Performance comparison of a custom emulation-based test environment against a real-world LTE testbed

Notwithstanding the value of Long Term Evolution (LTE) towards an improved user experience in next-generation networks, its associated high complexity is known to place computational and time burdens on testing tasks involving real-world platforms. Simulation is currently the tool most widely used to tackle this issue. LENA, for instance, is an open source simulator based on ns-3 that allows the design, evaluation, and validation of LTE networks. Despite of modeling the main LTE elements and interfaces, one limitation of LENA is that it docs not support the use of external traffic entities in conjunction with the simulation. In this paper, we describe how the ns-3 LENA LTE framework can be customized for use in an emulation-based test environment that allows a wider variety of real-world applications to be run over the simulated links. To validate our emulation results, we use as benchmark a testbed that differs from the aforementioned test environment in that the ns-3 server running the simulated network is replaced with a network made up of real-world platforms. Initial validation results, based on limited tests using an industry-standard VoIP test tool and iperf throughput tool, demonstrate that ns-3 LTE models can deliver voice quality and latency as good as an experimental testbed using actual LTE equipment over a range of signal-to-noise ratios. Similar conclusions arc also drawn for throughput, thus confirming the suitability of our emulation approach as a viable means to predict performance in real LTE networks. The good agreement of our experimental results is possible not only because the same functionality is implemented in both experiments but due to the use of the same traffic generation tools in the simulated and real-world LTE networks not -- possible in standard LENA simulation.