Yesterday, I arrived at the Brussels Midi station in a rainy, grey morning—not a very inspiring start for the day to attend the 2nd workshop organized by European Commission on Broadcast-Broadband Convergence.
In the 5-hour workshop, the consulting company Plum hired by EC presented their analysis aiming at figuring out whether the same infrastructure should be used for TV broadcast and mobile services in the UHF band. Their findings suggest that there are certain (high) cost to implement such a converged platform, but the benefits are far from certain and dependent on many factors such as terrestrial TV service penetration and the growing trends towards Video-on-Demand (VoD). More details on their motivations, objectives and work plan were presented in another workshop in March.
The attendees were mostly from national/EC regulators, policy people from vendors and of course, the most outspoken broadcasters. Interestingly, again, not many other ‘visible’ academics among the 100+ attendees.
The presentation started with the technical analysis done by Professor William Webb, discussing different alternatives for the ‘converged platform’ to provide TV coverage and content deliver equivalent to the High Power High Tower (HPHT) counterfactual (DVB-T/T2 as we know today). The basic assumption is that the 700 MHz band will be vacated by Digital Terrestrial TV (DTT) between 2017-2025, which commented by others as being politically correct but unrealistic. Then it further assumes a number of TV content equivalent to either 60Mbps or 180Mbps (two different cases) must be delivered within the remaining UHF broadcast band (470 MHz -694 MHz).
In the end it is concluded that HPHT with LTE technology is not a good idea, because it simply cannot compete with HPHT and DVB in terms of spectral efficiency. But Low Power Low Tower (LPLT) options are considered more favorable because the reduced size of interference zones between different contents enables large scale Single Frequency Networks (SFNs). Basically a spectrum reuse factor of 1.1-1.2 is envisaged for LPLT option (in comparison to a reuse factor of 3-4 for HPHT). A further ‘contentious’ assumption is that the spectral efficiency of an modified version of LTE-eMBMS could achieve 2bps/Hz even in rural area with SFN. This value is used as the base line for the rest of the calculation (William explains this number is provided by Qualcomm. And even though he has reservation about it, he still used it in the analysis for reasons we will explain later). In comparison, DVB-T2 SFN is assumed to have >3.5 bps/Hz spectral efficiency.
Afterwards, David Lewin presented the cost benefit analysis to show what is the incremental cost to deploy such platform and what is the benefits in terms of spectrum savings translated into monetary values. It looks at a hypothetical European country of 10 Million households and 10000 base stations, i.e., 1000 household per base station (our data on the greater Stockholm region indicates ~800 households per base station). The analysis results show that the only scenario with net benefits would be when DTT penetration level reduces to <30%. A bit surprising fact to me is that the primary cost for LPLT with LTE broadcasting apparently comes from the user equipment upgrade, which is estimated at 70 euro per house hold, plus 70 euros for 20% of the household to reorient their antennas. The ‘incremental’ cost in upgrade the base station for LTE broadcast and the extra backhaul capacity is less significant because the operator would need to upgrade them anyway in order to cope with the grow data traffic.
In the end, the conclusions seem to be painting a rather dim prospect for the converged platform. Because even with the ‘optimistic’ assumption on spectral efficiency, a positive net benefit is observed in a low DTT penetration scenario. And the difficulty to find enough spectrum for simulcast during the transition period makes it even less attractive.
To a large extent I agree with their analysis method and I am totally fine with their seemingly simplistic and ‘optimistic’ assumptions on LTE spectral efficiency. Because the logic behind such assumption is very clearly motivated: if the proposal won’t work even in the best case scenario, why bother arguing about whether the spectral efficiency is 0.2 or 0.5 bps/Hz too high?
However, I am quite concerned by the fact that the unicast capability of LTE is not highlighted at all in the analysis. Their argument against any substantial benefits of providing interactivity is based on the estimates that currently 90-95%of viewing remains in home where other hybrid TV solution already exists. But I would argue that the very reason so little viewing happens on the move is precisely because there is no cheap/good options for on the move viewing with interactivity, not because of the lack of demand!
This concern becomes more relevant if we recall that the time frame under discussion is about 5-10 years or more in the future. Because while unicast might not work currently due to the high number of viewers per TV channel per cell, we clearly see the trends where younger generation have much higher demand for VoD type service and the increasing number of TV channels. In such case, the interactivity provided by unicast and its capability to deliver a unlimited number of TV channels to a limited number of viewers would become extremely valuable. This aspect perhaps should be included in the consideration when comparing to the net benefit achieved by the satellite option, which appears to be the cheapest option (for a limited number of TV channels) but offers only broadcast service. In other word, the modeling or predication of the user behavior seems to be rather conservative in the analysis, which seems to be at odd with the ‘optimistic’ spectral efficiency assumption.
After the workshop, I took the opportunity to exchange some ideas with Prof.William Webb who shared the view that it might be necessary to look into the potential of unicast when demand for VoD service increases. Their next workshop on presenting the final verdict would be in November this year.
The slides of this workshop can be found here. One of our latest paper also looked at the similar issue and presented in DySPAN’14, the poster of the paper which won us the Best Poster Awards could be found here.