Twenty-Five Years of Signal Processing Advances for Multiantenna Communications: From theory to mainstream technology

Emil Björnson, Yonina C. Eldar, Erik G. Larsson, Angel Lozano, H. Vincent Poor

This review paper, published in the IEEE Signal Processing Magazine, provides a comprehensive retrospective on how multiantenna technology evolved from a theoretical niche in 1998 to the mainstream backbone of 5G networks today. It maps out the signal processing (SP) breakthroughs that allowed us to move from simple voice calls to the data-heavy, "always-connected" world of the 2020s.

TL;DR

The paper chronicles 25 years of multiantenna communications, tracing the shift from basic spatial diversity to the high-capacity "Massive MIMO" systems of 5G. It highlights how signal processing transformed theoretical concepts like channel hardening and TDD reciprocity into practical technologies that now support billions of subscribers.

What Problem Does This Paper Address?

In 1998, wireless communication was primarily about telephony and text. As user demand for data exploded (doubling every 18 months, per Cooper’s Law), engineers faced a fundamental bottleneck: limited radio spectrum.

The challenge was to increase "spectral efficiency"—squeezing more bits into the same frequency band. While early systems used multiple antennas merely to prevent dropped calls (diversity), the field needed to figure out how to use those same antennas to transmit multiple data streams simultaneously (spatial multiplexing) without causing self-interference. This required overcoming massive hurdles in hardware complexity, power consumption, and the acquisition of accurate "Channel State Information" (CSI).

Methods

The authors categorize the evolution into five key signal processing advancement areas:

  1. Spatial Multiplexing (MIMO): Moving from using antennas for reliability (diversity) to using them for speed. The key was the realization that multipath propagation—once considered a nuisance—could be exploited to create parallel "data pipes" in the air.
  2. Massive MIMO: Scaling base station antenna counts from 2 or 4 up to 64 or more. This relies on TDD (Time-Division Duplexing) Reciprocity, where the base station learns the channel from uplink pilots, a method that doesn't become more difficult as you add more antennas.
  3. Millimeter-Wave (mmWave) Processing: To use the massive bandwidth available at high frequencies (30 GHz+), the paper discusses Hybrid Analog-Digital Beamforming and Low-Resolution ADCs (Analog-to-Digital Converters) to keep power consumption manageable.
  4. Dimensionality Reduction: Using Compressed Sensing to exploit the "sparsity" of high-frequency channels. Since signals at high frequencies only bounce off a few objects, the channel can be described with much less data than a traditional model would suggest.
  5. Machine Learning (ML) Refinements: The shift from purely model-based algorithms to data-driven ones. A standout technique mentioned is Algorithm Unrolling (or deep unfolding), which turns classical iterative algorithms into neural network layers to improve speed and accuracy.

Datasets

Not discussed in the paper. As a review article, it focuses on theoretical frameworks, historical milestones, and algorithmic trends rather than specific experimental datasets.

Key Insights and Ideas

  • From Countering to Exploiting Multipath: The "aha moment" of MIMO was realizing that a "messy" radio environment with many reflections actually makes it easier to separate multiple data streams.
  • Channel Hardening: In Massive MIMO, as you add more antennas, the random fluctuations of the signal (fading) start to average out. The channel becomes "hardened" and behaves almost deterministically, simplifying the signal processing.
  • The TDD Advantage: Massive MIMO is only practical in TDD mode. In FDD (Frequency-Division Duplexing), the overhead of reporting channel conditions back to the base station grows with the number of antennas, making it unscalable.
  • Model-Aided ML: Rather than using "black box" AI, the field is moving toward "deep unfolding," where we take a proven math-based algorithm and use machine learning to fine-tune its parameters.

Key Results

  • Subscriber Growth: The technology supported a leap from 300 million subscribers in 1998 to over 6 billion today.
  • Capacity Gains: 5G base stations now commonly use 64-antenna arrays to serve multiple users (Multiuser MIMO) simultaneously on the same frequency, a feat that was considered computationally impossible 25 years ago.
  • Bandwidth Expansion: Signal processing enabled the move from 20 MHz channels in 2G to 100–500 MHz channels in 5G.

Limitations and Future Work

The authors acknowledge that as we scale to even more antennas and higher frequencies, several challenges remain:

  • Computational Complexity: Algorithms like regularized zero-forcing scale cubically ($MK^2$) with the number of users and antennas, threatening to outpace hardware improvements.
  • The "Boxes-in-a-Tower" Limit: We may be reaching the physical limit of how many antennas we can bolt onto a single cell tower.
  • Future Directions:
    • Cell-Free Massive MIMO: Distributing antennas across many locations rather than clustering them on one tower.
  • ISAC (Integrated Sensing and Communication): Using the same radio signals for both data transfer and radar-like sensing of the environment.
  • RIS (Reconfigurable Intelligent Surfaces): Using "smart" materials on buildings to reflect signals in specific directions, essentially making the environment part of the network.

Summary updated

Abstract

Wireless communication technology has progressed dramatically over the past 25 years, in terms of societal adoption as well as technical sophistication. In 1998, mobile phones were still in the process of becoming compact and affordable devices that could be widely utilized in both developed and developing countries. There were "only" 300 million mobile subscribers in the world [1]. Cellular networks were among the first privatized telecommunication markets, and competition turned the devices into fashion accessories with attractive designs that could be individualized. The service was circumscribed to telephony and text messaging, but it was groundbreaking in that, for the first time, telecommunication was between people rather than locations. Wireless networks have changed dramatically over the past few decades, enabling this revolution in service provisioning and making it possible to accommodate the ensuing dramatic growth in traffic. There are many contributing components, including new air interfaces for faster transmission, channel coding for enhanced reliability, improved source compression to remove redundancies, and leaner protocols to reduce overheads. Signal processing is at the core of these improvements, but nowhere has it played a bigger role than in the development of multiantenna communication. This article tells the story of how major signal processing advances have transformed the early multiantenna concepts into mainstream technology over the past 25 years. The story therefore begins somewhat arbitrarily in 1998. A broad account of the state-of-the-art signal processing techniques for wireless systems by 1998 can be found in [2], and its contrast with recent textbooks such as [3]-[5] reveals the dramatic leap forward that has taken place in the interim.
BibTeX
@article{bjornson2023twenty,
  author = {Emil Björnson and Yonina C. Eldar and Erik G. Larsson and Angel Lozano and H. Vincent Poor},
  title = {{Twenty-Five Years of Signal Processing Advances for Multiantenna Communications: From theory to mainstream technology}},
  journal = {IEEE Signal Process. Mag.},
  year = {2023},
  doi = {10.1109/msp.2023.3261505},
}
Figure