One of the trends that is sure to attract a lot of attention at the mobile industry's biggest trade show, 3GSM, next week is the concept of delivering TV to handsets. But there are three questions mobile TV must answer before it has a chance of success. Does anyone want it, can it be made to work reliably and will it make money?
These are questions to which nobody knows the answers -- least of all the people who failed to ask the same questions of 3G before investing. Yet these are the people who are keenest to sell TV on mobile phones and PDAs to the masses.
History isn't kind to the idea of mobile TV. Pocket TVs were first promised in 1966, first available in 1977 and first affordable in 1988: by 1999 most of the manufacturers had lost enthusiasm for the market. Most pocket TVs are bought in a mild gush of gadget desire and then left in a drawer until the batteries leak.
On the other hand, the Sony PSP and the video iPod have made the small screen sexy again: with mobile phones fully equipped to display telly on the move, the only thing that can stop operators pushing the service is if nobody wants it.
Maybe not even then. There have been a spate of user trials, with mixed results. With so many variables -- content, number of channels, handset type and technical characteristics of each broadcaster are different each time -- there is no easy way to compare even those trials in one country, let alone those around the world.
In the UK, O2's trials saw between three and five hours of TV consumption a week, but BT/Virgin Mobile could only clock up just over an hour -- and a marked reluctance to spend more than a fiver doing so. So an operator determined to find evidence that mobile TV is a good idea will be able to do so -- and with the fear that everyone else is doing it, that determination will be easy to conjure up.
In an ideal world, mobile TV would be just another digital service delivered alongside connectivity and voice over the existing 3G system. Unfortunately, the bandwidth demands it makes are well beyond any current or projected network capacity: even with full HSDPA, a 3G system will be saturated if half the subscribers watch just seven minutes a day.
Put another way, five minutes of TV a day takes as much bandwidth as 2,000 minutes of voice a month -- but will never be able to generate an appreciable fraction of the revenue.
Any viable mass-market service will have to use another band and to broadcast its signals, so that one transmission reaches hundreds or thousands of subscribers. And while mobile TV needs much less spectral space than existing telly -- because it can use more efficient digital encoding and because it can be optimised for the much smaller screens of hand-held devices -- finding the right spectrum is one of the hardest problems to solve.
Choice of frequency is key. Mobile television needs tens of megahertz, which requires high frequencies, but also needs to be able to work reliably inside buildings and on the move, which favours lower bands. The ideal frequencies are those already used for UHF TV, a number of which will become available as countries switch over from analogue to digital transmissions. Less temptingly, these frequencies are heavily in demand for other services and will be expensive to win at auction, and will vary from country to country.
Some operators already own or have control of individual bands in some countries; Qualcomm's MediaFlo being a case in point with its occupancy of 700 MHz -- UHF TV channel 55 -- in the US. It will be broadcasting multichannel TV on this and selling the service to other network operators. In the UK, O2 has been testing with DVB-H on UHF, a modified version of the existing terrestrial TV standard which is also Nokia's choice. In Australia, Nokia is testing DVB-H together with Telstra and The Bridge networks. Others are finding ways to transmit television over existing services such as digital radio with standards such as DMB, here the bands are internationally allocated -- at least in some regions -- and in some cases only lightly used, but generally there are bandwidth limitations and contention from other users.
The proposition by IPWireless, TDtv, is to use allocated but unused 3G allocation -- the unpaired frequencies. Most operators have two classes of frequency, paired and upaired, with the paired being use for two-way services such as voice and data. The unpaired is intended for asymmetrical services such as data, but is mostly unused. Because it's adjacent to the voice and data allocations its easy to modify existing base stations to transmit there, and of course licensing issues are eased. Other advantages are that it should be relatively easy and low-cost to include the extra radio in handsets, a problem that becomes more complex to solve the further away the TV band is from 3G, and that roaming will be easier.
Whatever bands or sorts of transmission are used, though, the operators will be wary of investing too much on infrastructure before the size and nature of the market is better known. It's even possible that mobile TV will never be that much of a revenue earner: people already spend a lot of money on multichannel TV and mobile services, and may be reluctant to hand over even more for something that seems to largely replicate stuff they've already paid for. Instead, mobile TV may be something that operators have to offer to stay competitive, or as part of a triple-play package to keep users faithful to a single supplier of multiple services. This may also be more appealing to companies who already have content deals with TV channels and production companies.
Until the revenue and cost implications are better known, expect mobile TV services to be offered over normal 3G. This may not have the bandwidth to support anything like a true mass market, but it will provide early indications of how the full service should be deployed. TVtd has an advantage here, in that by allowing the network to be rolled out on a cell by cell basis it can be installed only in areas where demand justifies: systems with separate networks will find this more difficult. In any case, customers will want on-demand and obscure channels that can't be efficiently delivered over a full broadcast network with its limitations of a few tens of channels. Here, 3G will be used to deliver the personalised services, probably seamlessly integrated with the broadcast sector.
Even without the promise of massive revenues -- and without the proof of massive user demand -- it seems certain that mobile TV will be arriving in some form or another from most mobile operators in the short to medium term. Feel free to ignore it, though: even the most advanced multmedia delivery system will still come, mercifully, with an off button.