February 23, 2012
By Dom Robinson Contributing Editor
Featured Articles
For the rest of the Spring 2012 (Sourcebook) issue of Streaming Media magazine please click here

Buyer's Guide: CellMux, Link-Aggregator, and Channel-Bond Devices

This article appears in the February/March issue of Streaming Media magazine, the annual Streaming Media Industry Sourcebook. In these Buyer's Guide articles, we don't claim to cover every product or vendor in a particular category, but rather provide our readers with the information they need to make smart purchasing decisions, sometimes using specific vendors or products as exemplars of those features and services.

In 2011, cellular multiplexing moved from an innovator-only technology into the early stages of adoption.

To understand why this happened, let’s briefly take stock of how the breaking news story has changed.

In 1991, an event would occur, and local observers would contact either the police or the local media. Phone calls, telegrams, faxes, and possibly bulletin board systems and email would start to notify local representatives of news agencies. They would mobilize reporters and outside broadcast uplink trucks from local depots. Given the portability of the technology required, radio reporters would probably get online first using public switched telephone network and integrated services digital network codecs to get the first reports flowing. As the TV crews arrived and got set up, pictures would start to appear via satellite, being downlinked and fed into scheduled playout on TV broadcast channels. Only really significant events would justify the thousands of dollars for reporters and TV satellite uplink to then provide “live” coverage. All in all, as far as the end consumer was concerned, news was something that evolved on a 24-hour basis; only really dramatic events received live coverage; and all content would be played out only if it achieved high-quality production standards.

By 2001, the internet had arrived. RSS, email, and push technologies used by local observers accelerated the notification process since more of the general public was versed in the technology and could notify the news agencies directly. Indeed, trend-based searching had emerged, and alerts and other such triggers could spot news as it developed and mobilize the technical teams much quicker. Once mobilized, the reporters and contributors would still need to travel to the event and link up. Despite webcast technology’s ability to deliver reasonably good quality images to the TV news companies, its use was very nascent. Some low-quality footage would occasionally make it to broadcast from camera phones or handycams, but all in all, the TV broadcast industry still viewed with caution any video content that was not produced to a very high-production standard. The general public still relied primarily on TV broadcast news media as the earliest sources for breaking news.

By 2011, the landscape had changed completely. Today, without a doubt, Twitter is the technology that breaks the news. Disintermediating the entire news workflow of a decade earlier, the local observers themselves can communicate directly to the audience.

As soon as the event occurs, observers can tweet from their phones, and these tweets can include images. This low-bitrate general packet radio service or 3G stream gets the story to the audience at the same speed that it formerly took to notify the news agencies that the story existed. As soon as Twitter has announced the event, we see links to uploaded videos, shot on smartphones and published on sharing sites such as YouTube. There may be a scattering of capable users who are familiar with live streaming services, such as Ustream or Livestream, and have a smartphone app that enables them to stream the events live. The probability of such a user being near an event is increasing by the day. Accordingly, even the most extreme, unexpected, and remote events are finding their ways to audiences live, in a way that was never before possible because of the scarcity and cost of the technology that could provide the bandwidth required to stream live video. Video footage documenting events is now widely available as fast as the traditional news agency has just about digested what has happened and begun to react. In fact, the agency itself is more or less relegated to being in the audience.

This is transformational.

Still, there is a demand for quality in broadcast media, and mobile/smartphone live video streaming is usually limited by the uplink capabilities of the smartphone’s connectivity. Typically, this means that the single data stream enabled by a smartphone’s SIM card is a limiting factor, maximizing throughput at anywhere between 64 Kbps and 384 Kbps on typical 3G networks, and only then if the user is near a cell mast to get a stable, clean signal.

Enter the cellular multiplexer (CellMux). The CellMux comprises multiple SIM cards, each representing a discrete data path. However, by adding some relatively simple intelligence at both ends of the data path, multiple paths can be aggregated together. And in a packet video environment, a multiplexer can be used to break up the video stream, sending bits of video down one path and other bits down another path, to achieve a throughput several times greater than the single-path-max of 384 Kbps. So broadcast quality video at around 1.5 Mbps or even greater can then be sent live over a group of cellular data networks.

In theory, at the camera/encoding end, there is no limit to the quality that can be filmed, encoded, multiplexed, and sent over the cellular data network. However, limits do exist in the cellular networks themselves. A typical cellular mast has from 3 to 9 T1 lines connected. Since each of these T1’s is 1.5 Mbps, it’s easy to see how even a couple of CellMuxes that live in the same area will saturate the mobile networks’ backhaul, and contention for the limited bandwidth could cause problems for all data services on that cell. In some limited cases, cellular operators will allow CellMux operators to pay for a premium class of services, but this is still very rare and expensive, and it requires planning, which would negate all of the opportunity that CellMuxes present.

One of the CellMux vendors, LiveU Ltd., has patented on-board techniques for reaching cell masts farther away than could be reached by a typical smartphone. This is, for me, a key and unique selling point that a buyer needs to consider. Not surprisingly, LiveU is not currently the cheapest option on the market.

However, if you are relying on getting that feed back to base, and for some reason a handful of other operators or smartphone users are saturating the local cellular mast’s backhaul, this feature may be the killer for your application. As far as I know, this feature is only available on LiveU devices, and it claims to have a patent.

Indeed, as the granddaddy and pioneer in this space, LiveU was well-positioned among news agencies to provide some of the earliest and highest-quality footage of a variety of events during the London riots and the Arab Spring in 2011. Much of the footage from Libya, Egypt, and London was backhauled over LiveU’s devices.

I mention this because the London riots were a good exercise of this patent. Not only were the TV companies that use LiveU streaming from the street as they provided live coverage, but so too were emergency services and the rioters themselves. The local cellular masts were extremely busy. LiveU’s patented technology allowed the TV companies to reach cells that the rioter’s BlackBerries couldn’t, and so they found clear(er) data paths to deliver the video.

Since these devices are subject to the cellular networks’ service quality, they are not always a suitable option for live streaming, where quality guarantees are required. For this reason, there is a trade-off between operational flexibility and quality, and while I would expect to see new angles and interesting footage, for example, at sporting events, I would be very surprised if a planned sporting event was to consider a CellMux as its only backhaul technology, given that satellite is far more reliable and comes with service guarantees. Satellite is, however, more expensive and requires either a self-alignment system built into a vehicle (limiting its portability) or a skilled engineer to align the dish. Satellite is more “ad hoc” than fiber, but considerably less “ad hoc” than CellMux. So where portability and time-to-live are the key performance indicators, CellMux will be the best choice.

Let’s get down to a quick guide to choosing your CellMux.

My initial review in the autumn issue of the European edition of Streaming Media featured just three devices, and at the time, only AVIWEST’s product was self-contained in what I called a “chocolate box” format—a stand-alone device that fit on a camera. Within a short space of time, both LiveU and TVU Networks, Inc. quickly released small form-factor devices too.

But that’s not the end of the story. Several devices are on the market that, when used with a separate encoder, can provide the ad hoc connectivity in the same way.

Mushroom Networks, Inc. has released Streamer, a channel-bonding router that’s not specifically for video, but, working in conjunction with a separate video encoder, can mux any data over the link. Viprinet GmbH also produces a range of bonding routers, but we have yet to review these fully at the point of writing.

Attila Technologies licenses network technology that can be used to build such devices, and indeed the IEEE 802.3ax and associated protocols are also now in the public domain. Implementations, such as the Link Aggregation Control Protocol, are even bundled with standard Linux installations, although they are not necessarily trivial to configure.

I am also tracking some manufacturers in Taiwan that purport to have been producing link-aggregation routers that support video.

The most recent entry, and undoubtedly one of the most eye-catching devices, is Teradek, LLC’s Bond, which includes Teradek’s establishing encoding box, the Cube. Not only does it look like it might walk to the event itself, but its cost is eye-catchingly low.

So what are the things to look out for? The killer app has the ability to reach a wider range of cells than those in immediate proximity, since doing so ensures that you can film at the center of the action, but you don’t have to contend with all the other network activity that such an event is likely to produce. I can see that devices that lack that capability will, at that critical moment, fail—they are not ideal for mission-critical content workflows!

The other factors that I think will be relevant for most users will be as follows:

Portability: Devices require a small form factor that doesn’t compromise function/capability.
Ruggedness: This is a field kit; if it can’t survive in the field in challenging conditions, then it is going to fail when it is most needed.
Power: The unit can share the same power as the camera; if you have power for either, you have power for both.
Low latency: In this environment, latency greater than 150 milliseconds will cause problems with sending two-way live talk back to the studio. While not all applications will require this, enough will, which will make this a must-have option, even if it is only available in a low-quality mode.
Cost: In some traditional broadcast environments, outside broadcast is known as occasional use. The total cost of ownership of such technology has to align with the potential revenue that it enables. Some providers lease the technology as part of a service model. While the larger news agencies can afford to take the risk of committing to a service model where they may or may not have an event to cover, smaller players need to think carefully before they make such commitments. For some, the cheaper options may appeal more.

While there are many other factors that any one buyer may find critical, this list is intended to cover the key features that, in my mind, are critical to the intended audience of these devices. In the following table, I use a crude scoring system. At some point, I hope to have all the devices for a true shootout, but so far, this has proven difficult to arrange with the vendors. Perhaps there is something they don’t want me to see?

Some of the data is not easy to obtain, so I have used 3 as a neutral point where I am not sure if the feature is included.

CellMux Chart

To sum up, if mission-critical is paramount, and cost is less of a constraint, then LiveU’s patented high-gain antenna gives it a clear edge in the field. Its nearest rival, TVUPack, may well also include a high-gain antenna, but I foresee a patent issue if LiveU challenges it.

For price alone, Teradek’s Bond is amazing—right in the sweet spot. At far less than $3,000 (not including the Cube encoder) and with no commitment, this is, in my mind, the best-priced unit of the true CellMuxes. The buyer must understand the implications of adding several data services on 3G to this device, and this will vary considerably with use and territory. It should also be noted that LiveU, TVUPack, and AVIWEST’s IBIS all come with preplanned data options, which are negotiated by the companies to help manage cost. Each use case will be different, and I would be very interested to hear from users about the costs of operation.

The last three entries in the table contain the link- aggregators and channel-bond devices only. Although they are notionally considerably cheaper than the CellMux devices and offer greater flexibility since they present high-capacity Ethernet in the field for use with any application, not just video, you will also need a video encoder to create the live stream. This reduces the portability; and clearly, a lot of these devices are designed to live in a rack in a permanent setup, rather than being field-ready. For versatility, these options are better priced, but for the specifics of live video streaming, a “true” CellMux with an integrated video encoder will be better in the field.

I am going to continue to try to cover the evolution of this technology throughout 2012, so I am interested in any case studies and feedback on these and any other similar devices. I am also exploring producing a “how-to” article explaining how to build your own link-aggregation and channel-bonding device.

So keep your eyes on this space, and bear in mind that a considerable amount of the “from the field” news coverage you are now seeing is being produced using one of these devices. They have truly come of age.