Earlier this week I wrote that “Mobile networks aren’t for Video on Demand” which was read by many visitors.  The fundamental problem with mobile VoD is that each video user can ask for more network capacity than all of the other phone and web users combined. Mobile network cells are designed to simultaneously service hundreds of users and not 20 or fewer video users.  But this isn’t the end of the road for video on mobile networks and the problems can be overcome with IP multicasting.

Multicasting is an analogue of traditional broadcasting and it is very effective on shared medium networks like 3G mobile networks when delivering popular live television content.  If a high quality (relatively speaking for mobile) widescreen 480P video stream requires 700 Kbps, the network would allocate 700 Kbps regardless of whether there are 10 users or 1000 users watch the same stream.  The same video using unicast technology would deliver a unique copy to every user which means that 20 video users would fully saturate a 3G network and bring it to a halt.  480P would be the high end of the mobile video spectrum since only the high end smartphone devices have 800×480 resolution display or better.  240P resolution video could easily be multicast at 225 Kbps and this would be more appropriate for the majority of phones (and nearly all portable TVs 9 inch and smaller) on the market.

Multicast video would still be limited by how much mobile capacity a cellular operator allocated, but it would no longer be limited by the number of end users watching the video.  If 7 Mbps of bandwidth was allocated, 10 high quality channels could be delivered or 30 standard quality streams could be delivered.  There could be a mixture of 480P and 240P content and higher quality be allocated to the most popular content which would be justifiable based on the amount of bandwidth allocated per user.

An ideal system could dynamically allocate bandwidth based on consumer demand where the more popular content would get more multicast bandwidth.  The Superbowl or Olympics for example could justify higher bandwidth allocation because more people are viewing it.  If 700 people on the same cell tower are viewing the same 700 Kbps video stream, each person is only using 1 Kbps which is 10 times less than the bandwidth consumed by a wireless phone call.

There are competing mobile TV systems already on the market such as Qualcomm’s MediaFlo and the coming Mobile DTV standard which is being heavily pushed by the TV broadcasters.  The biggest problem with these two solutions is that they require an additional radio and antenna whereas an IP multicast solution could be implemented on existing smartphones with their existing antenna and radios.  People don’t like the additional cost and weight of these systems nor do they like having “rabbit ears” protruding from their cell phones.  The other problem is that MediaFlo and Mobile DTV are limiting themselves to the lower resolutions while a cellular solution could be more flexible.

Mobile DTV has the advantage of being free-to-air but consumers have largely fled free-to-air broadcast television because they haven’t found it sufficiently compelling.  Mobile DTV uses the ATSC-M/H standard and the broadcasters are talking about allocating several Mbps of bandwidth to 240P video streams which seems wasteful since 480P could be supported with 600 Kbps and 720P could be supported with a minimum of 2000 Mbps.

At the end of the day, a large number of consumers want access to high quality video content on their mobile phones and they’re less inclined to carry another device or a separate radio and antenna.  What they really want is video on demand on their smartphones but that really isn’t feasible.  Perhaps the next best thing is video delivery through IP multicasting.