IoT 15(2): e1

Research Article

An automated testbed for profiling the packet send-time accuracy of embedded devices

Download942 downloads
  • @ARTICLE{10.4108/icst.tridentcom.2015.259826,
        author={Ricky K. P. Mok and Weichao Li and Rocky K. C. Chang and Kwok-Wun Yung and Ching-Ho Chan and Yat-Sing Poon},
        title={An automated testbed for profiling the packet send-time accuracy of embedded devices},
        journal={EAI Endorsed Transactions on Internet of Things},
        volume={1},
        number={2},
        publisher={EAI},
        journal_a={IOT},
        year={2015},
        month={8},
        keywords={embedded devices, network measurement, packet send-time accuracy},
        doi={10.4108/icst.tridentcom.2015.259826}
    }
    
  • Ricky K. P. Mok
    Weichao Li
    Rocky K. C. Chang
    Kwok-Wun Yung
    Ching-Ho Chan
    Yat-Sing Poon
    Year: 2015
    An automated testbed for profiling the packet send-time accuracy of embedded devices
    IOT
    EAI
    DOI: 10.4108/icst.tridentcom.2015.259826
Ricky K. P. Mok1, Weichao Li1, Rocky K. C. Chang1,*, Kwok-Wun Yung1, Ching-Ho Chan1, Yat-Sing Poon1
  • 1: The Hong Kong Polytechnic University
*Contact email: csrchang@comp.polyu.edu.hk

Abstract

Embedded devices, such as home routers or single-board computers, are becoming more powerful and affordable. Many of the existing Linux-based network measurement tools can also be run on these devices through cross-compilation. These features provide more incentive for network administrators or researchers to employ them as network measurement probes. Large-scale measurement projects, such as BISMark and RIPE, have already deployed more than hundreds of home routers as vantage points to measure the Internet from the edge.

Our previous work showed that these embedded devices have much lower packet send-time accuracy compared to commodity PC. The lower accuracy limits their ability to acquire sound measurement results in high-speed networks. In this poster, we present an automated testbed to systematically benchmark the performance of four popular embedded devices (Raspberry Pi I & II, ECS LIVA, and a TP-Link Travel Router) and suggest the network types that they can accurately measure. The testbed can also evaluate the performance of the kernel-space tools using OMware. The results can be visualized through a web interface, thus enabling users to compare various devices easily.