High-Bandwidth Tactical-Network Data Analysis in a High-Performance-Computing (HPC) Environment: Voice Call Analysis.

摘要

Analyzing network traffic for Voice over Internet Protocol (VoIP) calls for large test events, complex correlation, and aggregation of packet flows--between multiple nodes in a large network (typically 50 or more instrumented nodes), across multiple application flow protocols (Transmission Control Protocol call signaling and Real-time Transport Protocol audio data), and over significant lengths of time (multiple hours)--while also accommodating incomplete information in the captured packet streams. This report discusses the methods by which the US Army Test and Evaluation Command's Aberdeen Test Center and the US Army Research Laboratory's Computational and Information Sciences Directorate collaborated to utilize high-performance-computing (HPC) resources to extract VoIP call information and metrics from very large volumes of network traffic recorded in a tactical environment. The extracted voice call metadata and associated metrics were stored as part of the command, control, communications, and computers Data Model, which was delivered to the analysis community for evaluation. In addition, the HPC processing also extracts actual voice conversations (as audio files) from the traffic flows for analysts to review.