IMR: MT: Fine-Grained Telemetry for Next-Generation Cellular Access Networks (NG-Scope)

Princeton University, Princeton NJ

Investigators

Abstract

The Next Generation Cellular Wireless Radio Access Network (RAN) is emerging as the dominant data network for Internet access, supporting capacity-enhanced mobile broadband, ultra low latency, and high connectivity machine-to-machine scenarios alike. Insightful performance measurement of such applications depends on one or more of delay, jitter, and network capacity measurement data (telemetry) at millisecond-level timescales. This necessitates a clean-slate approach that marries high resolution physical-layer telemetry with end-to-end network- and transport-layer telemetry. This proposal develops NG-Scope, a measurement tool that provides a network telemetry stream to higher software layers. NG-Scope's telemetry succinctly summarizes wireless capacity, RAN-added latency, and network management data. Rather than relying on crowd-sourced data from deployment on most or all user equipment, NG-Scope provides visibility into all users' telemetry data in a given cell. NG-Scope accounts for and leverages the latest 5G innovations, but abstracts these details behind a proposed standardized API (Application Program Interface) that measurement tools, applications, and transport layer protocols access. NG-Scope gives network measurement researchers visibility into the RAN at fidelity levels and user population scales not previously possible. Its proposed RAN-specific processing techniques bridge the gap between physical- and transport-layer telemetry data, enabling inference of the sources of network phenomena not previously observable, such as the ability to differentiate between wireless channel conditions and RAN versus core network congestion as the cause of packet latency and jitter. NG-Scope does not require the incentivization or cooperation of any service provider, enabling unfettered access to RAN telemetry and thus broadly accelerating innovation in network measurement and design. NG-Scope will benefit basic longitudinal internet measurement research, network operators' network monitoring tasks, and internet application architects' design and development efforts. It will be developed in an open-source model so that internet measurement researchers can deploy the tool locally. Results of data measurement campaigns will be made publicly available, to facilitate trace-driven emulation based experiments of other researchers and network protocol designers. Researchers will be able to remotely prototype and evaluate their own applications and network protocols in a virtual programming environment with remote access. The project will engage with under-represented groups through summer research opportunities, REU sites, and the undergraduate junior/senior thesis. Project website: ngscope.cs.princeton.edu This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.