[Paper Review] An Active-Passive Measurement Study of TCP Performance over LTE on High-speed Rails

Jing Wang, Yufan Zheng+11 more|arXiv (Cornell University)|Dec 12, 2018

Network Traffic and Congestion Control29 references4 citations

TL;DR

This paper presents a large-scale active-passive measurement study of TCP performance over LTE on high-speed rail (HSR) networks in China, analyzing 1732.9 GB of data across 135,719 km of travel. It identifies severe throughput degradation due to frequent handovers and high mobility, and proposes a BBR-based congestion control algorithm that improves throughput by up to 36.5% in extreme mobility scenarios.

ABSTRACT

High-speed rail (HSR) systems potentially provide a more efficient way of door-to-door transportation than airplane. However, they also pose unprecedented challenges in delivering seamless Internet service for on-board passengers. In this paper, we conduct a large-scale active-passive measurement study of TCP performance over LTE on HSR. Our measurement targets the HSR routes in China operating at above 300 km/h. We performed extensive data collection through both controlled setting and passive monitoring, obtaining 1732.9 GB data collected over 135719 km of trips. Leveraging such a unique dataset, we measure important performance metrics such as TCP goodput, latency, loss rate, as well as key characteristics of TCP flows, application breakdown, and users' behaviors. We further quantitatively study the impact of frequent cellular handover on HSR networking performance, and conduct in-depth examination of the performance of two widely deployed transport-layer protocols: TCP CUBIC and TCP BBR. Our findings reveal the performance of today's commercial HSR networks "in the wild", as well as identify several performance inefficiencies, which motivate us to design a simple yet effective congestion control algorithm based on BBR to further boost the throughput by up to 36.5%. They together highlight the need to develop dedicated protocol mechanisms that are friendly to extreme mobility.

Motivation & Objective

• To understand the real-world performance of TCP variants—CUBIC and BBR—on high-speed rail (HSR) networks operating at speeds above 300 km/h.
• To quantify the impact of frequent and often failed handovers on TCP throughput, latency, and loss rate in extreme mobility environments.
• To analyze on-board user behavior, application usage patterns, and TCP flow characteristics in HSR settings.
• To identify protocol design inefficiencies in existing transport protocols under high-mobility conditions and motivate the design of mobility-aware congestion control.

Proposed method

• Conducted large-scale active and passive measurements across three high-speed rail routes in China, collecting 1732.9 GB of data over 135,719 km of travel.
• Performed controlled experiments and passive monitoring at the LTE gateway, capturing L1/L2 signaling, PHY rates, handover events, and TCP-level metrics.
• Correlated lower-layer LTE events (e.g., handover, SNR, MCS) with upper-layer TCP behavior (goodput, RTT, loss rate, window size) to isolate root causes.
• Analyzed application-level traffic from on-board WiFi, classifying flows by protocol, size, duration, and content type to understand user behavior.
• Designed and evaluated a modified BBR congestion control algorithm that accounts for RTprop under high mobility, improving throughput by up to 36.5%.
• Developed a taxonomy of handover types and quantified their performance impact based on success rate, duration, and timing relative to TCP RTT.

Experimental results

Research questions

• RQ1 How does extreme mobility (above 300 km/h) affect TCP performance metrics such as goodput, latency, and loss rate on HSR networks?
• RQ2 To what extent do frequent and unreliable handovers degrade TCP throughput, and how do they interact with congestion control protocols like CUBIC and BBR?
• RQ3 What are the dominant application usage patterns and TCP flow characteristics among on-board passengers on high-speed trains?
• RQ4 Why do existing transport protocols like CUBIC and BBR underperform in high-mobility environments, and what protocol design flaws contribute to this?
• RQ5 Can a modified BBR-based congestion control algorithm effectively mitigate performance degradation caused by high-speed rail mobility?

Key findings

• At 350 km/h, CUBIC and BBR throughput decreased by 47.5% and 40.1%, respectively, compared to 300 km/h, primarily due to frequent handovers and reduced PHY rates.
• Despite performance degradation, BBR maintained lower RTT, loss rate, and bytes-in-flight than CUBIC, preserving its inherent efficiency advantages.
• Handover frequency reached 13.7 and 8.6 seconds on average at 300 km/h and 350 km/h, respectively, with a significant increase in failed handovers at higher speeds.
• Successful handovers were typically too short (tens of ms) to allow TCP to react within its normal RTT, leading to unnecessary window reductions.
• Over 94% of on-board traffic was HTTP(S), with 95% of flows being small (under 100 KB), slow (under 1 Mbps), and unlikely to complete if larger than 1 MB.
• A modified BBR-based congestion control algorithm improved throughput by up to 36.5% by better estimating RTprop and adapting to high-mobility conditions.