On the road to Bandol

I have chaired a session at the SIM Summit, whose topic was the technical challenges to be met in order to satisfy operator requirements. The speakers were mixed between operators and vendors. The session was very informative, especially because there seems to be a good convergence between the actors. In order, the hot topics are as follows:

1. NFC
2. Smart card web servers
3. High-capacity cards

For NFC, it even seems that the technical people have done most of their job. This technology is getting very close to usability, especially as the SWP protocol seems to become the accepted way of communicating with the SIM. The good news for Java Card is that we are ready today to deal with this, as Java Card 2.2 is a perfect base for NFC applications.

Furthermore, as I mentioned in my own presentation, the use of NFC on mobile phones will undoubtedly raise interesting questions. As application providers will not resist the urge to use the mobile device’s display and keyboard, the questions of security and trust extend to the mobile device. Potential phone evaluations in turn raise the issue of minimizing the size of the target of evaluation, i.e., of the part of the phone that must be trusted. Naturally, one of the solution is to use the phone’s browser as trusted subsystem, and this is a natural way to move to the next generation of Java Card.

Of course, there is a catch. Operators and application providers are still struggling to define business models for NFC, and this could become a major issue is no progress is made rapidly.

The next topic, smart card web servers, was not much discussed, and the link with Bandol is so obvious that I will skip this part.

The last topic, about high-capacity cards, is much more interesting. Operators may be tempted to attract more business into their SIM cards, but providing high-capacity cards also means that they will need to bear the cost of providing memory. Whether or not the customer directly pays for the extra memory, two important problems remain: selling memory is not the natural operator’s job, and the integration of large memory into smart card tokens will always take time, which means that the memory on smart cards will always be at least 30% more expensive then on standard meomry cards. Cedric Nicolas, from Bouygues Telecom, clearly stated his preference for external memory cards, and I must admit that his talk made a lot of sense to me.

The next question is of course to figure out how much memory a Bandol-based SIM card should have. Even if we remove mass memory from the picture, the objective still remains to store many more applications, as well as provide interesting personalization for operators. My feeling is that, although putting gigabytes of data in such a card is futile (as I mentioned before, I don’t want to store my music on my SIM card), a few megabytes could become handy for the storage of a few nice servlets. If we ant the SIM card to become some kind of personal trusted server, we need to have the means to do so.

In addition to this priority list, one additional topic appeared several times: Wavecom has introduced a non-removable UICC in one of their products, and some guys seem to like it, in particular Jean-Gabriel Steinmetz, from PSA Peugeot-Citroen. However, Sergio Cozzolino, from TIM and also Chair of the GSM Association’s smart card group, also had strong feelings about this, and they were definitely not good. The basic idea is that the SIM card is the only thing that the operators own in a phone, and they wan to keep owning it.

Finally, there was something disappointing in this session. The presentation from TIM about Zigbee was very bad, only hype and no facts. I doubt that the guy converted anybody to the technology. The only interesting bit in this presentation is the ability in Zigbee to decide at the beginning of a transaction what the accepted range will be, allowing the same wireless controller to deal with a proximity payment transaction (10cm range) and with a TV remote control application (10m range).