Month: May 2010

Most public Wi-Fi hotspots use no encryption and hence, communication is not very secure. Using a VPN as discussed here and here solves the issue but very few people are actually aware of the problem and willing to take such measures. So far I thought there is little that can be done from the network side as the WPA Pre-Shared Key (PSK) method is ineffective if everybody uses the same key (password) as network monitoring tools can decode the encrypted traffic if the key is known and the authentication and ciphering dialogue is captured. But then I remembered that the University of Vienna offers secure Wi-Fi Internet access so I checked out how they are doing it.

It turns out that they are using individual EAP password authentication from which a Wi-Fi ciphering key (WPA2, AES)  is then calculated. The username and password used in the Wi-Fi authentication process is the student's username and password for the campus network, stored at a central place for all sorts of purposes, including Wi-Fi authentication and encryption. As each student uses individual authentication credentials, monitoring the authentication dialogue will not yield the keys to decode the ciphered traffic later-on. A very elegant solution that just requires support in the Wi-Fi access point for WPA2 enterprise authentication. On the client side, support is already built into the operating system. It's quite clumsy to set-up with Windows XP but with Windows Vista, Windows 7, Linux and Mac OS the configuration is straight forward. It even works with Symbian and Android devices and the iPhone.

The only catch of this solution: The server certificate is not provided, that would have to be done offline, i.e. it's too complicated. That means that the device can't authenticate the network and hence a rouge access point could be used for a man in the middle attack.

Yes, bandwidth requirements are rising, especially when you have a big screen and lots of GHz available for things like high resolution video telephony. I use Skype video telephony quite often these days and when the other end also has a multi-megabit per second uplink available and a good camera, the video quality is just awesome and the stream easily exceeds one megabit per second in each direction. In other words, during a 60 minute video call, over 1 GB of data is exchanged.

Let's compare that to a mobile voice (only) call that uses a 12.2 kbit/s bearer for its codec over the air interface. 2 * 12.2 kbit/s * 60 seconds * 60 minutes / 8 bit = 11 MByte per hour. There's two orders of magnitude of difference here, i.e. a single high quality Skype video call uses the same bandwidth as 100 mobile voice calls! In fixed line networks, voice calls are usually transported in 64 kbit/s channels but the difference is still 1:20. And I imagine video telephony in full-HD resolution is not too far away anymore pushing the numbers even further.

Interesting result from the Dutch 2.6 GHz spectrum auction and one I have difficulties to interpret. Three incumbents and two newcomers have bid for the 2x 70 MHz of spectrum resulting in:

• one newcomer getting 2x 25 MHz
• the second newcomer getting 2x 20 MHz
• two incumbents each getting 2x 10 MHz
• one incumbent getting 2x 5 MHz

Lightreading's Michelle Donegan is the only one on the net I've seen so far writing a meaningful report about it and calling the stunningly low result of €2.6 million paid by the five “some loose change they [the network operators] found down the back of their car seats”.

According to Lightreading, a bandwidth cap was in place to prevent incumbents from bidding for all of the spectrum but I don't quite understand how much that was in practice. In any case the resulting spectrum the incumbents now have in 2.6 GHz seems very strange to me. Having 20 MHz is something you can build a fat carrier with and get speeds far beyond what is possible with HSPA today. But 10 MHz is an awfully small carrier for LTE in this band and I completely fail to see what you do with just 5 MHz!?

Also I haven't seen a country yet where 5 network operators have really made it over time. So instead of fighting it out over an auction, are some companies speculating with a merger down the road to get some fixed line or wireless assets and further spectrum in the 2.6 GHz band? Not sure if the auction rules allow for mergers later on but at least the money lost would be negligible in case the spectrum would have to be returned.

With the spectrum auctions currently ongoing in Germany these days and LTE being the hot topic a number of people have independently asked me recently when I think UMTS will be switched-off. A refreshing variant of the question when GSM will be switched-off. I find the question quite interesting and my answer is that I personally think that UMTS won't go away anytime soon. Having reached almost nationwide coverage in many countries, offering broadband speeds and continuing development ensuring competitiveness, the only reason I can see why to switch it off at some point is to save cost. But until it can be switched-off a number of things have to happen:

• LTE must reach a similar coverage as 3G networks today.

• Most mobile devices requiring a fast mobile and wireless Internet connection have to have LTE built in.

• A voice solution for LTE must be found as falling back to GSM (which is not switched-off either…) for voice calls is from my point of view not a viable option.

So when will those things have fallen into place? I seriously doubt that this will happen within the next 5 years. And once we get there, will there still be a need to switch 3G off or will multi-mode base stations that can generate GSM, UMTS and LTE signals just make it unnecessary?

I see a coexistence of GSM, UMTS and LTE for a very long time to come. So instead of working on phasing out UMTS, it might make more sense to work on solutions to integrate the different radio systems.

As always, comments are welcome!