A Buyer's Guide to Cellular Multiplexers
Forget the sexy form factor and look deep inside the technology. Here are the three questions to ask before buying a cellmux.
Despite the explosion of choice in the cellular multiplexer (cellmux) sector in the past 18 months, the core things you need to think about when making a purchasing decision are actually quite limited.
Form factor, power, portability, and interface usability are all very important, but when it comes to determining whether or not the device will serve your workflow, there are really only three questions you absolutely must answer.
Three Questions to Ask When Buying a Cellular Multiplexer
- Is the device antenna compatible with the networks you'll be using it on?
- Do you really need LTE, or will another network interface suffice?
- What, exactly, is your video workflow?
In several recent articles on the subject, we have explored antenna issues. There are a number of schools of thought here. Antennas can be compact and portable but with optimized "high gain" so that the signal between the networks tower or mast and the unit will be strong enough from marginal locations such as basements or in the midst of large crowds.
However, sometimes these antennas may not be certified by the local radio authority or the telecom itself, and so they may operate "illegally." That may not be an issue for a war correspondent reporting from the front lines, but for a national broadcaster producing commercial television or live streaming, this is a serious consideration.
The telecom could well have the means to identify such offenders, and it can block the transmissions from those offended devices to its cell sites. But if the broadcast industry doesn't follow the regulations on how each frequency domain can be used, it can't be expected to be able to protect its own broadcast frequency from being interfered with by others. Its not just about the law of the land; it's also about industries respecting each other. Paul Shen, CEO of TVU Networks, Inc., points out that if the broadcast industry doesn't respect standards and legal frameworks within the telecom environment, then the telecoms will diverge from broadcast too and technology breakthroughs such as H.264 (an ITU standard) will not be possible.
We must remember that these standards are not the result of some selfish group trying to dominate a technology space but rather represent agreements between two organizations or states or between an organization and its members. They are essentially not intended to be prohibitive and are there to make things work better.
So if one can operate an antenna on a cellmux in an optimal way so that it makes the cellmux work better, why on earth would an operator choose to buy a cellmux that doesn't present the same optimizations purely so that they operate within the cellular network operator's guidelines and standards? Suppose Broadcaster X arrives at a rail disaster with its cellmux and begins to transmit using an illegally optimized antenna, resulting in all the other users of the local cell masts being "drowned out" by the cellmux. Whatever the value of the news story, the fact is that the emergency services and other users of the network must at least have equal priority to ensure correct emergency help is available.
So a key decision-making consideration should be ensuring that the device is compliant with the networks it will be bonding together. To be honest, you may have to take the manufacturer's word for it, unless you want to deconstruct the device in its entirety on the first day of owning it! I should imagine certification and badging by various operators will ensue. However, given that the whole point of these is that they work with multiple operators, it likely will be some time before proper certification appears.
This leads us nicely onto my second key criteria.
Range of Network Interfaces
Here is a typical list, copied from one vendor's data sheets: "HSUPA/HSPA/UMTS (W-CDMA)/GPRS/ CDMA2000/CDMA450/LTE."
Now before we go any further, do you, as a potential customer for these technologies, actually know what mobile network interfaces and services are available to you in your country?
LTE is becoming the most popular option to offer, but for the next year or so, LTE will be broadly useless in most of Europe, the U.K., and the U.S. Just check out http://go2sm.com/iphonelte to see Apple's list of 4G/LTE rollouts, which would make it seem like 4G/LTE is prolific. But if you look at U.S. carrier U.S. Cellular's map at http://go2sm.com/uscelllte (make sure you switch it to 4G LTE and note that 4G LTE is the dark green), you will see that LTE represents a very limited geographical option if you want that speed advantage from everywhere for broadcasting.
Also, with pricing in the U.K. standing at £36 (about $57) for 500MB of data transfer, it's worth noting that in some territories you will potentially see a hole in your pocket quickly appear when you start broadcasting a 1.4Mbps stream.
I strongly suggest you look at the type of events you are expecting to cover and understand the locations that they will be occurring in. If they will, for example, all be live nature shoots in the outback or news correspondence in the Middle East, you can probably save money by not opting for an LTE-focused system. Instead, go with a GPRS model with EDGE and similar 2G/3G capabilities.
If, however, you plan to replace awkward line-of sight satellite links with more flexible cellmux technology for rapid setup within urban zones in major first-world cities, then of course a system offering only LTE may be a much better option.
It is also worth thinking about the effect of 4G investment on 2G and 3G networks. Increasingly, the backhaul available at the towers should reduce the competition for all services, and this will increase the probability that you will be able to bond multiple network links together to achieve the highest throughput.
As these older technologies churn away, you may no longer need bonding at all since the available LTE (or other future 4G services) may simply be sufficient in a single channel mode from one provider. Indeed, it wouldn't surprise me if those cellular providers subsequently begin to offer broadcast grade service-level agreement, "leased line," and other such high availability services over LTE to broadcasters. This will potentially change the whole argument as to the need for a link-aggregating/bonding cellmux, since it may be that the network interface card in a standard video encoder will support LTE (or other 4G networks that may emerge). The whole trick of bonding multiple provider links together will then become redundant. For today, though, cellmux still offers reliability that no single-network solution can match.
Let me first explain the terminology: Alongside "cloud," the word "workflow" in this sector has suffered more abuse than almost any other. The two terms are wide open to interpretation, so one engineer's "cloud workflow" is another's "engineering process" and yet another's "business model." So perhaps we had better agree on a definition for "workflow" in this context.
As applied to live streaming, the continuous flow of the content, as it is transcoded, transrated, and transmuxed as it moves from CCD-in-camera through wire, silicon, electromagnetic encoded packet, and back out to LCD and retina is, for me, the "end-to-end workflow."
Obviously the domain for which most engineers are responsible is only a subset of that end-to-end workflow, and this scope of this domain is the video workflow that you must be clear about when buying your cellmux.
In making your purchase, it is important to understand what your own typical role and domain of responsibility is.
If your role is to be providing a link from a camera to a video switcher/mixer 100m away at a conference or a concert, you may be buying the wrong equipment. In this context, a 100m length of video cable will serve you much better.
While a cellmux can usually support "store and forward" transfer of your content back to the studio to be edited before it is delivered on air, ask yourself, "Do I need these clips back to the studio as soon as possible, or could I pop back to the hotel room after the shoot and upload them myself overnight so the editors can start work on Monday morning?" Again, in this instance, a cellmux is probably overkill.
However, it is very likely that if you have read this article this far, you are seriously thinking about flexibility in terms of the speed and quality with which you can deliver your video to a studio, as well as the location of that studio.
Assuming this, now put the studio (to whom you are using your cellmux to deliver the video) into one of two camps: Are you delivering to a studio (or platform) that publishes video online only, or are you delivering your content to a traditional TV studio infrastructure?
There are significant differences in what you may want to happen at the studio. Remember that most (if not all) cellmuxes are actually two-part systems. The studio-side part of the system-the cellmux receiver-is often forgotten, usually because it does not have a sexy form factor! But in many ways, its capability is more critical than the sexy field gear.
The field component acquires the video and audio and then compresses it, usually dynamically adjusting the compression to suit the variable network conditions, based on feedback from the link-aggregator and channel-bonding stages. Then it transmits the signal over multiple network routes (multiplexing) to the studio component.
The studio component receives the multiplexed packets from a variety of routes over the network, sequences them, recombines (demultiplexes) them, and then does something to them. That "something" will be critical to the output of your stage of the workflow; it is defined by the handoff you make to the studio.
Many TV studios will want a traditional SDI (or these days HD-SDI) video source, presented to them on a BNC-terminated connection. That signal must, in all senses, be a TV signal, and so it must use the right codec for the TV infrastructure, must have genlock or another timecode so that it can be properly synchronized with other video sources in the studio, and must perform correctly in all aspects of the TV facility. If the studio receiver component of the cellmux doesn't do this, you may need to transcode the signal with yet another device, and this may reduce quality, increase latency, and add complexity-none of which are desirable. It's much better if the cellmux receiver outputs what you need straight away.
Indeed, some studios may be more advanced and want MPEG-TS digital transport streams, and the output of the cellmux system may require an Ethernet link into an internet protocol (IP) LAN in the studio facility. Sometimes, that may be required as an IP Multicast so that multiple edit desks can receive the feed in parallel. All these requirements must be understood before you make your purchase.
In contrast, many online-only video publishers may not want the signal to ever move out of the IP domain. They may, for example, want an RTSP or an RTMP stream delivered directly from the cellmux into a media server such as Flash Media Server or Wowza Media server. Then, that signal may be handed off to a CDN for delivery onward to end users. The encoder in the cellmux is the only video compression in the entire process, producing super-low latency and minimal quality loss.
Naturally the models will have thousands of nuances, but it is really important not to get solely occupied with the form-factor of the field kit that connects to the cameras at the expense of the unit that gets installed in the studio. That unit has to translate from the proprietary world of the field-studio link into your workflow, and-however clever the cellmux is in the field-not all cellmux setups will have the interfacing you need at that studio end.
This article appears in the forthcoming 2013 Streaming Media Industry Sourcebook.