Title: Unveiling the 5G Mid-Band Landscape: From Network Deployment to Performance and Application QoE
Speaker : Eman Ramadan (University of Minnesota – Twin Cities)
Scribe : Yuntao Zhao (Xiamen University)
Authors: Rostand A. K. Fezeu (University of Minnesota - Twin Cities, USA); Claudio Fiandrino (IMDEA Networks Institute, Madrid, Spain); Eman Ramadan, Jason Carpenter (University of Minnesota - Twin Cities, USA); Lilian Coelho de Freitas (Federal University of Pará, Brazil); Faaiq Bilal, Wei Ye (University of Minnesota - Twin Cities, USA); Joerg Widmer (IMDEA Networks Institute, Madrid, Spain); Feng Qian (University of Southern California, USA); Zhi-Li Zhang (University of Minnesota - Twin Cities, USA)
Introduction
The paper conducts an in-depth study on the deployment and performance of commercial mid-band 5G networks across Europe and the U.S. It provides a detailed analysis of key channel configurations and parameters that influence the network’s physical layer performance, as well as the application-level Quality of Experience (QoE). The study reveals significant variability in 5G configurations and identifies the impact of channel bandwidth, MCS, and MIMO layers on user-perceived performance. Furthermore, it examines the temporal dynamics of 5G networks and their implications for application performance, particularly focusing on video streaming. The paper contributes to the optimization of 5G deployments and offers insights for improving application QoE over mid-band 5G networks.
Key idea and contribution:
This paper presents a comprehensive analysis of over 5600 minutes of 5G network data, amassed from seven major operators across Europe and the US, totaling more than 5 terabytes. The study meticulously compares 5G mid-band performance, focusing on network configurations, physical layer parameters, and deployment settings, and uncovers several key findings:
-
Divergent 5G Mid-band Deployments: The paper reveals a stark contrast in 5G mid-band settings and configurations between Europe and the U.S., as illustrated in the figure below. In Europe, a uniform deployment approach is evident, with all operators utilizing the n78 band and exhibiting more consistency in channel bandwidths, typically ranging from 80 MHz to 100 MHz. Conversely, in the U.S., a more diverse deployment strategy is observed, with operators such as T-Mobile employing carrier aggregation to overcome the limitations of fragmented mid-band spectrum, resulting in an aggregate channel bandwidth exceeding 100 MHz.
-
The Impact of Channel Dynamics: Channel bandwidth does not always yield higher throughput. PHY layer parameters like MCS and MlMO layers can play a bigger role.
-
Stability and Coverage Over Performance: The paper emphasizes that network stability and coverage are more critical for ensuring consistent user experience than merely boosting peak performance. The findings suggest that factors such as deployment coverage, tower density, and overall network infrastructure significantly affect performance, suggesting that operators should prioritize these aspects in their 5G rollout strategies.
-
Adaptation to Channel Dynamics for Enhanced QoE: The paper underscores the importance of mechanisms that can adapt to channel dynamics to improve application Quality of Experience (QoE). By examining the impact of 5G PHY parameters on video streaming, the study proposes that making applications “5G-aware” can lead to significant enhancements in streaming quality. This includes adopting smaller video chunk sizes to better accommodate channel variability, thereby improving average bitrates and reducing stall times.
The paper’s contributions lie in its extensive empirical analysis, its comparative study of 5G deployments, and its insights into the factors that genuinely affect user experience. By providing a detailed characterization of 5G mid-band performance and offering practical recommendations for network operators and application developers, the study contributes to the optimization of 5G networks and the enhancement of user experience in the burgeoning 5G ecosystem.
Q1: In your presentation, you mentioned channel variability. Did you conduct similar measurements to compare the mid-band and low-band of 5G? Because in the slides, I only saw you compare the mid-band with the high-band.
A1: The low band is not actually very interesting. We are interested in the millimeter wave and the mid-band of 5G because the millimeter wave offers the high bandwidth that everyone is looking for. However, as we actually observe, it is not stable at all. This is why we are trying to figure out why the mid-band of 5G is actually the ‘sweet spot’ compared to millimeter waves. Although millimeter waves have high throughput, due to their significant variability, in a sense, they are actually not very useful.
Q2: Is there an optimal density or ‘sweet spot’ for base station deployment that ensures maximum throughput? In other words, is there a point beyond which increasing the density of base stations no longer improves throughput, making further densification unnecessary?
A2: I believe that, generally speaking, an increased number of base stations leads to better performance. However, a key point we wanted to emphasize for operators is the importance of strategic placement of their base stations rather than a random allocation. By delving deeper into our results, operators can make informed decisions about the locations for deploying their base stations. It’s also important to consider the costs associated with densification, as it can be quite expensive. So, it depends on the operator’s strategy, but certainly, a denser deployment of base stations tends to yield better performance. We did not find a specific ‘sweet spot’ in terms of density that guarantees maximum throughput across all scenarios, as the optimal configuration can vary based on a variety of factors including geography, user density, and existing infrastructure.
Personal thoughts
This paper makes a contribution to the understanding of 5G mid-band deployments, providing a thorough comparative analysis between European and U.S. operators. I commend the authors for their extensive data collection and insightful findings, which offer a granular view of the 5G landscape. The paper’s detailed examination of network configurations, physical layer parameters, and deployment settings is particularly impressive, revealing the complexity and variability of 5G performance.
One of the aspects I appreciate most is the paper’s ability to demystify some of the performance nuances in 5G networks. The revelation that channel bandwidth isn’t always the key determinant of higher throughput and that MCS and MIMO layers play a more critical role is an important takeaway. Additionally, the paper’s focus on the impact of channel dynamics on application QoE is both timely and relevant, given the growing importance of video streaming and other bandwidth-intensive applications.
However, I find that the paper leaves some open questions unexplored. While it presents many interesting facts and insights, it does not delve into how these insights could be leveraged to inform actionable strategies. For instance, the paper could have discussed how application developers and network operators might use these findings to design schemes that optimize application performance and QoE. Specifically, it would be valuable to see recommendations or a framework for adapting application algorithms to better handle the observed channel variability and improve user experience under different network conditions.
Furthermore, the paper’s focus on the technical aspects of network performance leaves room for further exploration into user behavior and perceptions. How do users perceive the variability in performance, and how might this affect their overall satisfaction and choice of service provider? Understanding the user perspective, in conjunction with the technical data, could provide a more holistic view of the 5G experience.
In conclusion, the paper provides a solid foundation for further research and practical application in the field of 5G mid-band deployments. Its contributions to the understanding of network performance are clear, and the open questions it raises provide fertile ground for future work. Exploring how the insights from this paper can be applied to enhance application design and user experience will be an important next step in the ongoing evolution of 5G technology.