Mobile World Congress 2011 / 1

MWC 2011 Barcelona, Spain
Day 1: 2/14/11

Samsung Sets the Performance Standard for Phones & Tablets
InvenSense Shows Compelling Motion Platform
Fujitsu Struggles with Dual Screen Phone

Mobile World Congress 2011

Mobile World Congress 2011 Barcelona, Spain

MWC Impressions

This is a massive event which struggles with its size. One is reminded of CeBIT. MWC is an ecosystem conference focused mostly on the devices and operators. Apps are included but nowhere to the level which one would expect given the impact on smart phones. Just like CES it suffers from those who are missing, especially Apple, and also Nokia and Sony Ericsson. There were comments today about creating a 3 ecosystem, based on the Nokia and Microsoft alliance, but there is a huge void in the show floor, when Apple is not present here.

There are 3 conferences:

  • First are the main exhibitors in 5 halls. These are disjointed which makes coverage the same. There are massive exhibits such as Samsung and then those which are aggregates of companies from countries or regions.
  • Second are the many private booths. This is hardly an exhibition in that they are not showing products to the visitors. It is disappointing when companies such as Broadcom have a closed booth.
  • Third are the sessions. These cover a wide range of topics including health and apps.

Just sampling the phones one would think that only smart phones are the market. The low end high volume phone hardly seems of interest. On the other hand tablets are in many places. These are not a carrier-only play but the impact they are having on the overall market has them in many booths. One area of disappointment is the overall lack of the component technologies in the phone. In part, the chip companies are in closed booths. There is nothing we could find about the architecture of the phone, for example.

This is hardly a tablet conference but there was a sharp contrast between the booths that had them and those who did not. In the Samsung, LG and Motorola booths the tablet was just an extension of the mobile platform. The common OS made it all the more compelling as just another device. But when it is missing from the platform offerings there is an obvious hole.

The success of this venue is driven by one market dynamic – the world standard of GSM. All other standards, that is, CDMA, pale in comparison. As a result the success is what has created a 5 billion unit installed base. This then creates MWC which is European in its foundation.


Our first hall entrance landed in the massive Samsung booth. We tried the Galaxy S II phone and the Galaxy Tab. Other devices seen or tested included LG, Motorola, RIM, HP (no touching), HTC and found the Samsung compelling in terms of performance. The experience was fluid as we sailed through the experience. In the case of the Galaxy S II, which is 4.3” AMOLED and dual core processor, this phone stood out in terms of image quality and overall performance.

The Honeycomb-based Galaxy Tab sets the bar for performance also. After just sailing through the interface one really appreciated how well the OS, user experience and hardware were integrated.

We evaluated the Motorola Xoom and found it good but not to the level as this Samsung Galaxy Tab. The user experience seemed the same. It is here where Google has made significant effort that is paying off.

The Nexus S is the first phone with Gingerbread which is available on T-Mobile in the US now. The phone has a built in gyro, full support for VoIP services and 3D games. It uses an AMOLED display. The interface is smooth.


At the other end of the spectrum we evaluated the dual-screen smart phone in their large dedicated building/booth. Actually two designs were shown of a dual screen phone. The first was based on Symbian and it was polished and working. The integration of both the mechanical design and the UI was very good. One could flick content from one screen to another. There is no keyboard and all interaction is on the screen. The second was a concept demonstration based on Android. It was, based on the demo, only an example of what could be done. It showed only two screens in landscape mode on top of each other.

When asked when to market the responses seemed to reflect the current market dynamics. First, Fujitsu is waiting carrier responses to the concept of a dual-screen phone. Having the only working device in Symbian seems to be a poor starting point. However, the response to the Android version was – this will take a LOT of effort to tailor Android for dual screen and to do the mechanical design. We will only do this if we find an operator to sell it to.

We were left with the impression – good technical work but the device was overtaken by market events. This includes the OS in favor, the appeal of dual screens in a large screen smart phone, for example, the Samsung Galaxy S II, and the emergence of tablets. The time for a dual screen phone has passed.


Increasingly navigation on the smart phone is playing an important role. At CES, CEA stated that smart phone navigators are in the process of devastating the personal navigation market. Navigon stated to us that personal navigation is playing an increasing role in the market as individuals are using smart phones for walking navigation.


GMV has taken a software-only approach to GPS signal processing in the product called SRX-10. Basically they are saying that having GPS in a phone is only 200MIPS of processing power. This makes GPS available as a mass market feature on phones. It has been made operational on the XScale PXA270 and the ARM9 Samsung 2440. The TTFF is claimed to be <1min and the best case positional accuracy is <3m. This latter figure is subject to many restrictions that it would seldom be realized in use. They claim in a deep urban environment <10m with 50% and <25m at 90%. Coding to another processor is estimated to take a few months.


They are touting the “World’s Smallest GPS Module” at 7 x 7 x 1.4mm. Up to now the product has been focused on small GPS units such as those to attach to pets. Here at MWC they are seeking an entry into the mobile device market. The key attributes of the technology include: low noise processing, wafer-scale integration of the SiRFstarIV module and low power. Their module supports WAAS but the SBAS equivalent in Europe works poorly in limited areas. Thus, SBAS is only for the US. WAAS has no impact on power consumption.

The significant power consumption in GPS comes during TTFF which can go to 35ma. Once established the baseline current reduces to 5ma. Updates take about 1 sec. Thus, it is important to keep the module fixed all the time. This is one area where the low noise design of OriginGPS has an advantage. WAAS accuracy can get to 2.5m. Cold start to acquire a fix is about 30 sec.


Bordering on the edge of using the more accurate GPS solutions was Telemap. They provide location based services focused on both retail and social. In the booth they were showing what is called Ultra Local Experience. This is information and services down to a street level when an individual is present. When asked if they support Foursquare check in this was dismissed and it is assessed that Facebook will eventually own that market.


InvenSense has the most complete indoors navigation solution seen. Basically it is a strap-down navigation platform on a chip. This includes a 3 axis accelerometer, 3 axis MEMS gyroscope and 3 axis compass. The key to the success of this platform is three fold:

  1. Low noise processing of the signals from the sensors,
  2. Years of experience in the software which does the motion processing from the sensor inputs and
  3. Their own MEMS design.

The total package is only .9mm thick in a vacuum sealed unit. Current draw is 5ma.

The typical problem with a strap-down navigator is drift. This is worse in the lower cost units. Thus, over time, in an indoor setting, the accuracy declines until there is another update. InvenSense addresses this with the inclusion of a compass which will not drift. The problem with this sensor is the noisy sensing of the gravity field. It is here where the gyro plays an important role to damp the estimation of the location. Another limitation of the platform is an accurate estimate of stride length – an important factor when walking. If it is possible to pick up a WiFi signal, even limited information, will allow for a much more precise estimate of stride. As a result the location prediction improves significantly.

It was stated that a miniature laser gyro would improve the position accuracy but InvenSense still believes that the overall performance is based on the software which defines the Motion Processing Platform.

A demo was given on a phone while we walked the show floor. An initial fix was set based on the map of the show floor and the tracking was a very good. Some jumps in position were seen. While on the walk a Quake demo was shown playing the game in real time during the walk – gaming augmented reality.

Initially the company entered the market with its technology to provide motion compensation for camera movement in cell phones. This is the most difficult problem due to the high frequency of the movement of the camera – the cell phone not being a very stable platform. This requirement was stated to need very low noise detection of the motion and a high frequency response. A booth demo illustrated impressive results with a camera module from HYSONIC.

Another demo was shown of a window into a world on the screen of a tablet-like device. There was seamless motion between the device and apparent window. Again, InvenSense credited this with the role of the gyro.

When the market was discussed it was stated that for the last few years it was difficult to convince OEMs and operators the value of a gyro in the phone. This has changed. Currently the InvenSense technology is in some LG phones and many more in the future.

A startup is also in the process of creating a database of inside building maps:

InvenSense has been working with Google to help it with the inside building mapping API. It was stated that Google is well ahead of the industry in this critical database area.

When asked of the market there was no doubt that initially it will be gaming. A motion enabled phone is a new game experience. Eventually this technology will lead to a whole new class of products and services based on accurate indoor navigation.

More information on the platform can be found on YouTube video of a talk given at Google Headquarters by David Sachs of InvenSense.

> Sensor Fusion on Android Devices by David Sachs (Google TechTalk)

Here is an overview of the InvenSense technology.

  • Motion processing is much more than sensors. We learned this over 7 years.
  • The processing rate to create real time smooth output is 200Hz. This creates significant burden on the CPU. Our motion AP takes this burden and the application only needs to sample the location at a rate it needs.
  • Chip integration is essential. There are 4 chips with STMicroelectronics product but ours is one integrated package. Both the gyro and accelerometers must be in a vacuum seal. A single package is best in terms of cost and size.
  • The key is the software in the application processor – what we call the motion processor. We have support from Google but it is very cautious about showing preference to any one company. Our product is supported from Gingerbread on.
  • There are many issues to be addressed in getting a motion platform to work. One is calibration of the sensors. In conventional internal platforms this is a significant effort but in CE such activities cannot be tolerated from a cost standpoint. For example, every motion platform needs to be calibrated based on the chassis they are in because this influences the magnetic field around the compass. Out platform is self-calibrating.
  • There are many subtleties in the support of an inertial platform in a phone, for example. One is body placement. Two good places are the pocket or the shoe but, in large part, these are impractical. The reason is the motion is predictable do to body motion. BUT the worst place is the hand because the hand is decoupled from the rest of the body motion. However, given there are some constraints, such as readability, this provides clues which we use to determine location.

InvenSense has thought much about making location work on a device in closed physical environments.

Next article –> Read MWC 2011 Day 2

This entry was posted in Analysis, Events, Markets, News, Technology and tagged , , , , , , , , , . Bookmark the permalink.