Author: Martin

Back in 2009, I first reported on a growing trend of network operators outsourcing their network operation to third party companies, thus in effect becoming "Hollow Operators". The main drive for some network operators to doing so was to reduce their costs. With companies like Alcatel Lucent, Ericsson, Huawei, NSN and others eager to pick up such deals, competition must have brought down prices quite a bit, so perhaps they even got that.

But now it seems that many of these deals have not been very profitable for some of the service companies. Light Reading reports that Alcatel Lucent wants to get rid of or re-negotiate 25% of their managed service deals as they are not profitable. Other companies also seem to have a difficult time with some of their service contracts as the report notes.

By now, however, it's likely that a significant amount of local talent previously operating a network is gone, long having been replaced by supposedly cheaper labor in far away countries. Best of luck to those hollow operators who are now faced to re-integrate their network operation or to negotiate a service deal with someone else who will not take over resources not working locally anymore. Another choice is to obviously pay a higher price for ALU to continue the operation.

I wonder how that new price now compares with how much network operation cost while it was still done in-house including the speed, flexibility and control the company had while doing so!? It was not too long ago, so perahaps someone will still remember and have a metric or two.

Contrary to popular belief that voice and SMS use is diminishing in mobile networks, Vodafone Germany has posted some interesting quarterly results. Here are some highlights:

Growing Use of Voice Calling

35.8 Million users accumulated 14.25 billion (!) voice minutes in 3 months. That's 132 minutes per month per user on average. And that number is not declining, it is actually 1.16 billion minutes higher from the same quarter a year ago, or +8.9%. Voice calling in mobile networks might become cheaper all the time but definitely not less used. Other text based communication forms might become more popular and replace voice calling in some instances but I think these numbers make it quite clear that new forms of communication are still mostly adding to traditional voice calls and texting rather than replacing it.

Rising Use of Texting

Yes, there is WhatsApp, iMessage, Google Talk, etc. etc. but the number of SMS messages still keeps growing. The press report mentions a growth of 8.9%. It's not quite clear to me if this refers to revenue growth or an increase in number of messages being exchanged. But since SMS texting is not getting more expensive the number is likely to contain both. In other words no reason to be ashamed or feeling old fashioned for preferring SMS texting to WhatsApp anc Co. 🙂

Increasing use of Smartphones and Mobile Data

A couple of interesting data points here: 80% of mobile devices being sold with a contract this quarter are smartphones. On the prepaid side, probably not one of Vodafone's strengths as a premium network, smartphone sales are 1 in 3 at the moment.

Data excluding SMS and MMS now generated 28.5% of overall mobile revenue. To set this into perspective: Total voice + texting revenue was 1234 million Euros, data accounted for 492 million Euros, up from 411 million Euros (or 19.6%) from the previous year. That number is interesting as it shows that revenue from voice and text has been pretty much flat, slightly down by 1 million Euros from 1235 million Euros a year ago.

Recently, I bought a radio so I could listen to my favorite radio stations in the kitchen. It's not an FM radio, of course as my favorite radio stations are not broadcast over the air and hey, the world has moved on that's too low tech these days anyway. No, it had to be a Wi-Fi web radio to connect me to radio stations in Italy so the Spaghetti become extra 'al dente' and to stations in the US for the fries to get extra crispy.

While the radio works well and sounds good there is one major disadvantage: When I switch it on it takes 20-25 seconds before I hear the first music. That's how long it takes for the radio to boot, to connect to the Wi-Fi access point, open a radio stream and buffer a few seconds. Hm, I would have almost accepted this, but then I remembered that my smartphone and tablet remain connected to my local Wi-Fi even while in standby mode with little power consumption and are instantly fully connected again when unlocking them. In other words, from a technical point of view there are tons of devices out there that show that it can be done a lot better.

The initial delay also doesn't compare very favorably to an FM radio that works instantly when switched on. That reminds me a bit of how it must have been in the times when vacuum tube radios had their high times. At least switching between stations is fast enough with a delay of around 2-3 seconds.

Do you have a Wi-Fi connected web radio? If so, how long does it take yours to connect to the network and a radio station before it starts playing when you switch it on?

Intomobile reports that experts from a security firm are estimating that 67.000 phones to be stolen during the Olympics, based on 50.000 phones that, according to them, are stolen every two weeks in London without the Olympics. That number sounds incredibly high, so I did some research myself.

Just to set things into perspective, the 50.000 lost or stolen phones in London over a two week period, that's 1.3 million phones over the year. There are 8.7 Londoners living in the city, and lets add 1 million travelers and round it up, so let's say 10 million people are in London on average. That means that on average every single person in London loses a phone once in 8 years. Hm, I am skeptical.

The Metropolitan Police also has some stats. They are saying that around 10.000 phones are stolen per month. That would be  120.000 phones a year, or less than a tenth of what is claimed above. When taking this as a basis the resulting number of phones that might be stolen during the Olympics would be about 6.200. Far from the number above, but still a gigantic number.

Back in May I ran a couple of posts on Dual-SIM phones to see how they work in practice (see here, here and here). The secondary purpose of the exercise was to give the phone afterwards to someone using several phones simultaneously due to frequent traveling to different countries and friends in those countries preferring to call and text local numbers.

So while the phone served the purpose of reducing the number of simultaneous phones to be carried very well it had one major disadvantage: No QUERTY hardware keyboard. I was hoping that the dual-SIM capability in addition to Android and the possibility of Internet access and apps would compensate for the inferior text input capabilities of the on-screen keyboard. But it turned out that it didn't.

Even after two months I still got complaints of how cumbersome it was to use the virtual keyboard and calls were frequently missed due to Android's feature of setting the loudness level of the ring tone in idle mode via the buttons on the side of the phone which were often inadvertently pressed.

Back to two phones (+ a third one) then, the QUERTY hardware keyboard of the old phones have prevailed over dual-SIM. Which shows that there is a market besides the main stream where usability is compromised for a bigger screen and a slimness. Hm, that makes me think of my good old N95 and how fast I was with T9 text input and navigating on web pages on the dial pad. Sigh.

Observation of the day: While the trend of people using their daily commute time for reading news and checking their Facebook news on their newly acquired smartphones and tablets keeps increasing I also noticed that there are quite a number of people using their somewhat older Nokia phones to browse the web. Note that I am not talking about those previously high end phones such as the N96 etc. No, I am talking about those 3 year entry level Symbian phones such as the Nokia 5230.

I am pretty sure that when people bought those phones 2 or 3 years ago, it was not with Internet access in mind. Those were not the tech savvy people at the time that would have done that when it was still a nerd thing. But still, for whatever reason, perhaps because they don't want to invest money for a new smartphone but still access their Facebook messages on to go, they've opted for an additional data tariff on their current prepaid or postpaid subscription and have found out that their 'old' phone is capable of doing a lot more than what they've bought it for initially.

For me that shows that Internet access on the go has really reached the masses.

A propeller head entry today for Linux hackers that is wireless network related: As I'm probably not the only one on this planet considering using Ubuntu 12.04 and connecting it to the Internet over a 3G dongle and uses OpenVPN over the link here's a quick tip how to overcome a little flaw in the network manager that prevents the proper setup of DNS IP addresses when the VPN tunnel is established. Let's hope Google will help you to find this page when you need it 🙂

In the previous Long Term Support version, DNS configuration was dead simple: /etc/resolv.conf contained the IP addresses of the DNS servers to be used and the file was updated after the 3G dialup connection was established and once again after the OpenVPN tunnel came up as it uses different DNS servers that go through the tunnel.

Unfortunately in 12.04 LTS this has considerably changed: /etc/resolv.conf is now a link to /run/resolvconf/resolv.conf. The DNS reference in this file is now always 127.0.0.1 because Ubuntu now uses dnsmasq to reach out and resolv DNS queries instead of the standard network mechanism. The real configuration file that is populated with DNS IP addresses when the Wi-Fi or OpenVPN interface comes up is /var/run/nm-dns-dnsmasq.conf.

The problem comes in when a 3G dialup session is established. It seems to be a bug but the DNS servers are not put into the dnsmasq config file but into resolv.config before the reference to 127.0.0.1. In other words, dnsmasq is not used for 3G dialup or only when the 3G dialup session DNS servers fail.

This is no problem until the point when an OpenVPN tunnel is established via the network manager, which puts its DNS server ip addresses into the dnsmasq config file. As the 3G connection has modified resolv.config these DNS servers are only hit once no answer is received from the DNS servers of the 3G dialup, which are not reachable through the VPN tunnel. Result: DNS queries have to timeout on the 3G dialup DNS servers which do not answer. Then 127.0.0.1 is hit in resolf.conv which takes several seconds which invokes dnsmasq which finally asks the DNS servers that can be reached over the VPN tunnel. In practice that is pretty much standstill. Somebody must have overlooked that unfortunately.

And here's a fix until somebody in the Ubuntu or network manager team figures this out and provides a real patch:

• Create resolv.conf-overwrite in the root directory that only contains 127.0.0.1 as the nameserver IP address
• Create a script with filename 99copyresolvconf in /etc/NetworkManager/dispatcher.d with the following content:

#!/bin/sh 
cp /resolv.conf-overwrite /var/run/resolvconf/resolv.conf

• Ensure to do a chmod +x to make the file executable!

Now everytime a network interface comes up the resolv.conf file is overwritten with the default one that only contains 127.0.0.1. It's ugly but it works..

For those of you who are interested in the hardware of one of the latest high end phones, the Samsung Galaxy S-III, here are two interesting resources today if you haven't seen them already: Chipworks and iFixIt have pictures and descriptions of what's inside here (for textual descriptions and zoom in images) and here (for high-resolution overall images). Interesting points they discuss:

• The Samsung Exynos (ARM) application processor also includes the RAM, i.e. memory no longer requires a separate chip.
• The baseband (radio) chip is made by Intel (Infineon). According to Chipworks its the same chip already used in the SGS-II.
• A 16GB Flash memory chip. Interesting to note here is that the chip is bigger than the application processor chip as can be seen in one of the images.
• A separate Wi-FI chip and a separate GPS chip.
• 3D accelerometer and gyroscope are in one chip, the electronic compass in another one. Downright incredible, a 3D gyroscope in a small chip.
• An NFC chip by NXP.
• There is a specialized audio chip of Audience to enhance voice call quality.
• And there's a dedicated chip for other audio applications (music?) from Wolfson. They also claim in their press release that their chip can do voice call quality enhancements. I wonder how that comes together with the Audience chip!?

Chipworks mentions that as soon as the North American LTE variant with a Qualcomm baseband is available they'll do a similar exercise. It's going to be interesting if they'll just replace the Intel basband or use a Qualcomm SoC solution that includes the baseband, the application processor and perhaps a couple of other functions that are on separate chips in this variant.

Hat tip to Stefan Constantinescu over at IntoMobile.

Today I noticed that the company producing the local phone book and Yellow Pages has started distributing the latest edition of the 'paper version' to all households. I can't but wonder in how many households it will actually be used and in how many it will just be thrown away because smartphones and PCs have taken over the job!? How many more 'paper versions' will there be before it becomes economically non viable anymore!? I guess it will be rather sooner than later with smartphones now accounting for over half the mobile devices sold in Germany and other countries.