Feeds:
Posts
Comments

Archive for the ‘Market Watch’ Category

Yesterday at MoMo Bangalore’s November event, it was yet another session of sensible arguments, discussions and insights which I have come to believe characterize every MoMo event. In this community, ideas are shared openly. Suggestions are welcomed and analyzed. In a world in which knowledge is power, the people of MoMo community realize there is a greater power in sharing knowledge. Yesterday’s session had a high level of interactivity and this in part must be credited to Deepak Srinivasan, the CEO of Mobiance Technologies, who opened the floor for Q&A from the outset of his presentation.

Mobiance is in the business of mobile Location Based Services (LBS). Unlike the usual method of using GPS for delivering LBS, Mobiance has adopted a different approach based on a patent pending technology. The technology involves locating the mobile using cell triangulation. In other words, a mobile’s location can be identified by it’s distance from at least three base stations. This is done using Timing Advance. I believe (the discussions did not get into details of the technology), no extra signalling needs to be generated on the air interface for this purpose because such data is usually readily available with the network. All signalling is at SS7 level between Mobiance system and MSC, BSS and OMC. Data from the signalling network is filtered and processed intelligently to keep Mobiance generated traffic to a minimum.

Coming from an engineering background, I know that none of this sounds difficult. In fact, there is nothing to be solved here. GSM as a standard has evolved over a few decades. It has solved the problems of mobility management to a large extent. The value that Mobiance brings is in applying the fundamentals effectively for services that matter and marketing them at the right level. For example, Mobiance claims that their method of using timing advance brings greater accuracy. In Metros, an accuracy of 150-200 m is possible; 250 m in other cities that include Bangalore, Ahmedabad and Hyderabad; 2.5-3 km in rural areas and highways. The latency of locating a mobile is only 3 seconds. Locations can be updated once a minute or once in fifteen minutes or even once in an hour. This flexibility is in-built into their platform that allows configurability to administer QoS policies. Some solutions may require only a location displayed as a text; others may require the location displayed on a map that delineate taluk boundaries. Such flexibility of solution delivery exists with Mobiance. The technology does not need changes to base stations or mobiles. Even a low-end handset can get services enabled by this technology.

What are the services that can be deployed on this platform? Most of us will be familiar with services for consumers. These are generally services available in the vicinity – ATMs, restaurants, theatres, internet cafes. Then there are navigation services that tell drivers/pedestrians how to get from point A to point B. For enterprises, LBS can be used to deliver fleet management systems or verifying if their distribution network is working as planned.

The company was started in October 2004. In March 2007, they signed a contract with Airtel. In July 2007, their Enterprise LBS solution was launched. The challenge for Mobiance is to bring in more services, use more accurate maps and GIS data, partner with more cellular operators and continue to build on the partnerships that they currently have. As I listened to the presentation with interest, I can attribute their success to these key factors:

  1. Local solutions for local needs: India has many languages. People communicate differently. Directions are given differently – “next to”, “opposite to”. Any LBS enabled service should consider the cultural environment in which it is deployed.
  2. A model of effective partnership: Mobiance has recognized that solutions that work in a Western context may not work in India where partnerships are important. While in the West one may license Navteq maps, in an Indian context what is needed is a strong ecosystem of providers, partners and users in which everyone benefits and progresses in tandem. Thus Mobiance partners with device vendors, cellular operators, GIS (Geographic Information System) and map providers, and the like.
  3. Do what you do best: Mobiance has created the platform for LBS. This platform can be used by partners in developing services. Mobiance is not into creating these services. They concentrate on getting the platform right leaving the creation of services to others. Their platform API is currently based on XML-RPC.
  4. Initial focus: one common reason why startups fail is that they try to do too many things from the start. While the host of services that LBS can enable is many, Mobiance is focusing only on its enterprise solutions (fleet management, verification, awareness). A success in these can lead to more services in various other segments.

Read Full Post »

Mobile Monday is a global community with various chapters all over the world. What does this community do? This is what they say of themselves:

Mobile Monday is a global community of mobile industry visionaries, developers and influentials fostering cooperation and cross-border business development through virtual and live networking events to share ideas, best practices and trends from global markets.

As often is the case, Bangalore is not far behind in whatever happens at world stage. A group of enthusiasts have combined forces to start the Bangalore chapter of Mobile Monday. This happened in June 2006. Their meetings are fairly informal and full of ideas, discussions and business networking.

As I understand, they meet on the fourth Monday of each month at various venues. I attended the October meet. A roundup of this meeting is on their website. My own review of the meeting follows. The meeting started with a presentation by Vanu India Pvt Ltd, a global startup expanding in India. This was followed by a demo of QuillPad by Tachyon Technologies, a small Indian startup.

Presentation by Vanu India Pvt Ltd

Vanu India Pvt Ltd is a sister company of Vanu Inc., a global startup founded in 1998 and focusing on Software Defined Radio (SDR). SDR is a technology in which some or all of physical layer processing is done in portable software rather than custom-built hardware. This enables base stations and access points to activate different access mechanisms from the same platform with ease of mainenance and operability. The interesting aspect of this software is that it runs on a general purpose processor rather than a DSP. This makes Vanu’s approach disruptive to existing market norms. Vanu’s advantage is that as processors get better and faster, the same implementation can be ported with little difficulty to leverage on the latest advancements.

Figure 1

Waveform-Device-Vanu

This can be better understood by referring to Figure 1. The top flow is a traditional model. The bottom flow is what Vanu is following. Traditionally, it used to be the case that the average lifespan of a device is much longer than the waveform model. This was only because implementations in software used to be easier to change. This is no longer true as software modules are large and complex. Software these days is doing a lot of what hardware used to do, particularly in the domain of SDR. This means that if a new device is released much of that software has to be rewritten to take advantage of the advances in the device. The cost of developing new software or updating old software is quite high. Why not write a portable software that needs no update?

For this approach to work, General Purpose Processors (GPP) need to keep pace with advancements in DSPs. Only then, Vanu’s approach will have a competitive advantage.

SDR is presently not practical on a mobile but Vanu sees this as a goal for the future. Current technology requires that the RF frontend be duplicated for different access standards. The business case for Vanu is also not obvious at present but there is much promise in an evolving market where multiple standards will continue to exist and interwork.

The many insightful questions that were asked during the presentation underlined the uncertainty that’s present in an evolving market. Only those who innovate boldly and find an apt business case will succeed. Vanu appears to be doing just that, although there are sufficient challenges to keep them busy.

Demo by Tachyon Technologies

Representation of Indian languages in the digital domain has made much progress in the last decade or so. We now have unicode representations. We have many types of fonts. We have websites purely in Indian languages. However, methods of input have failed to catch up with these advances. Or rather, what we have today are sorely inadequate to fulfill the expectations of users. Anyone who has attempted to type Indian characters will know the difficulty of the exercise. The challenge is greater if we attempt to do this from a mobile keypad.

Tachyon aims to make such input a lot easier and intuitive. The ordinary way of Indian language input is to transliterate Roman characters to Indian characters using predefined rules. Because there are many more alphabets in Indian languages than in English, the methods of input are often non-intuitive and sometimes involve special keyboard characters. Tachyon has taken a different approach of using a trained data set of common patterns that occur in words. This enables users to type more naturally. The language interpreter takes on the job of finding the right word based on context. If there are multiple options, the most likely one will be suggested. It is also easy to navigate across the options. This make Tachyon’s input tool, the QuillPad, a more user-friendly tool than Google’s Indic Transliteration. Tachyon has taken the competition head-on. They appear to be in the lead.

The demo at the meeting was on a Nokia handset loaded with QuillPad software. Before coming to this meeting, I tried the online QuillPad for Tamil. The product is by no means done and dusted. It needs improvements. For example, try typing “Thekkady”. It makes the same mistakes as Google. Hopefully, Tachyon will win the race.

Read Full Post »

I attended this event recently. It was a whole-day event at Le Meridien, Bangalore. Registration was free. Apparently, ARM has been organising this every year for the last 2-3 years. This year it was held in three places in India: Bangalore, Hyderabad and Noida. It had no direct relevance to me because I am more into telecom software which is not what ARM does. Nonetheless, I found it extremely useful because the divide between software and hardware is not B&W these days. Things done in software could be done better in hardware. Things done in DSP could be done better in the MCU. Some factors influencing such decisions include the processing power, parallelism, multiple core architecture and others which are shaping the future generation of ARM processors.

Briefly, ARM operates in four key market segments: embedded, enterprise, home and mobile. Across these, there are four verticals: Processor IP, Physical IP, System Design, Media Processing and Services. This was enlightening for me as I learnt that ARM does a lot more than what I had previously known. They are not just an IP provider for RTL but provide a whole ecosystem of IP and services to realize System-on-Chip (SoC) quickly. They provide system IP around the ARM core. They provide software IP in terms of libraries. They provide a whole suite of integrated tools to reduce the time to market.

Just as importantly, they have initiated and manage the ARM connected community. Just like open communities on the Web, developers share knowledge, problems and solutions. Technical symposiums as this are regularly organized across the globe. ARM presence in India started in 1996 and it has grown substantially in the last three years. In the next 4-5 years, ARM intends to grow to over 1400 employees in India and 50% of their R&D will be done from India.

In the mobile space, there is plenty of growth potential for ARM. ARM processor architecture is one of the main components in many mobile systems. As an example, Nokia N95 uses two ARM processors: ARM1136 and ARM9. This phone is deemed to be one of the best selling phones in Europe and ARM sees it as the first great success story for ARM11. The market for smart phones is expected to triple by 2010. Another example: Apple’s iPod Nano uses Dual ARM7TDMI processor.

For long ARM7TDMI has been popular. Although ARM9 has been around for sometime now, only recently it has overtaken ARM7 in terms of number of licenses. ARM11 is just taking off; it is likely to be the chosen core for multimedia processing. The next generation of processors are already out – the Cortex family. Figure 1 illustrates these.

Figure 1
ARM Families

What this means is that processors and their platforms are trend setters. Every new generation of processors spawns a new generation of interfaces, tools, operating systems, applications, and user expectations. If today we were still using Intel 8086, the evolution of Windows OS might have been quite different. If ARM’s best processor core were still ARM7TDMI, Nokia N95 would have designed very differently with different power consumption and real-time performance. Being ahead of the race, processors define what’s possible. Everything around it tries to “upgrade itself”.

The next thing I found interesting was Thumb-2. I remember the time when engineers needed to decide what code should be compiled to Thumb and what else to ARM instruction set. What these engineers were attempting to do was to get a balance between code density and performance. This trade-off was linear. Thumb-2 breaks this linearity. It achieves a performance comparable to ARM and a code density comparable to Thumb. It is difficult to explain how this has been achieved without going into the instructions themselves. Hopefully, I will make a separate post. Briefly, Thumb-2 uses Thumb’s 16-bit instructions and extends to 32-bit Thumb-2 instructions. Figure 2 is a summary of what has been described.

Figure 2
Thumb-2 performance vs code density

There were lots more stuff at the Symposium. Some briefs are below:

  • Long ago I used ARMulator to do some profiling. RealView is infinitely better.
  • The future will see multiple wireless standards implemented on a single chip. For example, WiFi, WiMax, Bluetooth could all be on the same SoC. My own take on this is that software radio is a formidable competitor.
  • ARM was thought unsuitable for Internet browsing. ARM was benchmarked against Intel. A presentation slide showed comparable performance.
  • For MCU, 32-bit architectures are set to grow rapidly in the coming years.
  • In ARM, 30% of their turnover goes into R&D.
  • For microcontrollers, Cortex-M3 is likely to take over from ARM7TDMI.
  • AMBA4 is currently in design stage.

Relevant stuff (which you could search online & which I need to research):

  • Movial: open source stack to enable applications on mobile platforms
  • SOI Industry Consortium
  • ARM forum
  • Mali Graphics Processor and OpenGLES
  • NXP Software: founded by Philips: has access to Philips’ patents: claims that its software has been included in more than 250 million phones: they gave excellent review for ARM’s RealView profiler
  • AAPCS
  • EEMBC: a non-standard guideline for benchmarking processors
  • CELF
  • LiPS: Linux for mobiles

Two ways in which you can contribute on the above items:

  1. You may add your comments to this post.
  2. If you want to write a detailed article, you can post your views here on Mobile & Wireless. I will add you to my blog either as an author (you write the complete article) or as a contributor (we collaborate on the article). You will need to be registered within WordPress.com

Read Full Post »

« Newer Posts