[Paper Review] 5G Coverage, Prediction, and Trial Measurements

Tristan Curry, Hassan Abbas|arXiv (Cornell University)|Mar 21, 2020

Advanced MIMO Systems Optimization2 references4 citations

TL;DR

This paper presents a 5G NR link budget and coverage prediction for the 3.5GHz n78 band in Sydney, using Atoll RF planning with the CrossWave propagation model and validating predictions against live trial measurements. It confirms that 200 Mbps throughput is achievable at an NRSRP of -91 dBm, with prediction accuracy validated via drive test data and Lee's method for slow fading analysis.

ABSTRACT

When planning a 5G network in the sub-6GHz bands, similar cell planning techniques to LTE can be applied. Looking at the Australian environment, the n78 band (3.3-3.8GHz TDD) is approximately 1GHz higher than the 2.6GHz band used in existing LTE networks. As a result, the coverage footprint can be similar, and therefore co-locating 5G NR (New Radio) on existing LTE base stations is a common strategy for initial network rollout. Any difference in coverage can be compensated by beamforming gain, less downtilting, or increasing the gNodeB's transmit power. This paper presents an initial link budget for data services, provides a coverage prediction, and measurements for a 5G NR NSA (Non Stand Alone) trial radiating at 3.5GHz with 60 MHz bandwidth. The coverage prediction is generated using RF planning tool Atoll, which is then compared to coverage measurements from the trial. These findings can be used to help plan a future 5G network in the Sydney metro area or similar environment.

Motivation & Objective

• To evaluate the feasibility of using existing LTE cell planning techniques for initial 5G NR deployment in the sub-6GHz bands, particularly the n78 band (3.3–3.8 GHz) in Australia.
• To establish a 5G NR link budget for data services in the 3.5GHz band with 60 MHz bandwidth, accounting for path loss, beamforming, and transmit power.
• To generate a coverage prediction using the Atoll RF planning tool with the CrossWave propagation model and high-resolution terrain and building data.
• To validate the prediction accuracy by comparing it with real-world NRSRP and throughput measurements from a live 5G NR NSA trial site.
• To assess the impact of beam selection, fast fading, and slow fading on signal quality and throughput in urban environments.

Proposed method

• Constructed a 5G NR link budget using 3GPP-recommended parameters, including 60 MHz bandwidth at 3.5 GHz, 64-antenna gNodeB, and 3D vector building data.
• Applied the CrossWave propagation model in Atoll with 2m-resolution digital terrain maps and clutter data to simulate path loss and coverage.
• Used Lee’s method (2L = 40λ, N = 36, d = 0.8λ) to compute local signal envelope averages, isolating slow fading from fast fading in drive test data.
• Conducted drive tests over two days, measuring NRSRP, NRSRQ, SINR, and throughput using a scanner and speedtest tools.
• Correlated beam-specific NRSRQ and SINR values with beam direction and path characteristics to assess beamforming performance.
• Applied the theoretical SINR–NRSRQ relationship: $\text{SINR} = \frac{1}{\frac{1}{n \times \text{NRSRQ}} - x}$, to interpret signal quality variations.

Experimental results

Research questions

• RQ1 Can traditional LTE cell planning techniques be effectively applied to initial 5G NR deployments in the n78 band (3.5 GHz)?
• RQ2 What is the achievable data rate at a given NRSRP level (e.g., -91 dBm) in a real-world 5G NR NSA deployment?
• RQ3 How accurate is the Atoll-based coverage prediction using the CrossWave model compared to real-world NRSRP measurements?
• RQ4 To what extent do beamforming and multipath propagation affect NRSRQ and SINR in urban environments?
• RQ5 How do fast fading and slow fading components influence the reliability of NRSRP measurements and prediction models?

Key findings

• A 5G NR link budget predicts 200 Mbps throughput at a UE NRSRP of -90.62 dBm, aligning closely with measured performance.
• Measured throughput averaged 200 Mbps at an NRSRP of -91 dBm ± 1 dB, confirming the link budget’s predictive accuracy.
• The coverage prediction generated by Atoll with the CrossWave model showed strong alignment with measured NRSRP after applying Lee’s method to remove fast fading.
• Beam index 6, directed at the test location, achieved the highest NRSRQ and SINR, while beam index 2 showed high values due to a dominant reflected path.
• NRSRQ exhibited less fluctuation than SINR due to the non-linear relationship between them, particularly in the range of interest.
• Latency showed high variance (mean 58.5 ms, median 19.5 ms), with outliers up to 300 ms likely caused by IP routing issues and packet loss.