MIMO Networks With One-Bit ADCs: Receiver Design and Communication Strategies

Abbas Khalili, Farhad Shirani, Elza Erkip, Yonina C. Eldar

TL;DR

This paper tackles the power consumption challenge in modern wireless systems (like 5G mmWave and massive MIMO) by proposing new receiver designs that use low-resolution, one-bit Analog-to-Digital Converters (ADCs). They introduce "hybrid blockwise" and "adaptive threshold" receivers that leverage analog processing and temporal delays to achieve near-optimal data rates, even with limited ADCs, outperforming previous hybrid designs.

What Problem Does This Paper Address?

Modern wireless communication systems, especially those aiming for high data rates like 5G millimeter-wave (mmWave) and massive MIMO, rely on large antenna arrays. At the receiver, each antenna typically connects to a high-resolution ADC to digitize the incoming signal. However, these high-resolution ADCs are power-hungry, and their power consumption scales with the number of antennas and bandwidth. This makes them a significant bottleneck for energy efficiency in these advanced systems. While reducing ADC resolution (e.g., to one-bit) drastically cuts power consumption, it also leads to a substantial loss in achievable data rates. The challenge is to design receivers that can achieve high data rates while using a limited number of low-resolution ADCs, overcoming the inherent performance degradation.

Methods

The authors propose two novel receiver architectures: the hybrid blockwise receiver and the adaptive threshold receiver.

  • Hybrid Blockwise Receiver: This receiver extends the concept of "hybrid one-shot" receivers by incorporating delay elements in the analog domain. Instead of processing each incoming signal sample independently, it processes a block of l consecutive channel outputs jointly over l channel uses. This is achieved using a delay network and l pairs of analog combiners and threshold vectors. This temporal processing in analog allows for better utilization of the ADCs.
  • Adaptive Threshold Receiver: This receiver takes a different approach to improve performance. It uses adaptive thresholds for its one-bit ADCs, where the thresholds are updated based on previous outputs. This allows the receiver to extract more information from the signal over time. The paper shows that this can achieve optimal high SNR rates irrespective of the channel rank, unlike the hybrid blockwise receiver which has rank-dependent optimality.

Both proposed receivers leverage analog processing before quantization to mitigate the rate loss associated with low-resolution ADCs.

Datasets

Not discussed in the paper. The paper focuses on theoretical analysis and simulation results based on established communication models.

Key Insights and Ideas

  • Temporal Processing in Analog is Key: The core insight is that by introducing temporal processing (using delay elements) in the analog domain before quantization, the performance limitations imposed by one-bit ADCs can be significantly overcome. This allows for a more efficient use of the limited ADC resources.
  • Adaptive Thresholds Unlock Full Potential: The adaptive threshold receiver demonstrates that by dynamically adjusting ADC thresholds based on past observations, the receiver can achieve optimal high SNR rates regardless of the channel's rank. This is a significant advantage over designs whose performance is tied to channel conditions.
  • Trade-off Management: The paper highlights the fundamental trade-off between power consumption (driven by ADC resolution and count) and achievable data rates. The proposed receivers offer a way to navigate this trade-off more effectively.
  • Blockwise Processing for Rate Gain: Processing blocks of data in the analog domain (hybrid blockwise) allows for a higher number of effective "decision regions" at high SNR, directly translating to higher achievable rates compared to processing samples individually.

Key Results

  • High SNR Optimality: Both proposed receivers achieve the maximum achievable rates (or rate regions for multi-user scenarios) among all receivers with the same number of one-bit ADCs, especially in the high SNR regime.
  • Hybrid Blockwise: Achieves optimal high SNR capacity for single-user MIMO when the number of ADCs ($n_q$) is less than or equal to twice the rank of the channel ($n_q \le 2 \text{rank}(H)$) for large delay lengths ($l$).
  • Adaptive Threshold: Achieves optimal high SNR capacity for single-user MIMO ($n_q$ bits per channel-use) irrespective of the channel rank. It also achieves optimal sum-rates in MIMO MAC and per-user rates in MIMO BC, again, irrespective of channel ranks.
  • Rate Improvement: The proposed receivers demonstrate significant improvements in achievable rates compared to conventional hybrid one-shot receivers, particularly when the number of ADCs is limited relative to the channel rank. For instance, in a two-user BC example, the adaptive threshold receiver achieved a higher rate than hybrid one-shot with equal time sharing.

Limitations and Future Work

  • Complexity: The adaptive threshold receiver, while achieving optimality, is described as more complex due to its adaptive nature. The paper mentions that finding optimal power and ADC allocation for these schemes can be an NP-hard problem.
  • Practical Implementation: The paper notes that practical constraints like limited phase shifter resolution could impact the performance of SVD-based approaches.
  • Future Work: The authors suggest exploring the limitation of low-resolution DACs and optimizing the overall transmitter and receiver architecture. They also propose finding closed-form expressions for the capacity of hybrid one-shot and blockwise receivers for all SNR regimes. Investigating low-complexity solutions for optimal power and ADC allocation is also mentioned.

Summary updated

Abstract

High resolution analog to digital converters (ADCs) are conventionally used at the receiver terminals to store an accurate digital representation of the received signal, thereby allowing for reliable decoding of transmitted messages. However, in a wide range of applications, such as communication over millimeter wave and massive multiple-input multiple-output (MIMO) systems, the use of high resolution ADCs is not feasible due to power budget limitations. In the conventional fully digital receiver design, where each receiver antenna is connected to a distinct ADC, reducing the ADC resolution leads to performance loss in terms of achievable rates. One proposed method to mitigate the rate-loss is to use analog linear combiners leading to design of hybrid receivers. Here, the hybrid framework is augmented by the addition of delay elements to allow for temporal analog processing. Two new classes of receivers consisting of delay elements, analog linear combiners, and one-bit ADCs are proposed. The fundamental limits of communication in single and multi-user (uplink and downlink) MIMO systems employing the proposed receivers are investigated. In the high signal to noise ratio regime, it is shown that the proposed receivers achieve the maximum achievable rates among all receivers with the same number of one-bit ADCs.
BibTeX
@article{khalili2022mimo,
  author = {Abbas Khalili and Farhad Shirani and Elza Erkip and Yonina C. Eldar},
  title = {{MIMO Networks With One-Bit ADCs: Receiver Design and Communication Strategies}},
  journal = {IEEE Trans. Communications},
  year = {2022},
  url = {https://openreview.net/forum?id=UGtW0Fiw-PD},
}
Figure