Your Browser is Out of Date

Nytro.ai uses technology that works best in other browsers.
For a full experience use one of the browsers below
Computer Vision for Public Transportation - Phase 2
Overview Architecture concepts Solution architecture Validation
Validation
Validation overview
Camera simulation Offload simulation Metrics collection
Results and findings
Validate bus offloading video over5G Validate bus moved before offload complete Validate bus depot to data center video transfer High availability
Summary
Sizing the solution Conclusions References

Results and findings

Home > Workload Solutions > Computer Vision > Guides > Computer Vision for Public Transportation - Phase 2 > Results and findings

Results and findings

  • Validate bus offloading video over5G

    This use case tests the lab infrastructure built to determine how efficiently video recorded on the EPIC iO device (bus) can be offloaded over a private 5G network. This use case is consistent with how video offload would work in a real-world depot environment.

    Test Specification

    The steps conducted were as follows:

    • Used all 9 camera video streams ~14 Mb/s network transfer rate
    • Set up trace with Live Optics for 3 servers:
      • UIG
      • Depot
      • Data center
    • Recorded metrics with Live Optics for 12 hours of steady state workload

    Test results

    The first group of results are to investigate the transfer rates from the Dell 7090 device to the Cradlepoint device. All outbound traffic passes from the Dell 7090 to the Cradlepoint and then northbound over the 5G network. In this test, the Dell 7090 device is the only device that is connected internally to the Cradlepoint using a 1 Gb/s LAN cable. The traffic measured for inbound video data to the Cradlepoint is as follows:

    Figure 8. Inbound data to the Cradlepoint
    This graphic shows the traffic measured for inbound video data to the Cradlepoint.

    This data shows a periodic spiking pattern for the transfer rate. This is expected as video bitrates will vary with the average across the full time period of 13.6 Mb/s.

    The next transfer step for video data when it exits the Cradlepoint device. This is shown as "wan_out" in the graphics:

    Figure 9. Outbound video data from the Cradlepoint
    This graphic shows the traffic measured for outbound video data from the Cradlepoint.

    In this series it is easy to see by comparison with the above chart that the data coming in over the LAN (red) was forwarded directly out to the 5GWAN connection (blue).

    The next northbound hop for the video data is onto the NFS share in the bus depot. In order to capture these low-level metrics shown below, the Linux bmon tool was used.

    Figure 10. Linux bmon tool example
    This graphic shows a screen capture of the Linux bmon tool.

    The test ran for 12 hours and transferred over 60GB from the bus to the depot over our 5G network in the lab. Our main observations were:

    • Zero bytes lost of the 60 GB transferred
    • Zero lost packets from the 9.6 million received over 5G
    • No errors of any type. Therefore, the communication from the bus to the depot over 5G is stable and reliable for continuous data streaming.

    Findings

    • All data transferred in from the Cradlepoint is passed out over the 5G Network.
    • During detailed analysis of the depot device, it was shown that zero packets were lost over the 5G network.