Abstract
In this paper, we design coordinated beamforming at base stations (BSs) to facilitate interference cancelation at users in interference networks, where each BS is equipped with multiple antennas and each user is with a single antenna. By assuming that each user can select the best decoding strategy to mitigate the interference, either canceling the interference after decoding when it is strong or treating it as noise when it is weak, we optimize the beamforming vectors that maximize the sum rate for the networks under different interference scenarios and find the solutions of beamforming with closed-form expressions. The inherent design principles are then analyzed, and the performance gain over passive interference cancelation is demonstrated through simulations in heterogeneous cellular networks.
Introduction
In the push toward 5G and ultra-dense heterogeneous networks (HetNets), interference between base stations (BSs) has become the primary bottleneck for performance. Traditional solutions like Coordinated Multi-Point Joint Processing (CoMP-JP) require massive backhaul capacity to share user data, making them difficult to implement.
A more practical alternative is Coordinated Beamforming (CB), which only requires sharing channel state information (CSI). However, most existing CB techniques focus on "Zero-Forcing"—trying to make the interference zero at the receiver. This is often a waste of spatial resources. If interference is already very weak, it is better to just ignore it (Treat Interference as Noise, or TIN). If it is very strong, the receiver can decode the interference first, subtract it, and then decode their own signal (Interference Cancelation, or IC). The challenge lies in designing beamforming vectors that optimally "nudge" the interference into one of these decodable regimes, especially in "mixed" scenarios where one user sees strong interference and another sees weak.
System Model and Simulation Setup
The authors focus on a Multiple-Input Single-Output (MISO) interference channel. They evaluate their approach using Monte Carlo simulations based on standardized 3GPP channel models for path loss and Rayleigh fading. The simulation environment typically involves a two-cell setup (one macro-BS, one pico-BS) or an extended multi-pico-cell environment as illustrated in the system model.
The network is categorized into four scenarios based on the optimal decoding strategy:
- Weak Interference: Both users treat interference as noise.
- Mixed 1 & 2: One user cancels interference while the other treats it as noise.
- Strong Interference: Both users cancel interference.
Methodology
To solve the non-convex sum-rate maximization problem, the authors use high-SINR (Signal-to-Interference-plus-Noise Ratio) approximations to simplify the objective functions. The core contribution is the derivation of closed-form expressions for the beamforming vectors in each scenario. This is a significant step up from previous iterative or brute-force searching methods, as it allows for real-time implementation with low computational overhead. They also provide a geometric interpretation, showing how optimal vectors align with the altitudes of triangles formed by channel vectors.
Key Insights
- Proactive vs. Passive: Traditional IC is "passive"—it only works if the interference happens to be strong. "Proactive" IC (PIC) involves the transmitter intentionally shaping the beam to ensure the interference is strong enough to be decoded at the unintended receiver.
- The Power of "Mixed" Scenarios: In mixed interference, the BS of the "strong" user might actually strengthen its interference to the "weak" user. While counter-intuitive, this ensures the weak user can successfully cancel that interference, leading to a higher overall system rate.
- Geometric Elegance: The authors show that the optimal beamforming for proactive cancelation often lies in a two-dimensional subspace spanned by the direct and cross-link channels, effectively balancing signal strength and decodability.
Results and Performance
- Near-Optimal Performance: The proposed PIC scheme performs almost identically to the Pareto-optimal beamforming (the theoretical upper bound), which requires much more complex computation.
- Superiority over Zero-Forcing: PIC significantly outperforms Zero-Forcing (ZF-CB) and Maximal-SLNR schemes, especially as the number of pico-cells increases.
- Cell-Edge Gains: In a 20-pico-cell scenario, the PIC scheme achieved a cell-edge per-user rate nearly six times larger than passive interference cancelation (MF-IC). This suggests PIC is highly effective at improving the "ubiquitous" user experience in dense areas.
Limitations and Future Work
While PIC avoids sharing user data, it still requires accurate and timely CSI exchange between coordinated base stations, which can be challenging in high-mobility environments. Furthermore, the closed-form solutions rely on high-SINR/ISNR approximations. While these hold true for many modern cellular contexts, the performance might degrade in extremely low-power or highly shadowed regions. Future work could extend these proactive principles to MIMO (multi-antenna users), which would introduce significantly more degrees of freedom and mathematical complexity.
Summary updated