AVIWest DMNG Pro180 and Studio: Hands On Review
Paul commented that they had extensively tested using third-party APNs (not operated by the network operators). They found such providers gave significant improvements on throughput, and this was assumed to be down to lower contention at the private APN that would be found in the normal network operator APN. I intend to be exploring this further in a separate article in the future since it seems like a subject worth its own investigation for this audience.
With the SIMs online, our focus turned to the other end of the link—the “receiver” or, in the cellmux terminology, the “demuxer” that will terminate all the separate channels of data received over each data path, sequence them, combine them and pass back to either the Network card, the hard drive or the SDI / video output graphics card as a single video stream.
In the DMNG system the receiver/demuxer is termed the DMNG Studio, a 1RU unit shown in Figure 4.
Figure 4. The DMNG Studio
The Studio is driven by a web-based GUI (Figure 5), and there re some capabilities to remotely switch on and off recording and so on from the Pro180.
Figure 5. The DMNG Studio’s web-based GUI
We quickly found it on the network by looking at our local DHC reservations and then accessed the UI via a web-browser, and simply set it to record everything it received, as shown in Figure 6.
Figure 6. Setting up the DMNG Pro180 to record everything it receives.
Back to the Pro180 it notably has two encoding DSPs. This means that you can dual stream in multiple bitrates or formats, or you can archive locally on the DMNG at a high quality, for higher latency store and forward workflows. I asked for a local copy on the Pro180 to be recorded as well as the copy recorded remotely on the Studio.
I left it to Déborah and Paul to deliberate on the optimal bitrate target while Michael and I got my camera hooked up in the car.
Testing the DMNG Pro180
So let’s quickly recap:
- I have a SONY Handycam with a composite video out connected to the portable DMNG chocolate-box cellmux, the Pro180.
- That unit is connected by 7 SIM cards (from three operators) to the internet via each SIM card’s data service.
- The unit is going to encode the video to H.264 and is going to send bits of that encoded video over each of the separate SIM card’s data services.
- That data will be routed across the network operator’s backhaul network to the APN, where it will be billed and routed onto the Internet.
- From there the data will route to the DMNG Studio receiver that is located on my desk, connected to the internet by an Ethernet cable to my router. (My downstream into the receiver is “speed testing” at 20Mbps.)
Normally a user would connect the SDI out of the Studio device into their TV network playout or to a vision mixer or perhaps another encoder. It is also possible to receive the video live as an MPEG-TS over IP. In today’s test, since we were all going out in the car together to perform the test there was no need for local monitoring and so we set the DMNG Studio to simply record to disc.
Returning to this bitrate targeting; as I mentioned above I left this to Paul and Déborah and they settled on a 1Mbps target stream with a 2.5Mbps recording locally on the Pro180. They had a few concerns about the overall backhaul capacity of the masts we were going to pass being able to handle a very high bitrate stream, and because I wanted to look for “flatness” in my bitrate graph they went for a CBR stream of 1Mbps.
The picture quality was certainly very usable at 1Mbps.
I am hesitant myself to comment on video compression quality from the point of view of the application of a codec such as H.264; this is much more the type of article that I am sure will emerge in due course from my colleagues Jan Ozer or Tim Siglin (among others) as they benchmark the codecs that the cellmuxes deploy.
For now, you can look yourself at the quality. I have posted the video on YouTube (bear in mind they will all be encoded again by YouTube—but at least that transcode will be the same for all my test videos):
The journey was pretty uneventful however there are a few key checks I look out for on these runs:
There is a known “dead spot” in to the journey; a point where no cellular devices get signal. When I am trying to find ways to chart the quality, I always look for degradation at this point on the timeline. The DMNG’s on screen bitrate reading was a clear indication that we were losing throughput as we entered the area, and this eventually resulted in a temporary loss of signal. I detail this more below, but it was essentially the key thing I was looking for during the test. The dead spot clearly broke the link completely for a short while, although it auto-recovered quickly as one would expect. Knowing this had happened means that I know I have at least a “signal integrity break” to look for in the analysis.
The name's Bond. Bond II. And it's a cellular bonding device and encoder that delivers terrific performance at a price point that's almost unbelievable.
News gatherers can record and stream simultaneously, or conduct interviews with low latency.
What to look for in cellular multiplexing devices -- perhaps the fastest growing streaming technology.