Measurement Framework for Assessing Reliable Real-Time Capabilities of Wireless Networks

Abstract

Network latency and reliability are becoming critical performance indicators for future mission critical services based on wireless connectivity. In this article we present a portable and low-cost measurement framework for quantifying the endto- end (E2E) latency and reliability of communication systems based on single or multiple links. The proposed framework is tailored for characterization of access diversity mechanisms, for example, multi-connectivity or hybrid access methods, which are expected to be essential for ultra-reliable wireless communications in 4G and 5G networks. An active-probe measurement approach is adopted, consisting of a client that transmits timestamped packets to a server, and vice versa, via one or multiple communication links under test. The setup allows us to measure the one-way latency performance in uplink and downlink separately, and allows fully synchronized measurements of multiple links in parallel to understand the benefits of multi-connectivity mechanisms. Traffic generation is flexible in terms of packet size, transport protocol (TCP or UDP), and inter-arrival time interval of the generated packets, hence being a powerful tool for benchmarking the performance for different application-specific message flows. Special emphasis is put on the tail of the latency distribution, that is, the latency that can be achieved with a certain probability constraint, for example, 99.9 percent or beyond. Given the challenge of measuring such rare events, we measure and validate the stability of the measurement equipment to ensure that the observed performance is not compromised by undesired behavior or malfunction of the equipment. The system has a measurement error below 400 μs with 99.9 percent confidence and below 3.1 ms with 99.999 percent confidence. The capability and applicability of the measurement framework is illustrated by a measurement test of two simultaneous LTE connections that provide robust communications to an aerial drone.