A Survey on Low Latency Towards 5G: RAN, Core Network and Caching Solutions

Imtiaz Parvez, Ali Rahmati, Ismail Güvenç, Arif I. Sarwat, Huaiyu Dai

Abstract

Achieving sub-1ms end-to-end (E2E) latency is a cornerstone for the realization of the "Tactile Internet" and mission-critical 5G applications. This survey explores the architectural transformations required across the Radio Access Network (RAN), Core Network, and Edge Caching to meet these stringent requirements.

Introduction: The Latency Challenge

While 4G networks successfully delivered mobile broadband, they fall short of the requirements for next-generation services like autonomous driving, tele-surgery, and industrial IoT. These applications demand Ultra-Reliable Low-Latency Communication (uRLLC). Current 4G E2E latency typically resides in the tens of milliseconds, whereas 5G targets a sub-1ms threshold.

To bridge this gap, it is essential to understand the sources of delay throughout the network pipeline. Latency is accumulated through frame alignment, queuing, backhaul transmission, and core network processing.

End-to-end latency components in a mobile network architecture, highlighting radio, backhaul, core, and transport delays.

A Multi-Domain Solution Framework

The survey categorizes emerging low-latency technologies into three primary pillars: the Radio Access Network (RAN), the Core Network, and Caching solutions.

Taxonomy of 5G low latency communication solutions, categorizing techniques into RAN, Core Network, and Caching domains.

1. Radio Access Network (RAN) Solutions

RAN optimizations focus on the physical air interface. Key strategies include:

  • Shortened Transmission Time Intervals (TTI): Reducing the time duration of data frames to minimize waiting time.
  • Flexible Numerologies: Adjusting subcarrier spacing to suit different environments, such as wider spacing for mmWave to mitigate phase noise.
  • New Waveforms: Moving beyond standard OFDM to alternatives like Generalized Frequency Division Multiplexing (GFDM) and Universal Filtered Multi-Carrier (UFMC) to improve spectral efficiency and reduce latency.

2. Core Network Solutions

The core network is undergoing a shift toward Software Defined Networking (SDN) and Network Function Virtualization (NFV). These technologies decouple the control and data planes, enabling a "flattened" network architecture. This allows for Mobile Edge Computing (MEC), which brings processing power closer to the user, significantly reducing the round-trip time to the application server.

3. Caching Solutions

Caching aims to bypass backhaul and core network delays by storing popular content at the network edge—either at the base station or on user devices via Device-to-Device (D2D) communication. Research indicates that MEC-based caching can reduce access latency by up to 13%.

Key Insights and Empirical Results

  • Small Packet Efficiency: 5G traffic often involves small packets. The survey identifies Polar Codes as a superior alternative to 4G's Turbo codes for these short-packet scenarios.
  • Prototype Performance: Early 5G trials using short TTI and wider subcarrier spacing have successfully demonstrated air interface latencies of 1ms or less.
  • MEC Bottlenecks: While MEC is promising, early trials showed E2E latencies around 17ms, indicating that the software stack requires further optimization to reach the 1ms target.

Future Directions and Limitations

Despite significant progress, several challenges remain:

  • mmWave Complexity: Modeling high path loss and sensitivity to blockages in mmWave bands for high-speed users is difficult.
  • SDN Scalability: Centralized SDN controllers may become bottlenecks as IoT device counts explode; decentralized control is a key research area.
  • Mobility-Aware Caching: Maintaining low-latency connections for high-speed mobile users requires predictive caching algorithms.
  • Security Overhead: Balancing robust encryption with the strict 1ms latency budget remains a critical trade-off.

This summary is based on a survey of international 5G research projects, including METIS-II, MiWaveS, and 5G CHAMPION, alongside vendor field tests from Huawei, Ericsson, and Nokia.

Summary updated

BibTeX
@article{parvez2018survey,
  author = {Imtiaz Parvez and Ali Rahmati and Ismail Güvenç and Arif I. Sarwat and Huaiyu Dai},
  title = {{A Survey on Low Latency Towards 5G: RAN, Core Network and Caching Solutions}},
  journal = {IEEE Commun. Surv. Tutorials},
  year = {2018},
  doi = {10.1109/comst.2018.2841349},
}
Figure