When designing wireless communication systems with multiple transmit antennas, engineers face a fundamental choice: should the transmitter split its data into multiple spatial streams (spatial multiplexing), or should it concentrate all its power into a single directional beam (beamforming)?
Mathematically, this choice is captured by the transmit covariance matrix, $\mathbf{Q}$. If $\mathbf{Q}$ has a rank greater than one, the transmitter is distributing information across multiple spatial dimensions. If $\mathbf{Q}$ is strictly rank-one, it can be factored into a single vector ($\mathbf{Q} = \mathbf{w}\mathbf{w}^H$), meaning the optimal strategy is to focus all energy into a single spatial stream.
Intuitively, you might expect that adding complex network constraints—such as avoiding interference with neighboring users, hiding data from eavesdroppers, or transferring wireless power—would require a high-rank strategy to balance these competing goals. However, a remarkable body of research demonstrates that under a wide variety of practical constraints, rank-one beamforming is mathematically optimal.
The Core Intuition: Why Rank-One Suffices
To understand why rank-one solutions dominate, consider a Multiple-Input Single-Output (MISO) channel where a multi-antenna base station sends data to a single-antenna receiver. Because the receiver has only one antenna, it can only resolve a single spatial dimension at any given moment. Any secondary data stream sent by the transmitter along an orthogonal path cannot be separated by the receiver and instead manifests as self-interference.
This geometric reality underpins why complex space-time coding across multiple time slots often yields no capacity benefits when feedback is available. As shown by Che Lin et al. in To Code or Not to Code Across Time: Space-Time Coding with Feedback, concentrating power into the single most favorable spatial path identified via channel feedback consistently outperforms higher-rank schemes. Mathematically, this is verified by taking the derivative of the Shannon capacity formula with respect to the power allocated to secondary streams; Hongyang Chen et al. demonstrated in Optimality of beamforming condition for multiple antenna systems with mean feedback that when this derivative is zero or negative at the boundary, single-stream beamforming is strictly capacity-maximizing.
Navigating Interference and Spectrum Sharing
The optimality of rank-one beamforming persists even when the transmitter must actively protect other nodes in the network, such as in cognitive radio or multi-cell coordination. In these scenarios, a secondary transmitter must maximize its own data rate while keeping the interference caused to a primary user below a strict "interference temperature" threshold.
[ Multi-Antenna Transmitter ]
/ \
/ (Main Beam) \ (Null/Attenuation)
v v
[Intended User] [Protected Primary User]
Intuitively, one might think the transmitter should allocate one spatial stream to the intended user and a second stream to cancel out interference at the primary user. However, optimization theory reveals that a single, carefully steered beam is superior. In Cooperative Interference Management in Multi-Cell Downlink Beamforming and Cooperative interference management with MISO beamforming, Rui Zhang et al. proved that rank-one beamforming is mathematically sufficient to reach any point on the Pareto-optimal rate boundary of a MISO interference channel.
Instead of splitting power across dimensions, the optimal rank-one matrix shifts the direction of the single beam. It finds a spatial compromise that maximizes the projection onto the intended user's channel while satisfying the interference constraints. This structural property holds true even under probabilistic or outage constraints, as verified via Karush-Kuhn-Tucker (KKT) conditions by Jing Wang et al. in Transmission strategies with interference coordination in spectrum sharing cognitive radio networks under outage probability constraint.
The Geometry of Physical Layer Secrecy
Physical layer security introduces an adversarial constraint: maximize the data rate to a legitimate receiver while ensuring that an unauthorized eavesdropper intercepts nothing.
Lan Zhang et al. established an elegant mathematical bridge in On the Relationship Between the Multi-antenna Secrecy Communications and Cognitive Radio Communications, demonstrating that a secrecy constraint behaves mathematically like a virtual interference constraint. Because the legitimate user possesses only a single antenna, the optimal transmit strategy for the secure signal remains strictly rank-one, even when facing multiple distributed eavesdroppers.
Using semidefinite programming (SDP) and semidefinite relaxation (SDR), Wing-Kin Ma et al. proved in Optimal and Robust Transmit Designs for MISO Channel Secrecy by Semidefinite Programming and A robust artificial noise aided transmit design for MISO secrecy that the relaxed optimization problems yield a rank-one solution naturally. Spreading the legitimate signal across multiple spatial dimensions would degrade the signal quality at the single-antenna legitimate receiver without providing any geometric advantage in hiding the data. Instead, the system maximizes the "secrecy leakage gap" by steering a single, high-intensity beam toward the legitimate user.
Maintaining Optimality Under Uncertainty
In real-world deployments, transmitters never possess perfect Channel State Information (CSI). They must optimize for the worst-case scenario within an uncertainty region (e.g., spherical or ellipsoidal errors).
Remarkably, bounded uncertainty does not break the rank-one structure. In Robust Cognitive Beamforming With Partial Channel State Information, Lan Zhang et al. proved that under ellipsoidal uncertainty, the worst-case optimization problem still collapses to a rank-one matrix. Similarly, Jiangyuan Li et al. demonstrated in Explicit Solution of Worst-Case Secrecy Rate for MISO Wiretap Channels With Spherical Uncertainty that the optimal input covariance matrix retains a rank-one structure under spherical uncertainty, reducing a complex matrix search to a simple eigenvalue problem.
Why does this happen? Wing-Kin Ma et al. resolved this underlying mystery in Unraveling the Rank-One Solution Mystery of Robust MISO Downlink Transmit Optimization: A Verifiable Sufficient Condition via a New Duality Result by introducing a critical metric: the channel-to-uncertainty ratio. They proved that if the estimated channel strength significantly outweighs the estimation error relative to the target data rates, the optimization landscape naturally forces a rank-one solution. The uncertainty merely alters the orientation of the optimal beamforming vector rather than forcing the signal to split into multiple streams.
Balancing Information and Wireless Power
The rise of Simultaneous Wireless Information and Power Transfer (SWIPT) introduces a dual objective: a transmitter must deliver decodable data to an information receiver while beaming raw radio-frequency (RF) energy to an energy harvester.
/---> [Information Receiver] (Decodes 1 Stream)
[MISO Transmitter]
\---> [Energy Harvester] (Captures Raw RF Energy)
Because energy harvesters do not decode data—they merely rectify the incoming RF waveform—they are indifferent to the phase, modulation, or rank of the incoming signal; they care only about total received power. Consequently, allocating a separate, dedicated "energy beam" alongside an "information beam" is fundamentally inefficient.
As demonstrated by Chao Shen et al. in Simultaneous Information and Energy Transfer: A Two-User MISO Interference Channel Case, a single directed information beam (rank-one covariance) is mathematically optimal for maximizing data throughput under minimum energy harvesting thresholds. Zhao Chen et al. reinforced this in Energy-Throughput Tradeoff in Sustainable Cloud-RAN with Energy Harvesting, proving that single-stream beamforming is sufficient to reach the absolute performance frontier of joint information-and-power networks. The information beam itself doubles as the energy source, eliminating the hardware and computational complexity of multi-beam generation.
Summary of Applications
The mathematical elegance of rank-one optimality simplifies transceiver architectures across several domains:
- High-Frequency Channels: In mmWave and Terahertz systems, channels are naturally sparse and low-rank. Jingping Qiao et al. showed in Secure Transmission for Intelligent Reflecting Surface-Assisted mmWave and Terahertz Systems that rank-one channel conditions decouple the optimization of base station beamforming from the phase shifts of intelligent reflecting surfaces.
- Cooperative Relaying: When cognitive base stations act as relays to assist primary users, Gan Zheng et al. proved in Cooperative Cognitive Networks: Optimal, Distributed and Low-Complexity Algorithms that the optimal relaying matrix collapses to a rank-one structure, transforming complex matrix optimization into efficient vector matched-filtering and beamforming.
- Degraded Channels: For multi-user networks where channel qualities can be strictly ordered, Jubin Jose demonstrated in Sum capacity of K user Gaussian degraded interference channels that unit-rank matrix representations yield exact closed-form capacity solutions.
By proving that single-stream beamforming achieves the same performance as multi-stream multiplexing under these diverse constraints, this thread of research provides a solid theoretical foundation for building high-performance, low-complexity wireless hardware.
Key Foundational Work
To Code or Not to Code Across Time: Space-Time Coding with Feedback (2007) — contributed 3 concept elements to this direction.
Robust Cognitive Beamforming With Partial Channel State Information (2008) — contributed 2 concept elements to this direction.
Robust cognitive beamforming with partial channel state information (2008) — contributed 2 concept elements to this direction.