a great number of distributed access points (APs) to coherently serve a lower number of users in the same resource blocks. By combining the merits of distributed massive MIMO and network MIMO systems, cell‐free massive MIMO system leads to an ability to dramatically boost the channel capacity and energy efficiency 8,9 …”