Predictor Antenna: A Technique to Boost the Performance of Moving Relays

Hao Guo, Behrooz Makki, Dinh-Thuy Phan-Huy, Erik Dahlman, Mohamed-Slim Alouini, Tommy Svensson

TL;DR

The "Predictor Antenna" (PA) system solves the "channel aging" problem in high-speed vehicles by using a lead antenna to sample the radio environment that trailing antennas will encounter moments later. This "look-ahead" mechanism can boost backhaul throughput by up to 50% and supports vehicle speeds five times higher than traditional Kalman-filter-based prediction methods.

What Problem Does This Paper Address?

In modern wireless networks, users in high-speed vehicles (trains, buses, cars) often suffer from poor connectivity. To solve this, researchers use Moving Relays (MRs)—access points mounted on vehicle roofs that talk to the base station outside and provide a stable Wi-Fi/5G bubble inside.

However, at high speeds, these relays face a major hurdle: Channel Aging. By the time a base station estimates the "shape" of the radio channel and prepares a data packet, the vehicle has already moved several meters. This makes the Channel State Information (CSI) at the transmitter obsolete, leading to dropped packets and reduced speeds. While 4G/5G systems work well for pedestrians, their performance plummets at speeds above 120 km/h because they cannot predict the fast-changing radio environment accurately enough.

Methods

The authors propose and analyze the Predictor Antenna (PA) setup. The core idea is simple but elegant:

  1. The Hardware: Two sets of antennas are installed on the vehicle roof. The Predictor Antenna (PA) is placed at the front, and the Receive Antennas (RAs) are aligned behind it.
  2. The "Look-Ahead": As the vehicle moves, the PA samples the channel at a specific spatial point. Because the RAs are following the same path, they will reach that exact same spatial point a few milliseconds later.
  3. The Processing: The system uses the PA’s measurements to provide the base station with "future" CSI for the RAs.

The paper evaluates two main configurations:

  • Adaptive-delay setup: Dynamically adjusts the transmission timing based on vehicle speed so the data arrives exactly when the RA reaches the PA's previous spot.
  • Nonadaptive-delay setup: Uses spatial correlation and interpolation to estimate the channel when a perfect spatial match isn't possible (e.g., due to fixed processing delays).

The study covers both Urban scenarios (terrestrial base stations) and Rural scenarios (satellite or High Altitude Platform backhaul).

Datasets

This paper primarily uses sophisticated numerical simulations and references specific real-world field trials rather than a single static dataset.

  • Simulations: Based on the Jakes’ correlation model for spatially-correlated Rayleigh fading (urban) and shadowed Rice models (rural/satellite) to simulate tree foliage and blockage.
  • Field Trial References: The authors incorporate data and observations from previous trials in Dresden and Stuttgart, Germany, which validated the PA concept at speeds between 25–50 km/h and frequencies around 2.5–2.68 GHz.

Key Insights and Ideas

  • Beating the Kalman Limit: Traditional mathematical predictors (like Kalman filters) only work for very short horizons (0.1–0.3 wavelengths). The PA system extends this horizon significantly, allowing for accurate CSI even with the 5–10ms delays typical of modern networks.
  • The "Spatial Mismatch" Trade-off: The biggest challenge isn't the radio noise, but the physical alignment. If the vehicle speed doesn't perfectly match the processing delay, the RA won't be exactly where the PA was. However, the authors show that Massive MIMO beamforming can actually help "blur" this mismatch, making the system more robust.
  • Modular Superiority: No single prediction method wins at every speed. The authors suggest a "modular" approach where the system switches between PA-based prediction at high speeds and traditional diversity schemes at very high speeds where spatial correlation breaks down.
  • Satellite Synergy: In rural areas, where a train might be blocked by trees or buildings, having multiple antennas (PAs and RAs) across different wagons allows the system to "pick" the antenna with the clearest line-of-sight to the satellite.

Key Results

  • Throughput Gains: A single-antenna PA setup can increase backhaul throughput by up to 50% compared to standard open-loop (non-CSI) schemes.
  • Speed Support: At a carrier frequency of 2.68 GHz and a 5ms delay, the PA system supports speeds up to 120 km/h, whereas traditional Kalman predictors fail past 24 km/h.
  • Reliability: In the "Rural Train" simulation, the PA system maintained nearly constant Quality of Service (QoS) across a wide range of speeds, effectively neutralizing the negative effects of channel aging.
  • Beamforming Accuracy: When combined with Massive MIMO (32+ antennas at the base station), the PA allows for highly accurate "Maximum Ratio Transmission" (MRT), which significantly boosts the received power compared to systems without PA.

Limitations and Future Work

  • Standardization (IAB): The current 3GPP standards for Integrated Access and Backhaul (IAB) assume relatively stationary relay nodes. Integrating PA requires updates to how mobile terminals and distributed units (DUs) handle mobility.
  • Hardware Overhead: The system requires specific physical spacing between antennas, which might be difficult to implement on smaller vehicles compared to long trains or buses.
  • mmWave Challenges: While the paper touches on sub-6 GHz frequencies, the authors note that moving to millimeter wave (mmWave) will make beamforming even more sensitive to spatial mismatch, requiring even more precise PA timing.
  • Complexity: Implementing "Adaptive-delay" requires real-time, high-precision speed tracking and tight synchronization between the vehicle and the base station.

Summary updated

Abstract

In future wireless systems, a large number of users may access the networks via moving relays (MRs) installed on top of vehicles. One of the main challenges of MRs is rapid channel variation, which may make channel estimation and its following procedures difficult. To address these issues, various schemes are designed, among which the predictor antenna (PA) is a candidate. The PA setup refers to a system with two (sets of) antennas on top of a vehicle, where the PA(s) positioned in front of the vehicle is (are) utilized to predict the channel state information required for data transmission to the receive antennas aligned behind the PA. In this article, we introduce the concept and the potential of PA systems. Moreover, summarizing the field trials for PAs and the 3GPP attempts on (moving) relays, we compare the performance of different PA and non-PA methods for vehicle communications in both urban and rural areas with the PA setup backhauled through terrestrial or satellite technology, respectively. As we show, with typical parameter settings and vehicle speeds, a single-antenna PA-assisted setup can boost the backhaul throughput of MRs, compared to state-of-the-art open-loop schemes, by up to 50 percent.
BibTeX
@article{guo2021predictor,
  author = {Hao Guo and Behrooz Makki and Dinh-Thuy Phan-Huy and Erik Dahlman and Mohamed-Slim Alouini and Tommy Svensson},
  title = {{Predictor Antenna: A Technique to Boost the Performance of Moving Relays}},
  journal = {IEEE Communications Magazine},
  year = {2021},
  url = {https://openreview.net/forum?id=9NOUc2NtRB},
}

Loading related content...

Figure