Thursday, March 27, 2008

MHEG-5 headed for India

MHEG-5 is an open standard middleware solution, an application program interface (API), designed specifically for low-cost memory constrained devices -- particularly suitable for digital interactive TV (iTV) services and is platform agnostic. Simply put, MHEG-5 is a simple object-orientated programming language.

MHEG-5 was initially adopted in 1998 by ONDigital in the United Kingdom (rebranded as ITV Digital) for use in the world’s first pay-TV digital terrestrial television (DTT) network. ONDigital lead the industry wide specification and development task of the first MHEG-5 profile, which subsequently (after the collapse of ITVDigital) formed the basis for deployment of interactive services on the UK DTT platform (Freeview).

Developed by the ISO-MHEG group and DAVIC in 1995, it is intended as a UI for DAVIC interactive services and VOD. MHEG-5 has been standardized in ISO 13522-5 and adopted by the UK DTG in 1997.

Going by its history on IMPALA (The International MHEG Promotion Alliance), MHEG (Multimedia and Hypermedia information coding Expert Group) originally developed and standardized by Working Group 12 (WG12) of the ISO -- officially known as ISO/IEC JTC1/SC29/WG12. It was developed in the mid 1990s as part of the DAVIC (Digital Audio Video Council) standardization effort to support interactivity and navigation of multimedia services on various small footprint devices.

WG12 issued a suite of documents (MHEG parts 1-8) as part of MHEG standard covering extensions for scripting language (MHEG-6), testing and interoperability (MHEG-7) and support for encodings in XML format (MHEG-8). Part 5 of the standard, officially known an ISO/IEC 13522-5, or more commonly know as MHEG-5 that is of primary relevance to interactive DTV.

So what's hot about MHEG-5? Well, the standard's profile evolved to UK Profile 1.06 (current); ETSI standard ES 202184. And now, new international profiles extend the UK profile. These include: New Zealand -- extra Maori characters and EPG key; Hong Kong -- Traditional Chinese font; Singapore/China - simplified Chinese font. Certain other extensions are said to be under development within DTG for possible deployment in 2008. These include IP interaction channel (return path), improved graphics, HD compatibility and support, and PVR support.

Heading for India?
What's more significant is that MHEG-5 is said to be launching in India and in Hong Kong in 2008! Trials and evaluations are reportedly ongoing in Ireland, Malaysia, Singapore, Turkey, India and Russia. There is said to be interest from other countries in Europe and Asia as well.

MHEG-5 has no known essential IPR. The MHEG middleware software is typically less than US$1 per receiver. It has had wide integration into iDTVs in Europe. Finally, MHEG is proposed as the UI for new Common Interface spec (CI+), as well.

I hope to be speaking with IMPALA sometime soon about its plans for India and update you appropriately!

Friday, March 21, 2008

Will the solar bubble really burst in 2009?

Will this actually happen? Will the solar bubble burst in 2009 as supply exceeds demand? I don't know, but I came across this news and am pleased to share with you.

According to Lux Research, the overall solar industry revenues will grow to $70.9 billion in 2012, but oversupply and new technologies will squeeze today’s leaders! This has been reported in its new report titled "Solar State of the Market Q1 2008: The End of the Beginning."

As per Lux Research's study, the solar industry has been having a remarkable run, attracting the attention of all stakeholders. However, all of this activity has also led to the growth of a bubble, which will most likely burst some time soon.

Lux Research says that while growth will continue to be robust -- solar industry revenues will grow at a brisk 27 percent annual rate to reach $70.9 billion in 2012, up from $21.2 billion in 2007 -- the solar industry will look very different just two years from now!

"Government subsidies in countries like Japan, Germany, and Spain have helped make large-scale solar a reality, with annual installations reaching 3.43GW in 2007," said Lux Research Senior Analyst Ted Sullivan, the report’s lead author. "During this period, solar demand has consistently outpaced supply. But the market is now approaching a tipping point. We project that the supply of solar modules will exceed demand in 2009, leading to falling prices and a shake-out among companies that aren’t prepared to thrive in this new environment -– particularly crystalline silicon players that haven’t invested in new thin-film technologies."

Interesting comments! At the recently concluded ISA's Vision Summit, it was disclosed by Jairam Ramesh, union minister of state for Commerce, that the Indian government had received seven confirmed investments in the Hyderabad Fab city, with a total value of $7 billion for 10 years.

Five firms have been given principle approval with $1 billion investment. Proposals of three other firms -- Videocon, Moser Baer, and Hindustan Semicon Manufacturing Corporation (HSMC), have yet to be considered.

Reliance is said to have put forward a proposal for solar/PV manufacturing facility with an investment of $5 billion in Jamnanagar. This proposal is under consideration. The majority of firms proposed to set up in Fab city are mainly focused on solar/PV.

As I said in that blog, the last bit is the interesting part. If everyone focuses on solar/PV, who will focus on the other ancillaries required to complete the semicon ecosystem? What about the LCD, OLED plants, etc., that were mentioned in the semiconductor policy? Now, with the Lux Research report predicting a squeeze next year, these Indian companies investing in solar/PV fabs need to be careful about how they go about setting up their businesses.

The solar business will surely be a space worth watching over the coming years!

Thursday, March 20, 2008

Top 25 semicon vendors of 2007

Here are the top 25 global suppliers of semiconductors, according to iSuppli. First up, there are no surprises in the top 5 -- Intel, Samsung, Texas Instruments, Toshiba and STMicroelectronics retain their spots for this year too. The surprises occur in the second rung -- or, in the next five spots.

Renasas and Hynix exchanged places, with Hynix moving up from 7th position in 2006 to 6th position in 2007, and Renasas dropping from 6th last year to 7th. This is very interesting, because, despite memory market pains during 2007, South Korea's Hynix Semiconductor and Japan's Toshiba and Elpida Memory achieved memory-chip revenue growth of 15, 14.5 and 8.8 percent respectively in 2007, as per iSuppli.

Infineon, Sony major movers
The next three positions are the major surprises of the year. Well, the 10th position was no surprise to me -- AMD, dropping from 8th in 2006 to 10th in 2007. Sony and Germany's Infineon Technologies have been the biggest gainers of the year!

According to iSuppli, Infineon acquired TI's DSL CPE chip business and its wireless baseband semiconductor unit, boosting its revenue. Qimonda, which spun off Infineon, dropped from 12th in 2006 to 16th this year. This split had seen Infineon go out of the top 10 last year.

As per iSuppli, logic application specific integrated circuits (application specific standard products and ASICs) enjoyed the strongest performance among all semicon segments in 2007. Sony and Toshiba were key drivers of growth in this segment due to their sales of semiconductors for the PS3.

Fabless is surely in
The presence of Qualcomm and nVidia in the top 25 list speaks volumes of the power of fabless companies. Qualcomm moved up from 16th to 13th position this year, while nVidia moved up from 25th to the 20th position this year. There is every chance that we will see a fabless company in the top 10 next year! There is an even better chance that more fabless companies will make it to the top 25 companies next year and in future.

All other key players dropped in their rankings. NXP dropped from 9th to 11th; NEC dropped from 11th to 12th; Freescale from 10th to 14th; Micron from 13th to 15th; Elpida moved up from 19th to 17th; while Matsushita and Broadcom dropped a place each.

An iSuppli release says: "Overall, the top 25 semiconductor suppliers significantly outperformed the combined performance of companies ranked lower than them in 2007. The Top-25 as a group achieved revenue growth of 4.5 percent in 2007 while the combined growth of all other semiconductor suppliers was only 0.8 percent."

On a personal note, I would love to see names like SemIndia and HSMC making it to the list. If not now, then at least sometime in the near future. However, it seems from certain published reports that the Indian fab story has gone all wrong. I'll take up this topic in a future blog for sure!

Saturday, March 15, 2008

NXP India achieves RF CMOS in single chip

NXP Semiconductors India has developed the PNX4902, an ultra low-cost GSM/GPRS single chip, which was announced this February. The highlight -- the entire analog and RF work done has been in Bangalore! You might wonder what's so unique about this!

Well, let's start with what is tough about RF CMOS in single chip! CMOS is primarily a digital process. The analog circuit design in CMOS is tough, and the RF circuit design in CMOS is even tougher. Now, the co-existence of RF CMOS circuits with noisy digital in a single chip was (and is) considered the holy grail of chip design.

Next, cellular standards (such as GSM, EDGE) and specs are much tougher than other comparable standards like FM, Bluetooth, etc. Also, some key cellular parameters like RX sensitivity become tougher for single chips aimed at emerging markets. Especially, we all know that base stations are sparse in rural areas. Taking all of these as a whole -- RF design in presence of digital noise is the biggest challenge in a single chip!

Factors enabling single chip design
There are said to be three factors. One, RF CMOS is the high quality analog/RF design in CMOS. The high-performance RF blocks like LNA, mixers, etc., used to be the domain of BiCMOS, a higher cost technology. Next, fine-line CMOS (0.18mm and lower) provide high fT and lower noise.

Two, there are new architectures that minimize analog signal processing. Chip designers to convert the analog signal to digital -- so they might as well do it early -- analog-to-digital conversion at the IF, instead of at DC. There's also a need to move the final down-conversion and filtering into digital domain.

Three, the use of DSP to calibrate the analog performance. Things like temperature and process sensitivities in analog circuits need adjustments. Also, the digital engines can provide the ability to 'lock-in' the performance. Finally, a strong 'engineering culture' is a MUST to execute on complex chips.

Factors enabling AeroFone single chip design
NXP had acquired Silicon Labs Wireless group in 2007. Silicon Labs was a leader in RF CMOS, and so it also acquired numerous patents and trade secrets. Trade secrets for integration of RF CMOS circuits with noisy digital provide an edge over competitors as the integration intensifies.

Thereafter, NXP went on to form the NXP India single-chip design team. As single chip products are designed for emerging economies, NXP India invested heavily to develop the design expertise in Bangalore. The seed group of chip leads and system leads relocated from USA to India to start an analog/RF competency center for developing highly integrated chips.

The NXP India single chip design team has the vision to be the best center of competence in architecture and design of highly integrated circuits (ICs) for emerging market products. It is building the best analog/RF group in India.

Wednesday, March 12, 2008

VLSI as a career in India

It is a great pleasure to find people, most of whom I don't even know, either leave comments on my blog, or even leave a post asking for help. First of all, I would like to thank all of those who read my blog posts. Two, I would like to thank a reader, Paul, for his best wishes pertaining to my recovery. Thank you friend!

I would also like to address a query left by a young reader R. Shrivatsan. It is my pleasure to let you all know that he is going to do a PG course in VLSI in Bangalore. There are lot of opportunities in the VLSI space in the country, and it would do India a lot of good if more such youngsters took to this line. India needs people with experience in VLSI!

I am not a career consultant, nor am I qualified enough to give VLSI tips to people. However, I will try and list some sites who can offer that kind of guidance to people interested in VLSI as a career.

A very good place to start looking for VLSI related information is the VLSI Society of India. This is the site that covers all that is related to VLSI. There are some related sites as well, and I am sure it would help those interested in pursuing VLSI as a career. The India Semiconductor Association is also a good place to ask around for VLSI related information.

I'd like to add that the VLSI Society of India signed an MoU with the Visweswaraya Technological University (Belgaum, Karnataka) and the India Semiconductor Association in June 2005 to spread and improve the quality of VLSI education. Under the aegis of this MoU, colleges that offer M.Tech programs in VLSI Design and Embedded Systems have been identified as centers where champions from the industry will work in close cooperation with faculty champions to improve the quality of M.Tech programs.

I also came across certain sites offering some valuable information. The VLSI Chip Design site lists some of the leading ASIC and VLSI design companies in India. Then, there is this nice little blog -- India VLSI -- which also has some good indications as far as VLSI companies in India are concerned.

Next, there's a nice site -- VLSI Chip Design -- which also offers lots of answers, as well as news, jobs and views about the semiconductor industry in India.

Further, there a host of very good verification companies in India, such as EVE, Tessolve, etc. Lot of work is going on in this space as well. I also came across EDA 101 -- a new community for electronic design engineers, managers, students and consultants whose career includes use or management of Electronic Design Automation (EDA) tools in modern digital and analog design.

There is lot of information going around on opportunities in VLSI in India. My best wishes to Shrivatsan, and hope that India gets a great VLSI engineer.

Lastly, I would like to invite knowledgeable people to send me links, etc. on any good sites, education institutes in India, pertaining to VLSI. I will publish the links and information on my blog and also acknowledge those people. Many thanks!

Thursday, March 6, 2008

Using 'semicon' simulation for drug discovery

Most of us have heard of simulation in semiconductors. However, using the technique in drug discovery/pharmaceuticals is indeed unique. The Cellworks Group Inc. (CWG) has managed to achieve this unique status by developing in silico based physiologically aligned predictive analysis platforms based on a systems biology approach.

These contain quantitative and dynamic maps of metabolic and cell signaling pathways, which are relevant to several diseases like cancer, inflammation, diabetes, CNS, infectious diseases, etc. The company was founded in June 2005, with corporate office in California, USA and the R&D team stationed in Bangalore, India.

Taher Abbasi, CEO & COO, CWG and Cellworks Research India Pvt Ltd, said that drug researchers can "simulate" these in silico platforms with various disease triggers, biological target inhibition/activation to represent functional drug action, patient sub-types with the overall objective to understand the disease physiology and treatment therapies in the context of a physiological view.

Simulation in pharma
Modeling and simulation have been used very successfully in fields such as aeronautics and semiconductors. However, their adaptability within the pharmaceutical domain has been anemic until now. Physiology simulation and modeling is crucial for identifying the issues early in the development cycle and in creating more efficacious and safer drugs.

Abbasi, a former Cadence employee, said: "Coming from the VLSI/chip industry, we are used to using automation technologies, especially simulation. One of our co-founders is Dr. Shireen Vali, who has a deep scientific background. We looked at how we could bring in the concept of simulation into drug discovery." The company has just crossed 65 members and is growing. "We have covered a pretty broad range of diseases and we have also brought together people who are experts in their respective fields."

Cellworks took the idea of simulation into drug discovery by looking at diseases, identifying the biological processes involved with a particular disease, and then looking at the biological players -- such as proteins, enzymes, etc., finally leading to building a unique dynamic map -- which is underlined by mathematics, he added. "As it is mathematics, it is similar to say, looking at transistors," he said.

Cellworks has an underlying computational engine, which is an OD (ordinary differential equations) solver, and solves those. The first layer is biology, the second -- mathematics, and the third -- semiconductors. Abbasi said: "Under the hood, we have put in the best practices of the semiconductor industry, especially, the automation part. This should handle pathways, runtime capacity and analysis capability to probe the system for data."

Cellworks' goal is crystal clear. Its mission is to be the infrastructure supplier to the pharmaceutical industry, by enabling the different components of this industry to carry out drug discovery. Besides this, Cellworks also aims to become the equivalent of Goggle Maps/Earth for understanding the human physiology, disease progression and predict the biological responses and therapeutic effects of drugs.

Huge market opportunity
So what is the market opportunity in this domain? As per Cellworks' estimate, the total available market (TAM) is said to be worth $1bn+ for collaborative R&D, disease platforms and in-house drug research. The pharmaceuticals and biotech industries spend over $51bn yearly on drug research.

The vast majority of this money goes into projects that have to be cancelled because of unanticipated side-effects, toxicity or lack of potency which is discovered in clinicals when very large expenses have already been incurred. Abbasi added, "We will and we can create and own the largest and most comprehensive IP of the dynamic physiological maps and computational infrastructure for the in silico analysis of cell physiology and disease pathology."

He pointed out that the only way to reduce costs and improve on the first time succession is to prototype a drug in a virtual environment -- similar to simulation in semiconductors/chip design. "The key is to first get the biology right, followed by the focus on chemistry, which is similar to frontend and backend in chip design," he noted. "Only then can you drive the in vitro and in vivo experiments through in silico."

Cellworks has already tasted early success. So far, it has created seven base disease platforms, which are physiologically validated. It is also working on live projects with initial customers, collaborators on these platforms. These platforms contain 100+ validated sub-pathways, which model over 10,000 reactions and 30,000 parameters.

The seven base disease platforms include: oncology platform (Glioma, Breast cancer), inflammation platform (RA and IBD), metabolic-syndrome platform (Diabetes II), skin (Pigmentation), pre-term endocrinology, infectious diseases, and neuro-degenerative disease platform (Parkinson's).

Cellworks has established strong active collaborations/partnerships in several areas. Notable being the UC Davis relationship in oncology, neuro-science, pharmacokinetics and inflammation; Cedar-Sinai/UCLA on pediatrics/pre-term delivery research, NIMHANS in Bangalore, for CNS work; UoH for oncology and Indian Institute of Sciences, Bangalore for heat shock proteins.

Cellworks has created a fast and scalable software platform for validation and analysis of disease platforms. The software platform allows it to simulate and verify at the subsystem level as well as at the disease level. It enables easy integration and re-use of models, intuitive debug and analysis of the platform. Next, it has a natural language processing system for enabling pathway analysis and validation.

Abassi added: "We are working with Orchid Chemicals on an arthritis project. We had real success regarding the usage of our technology in various R&D and production projects." Most of the R&D is done out of India, with only two people based in the US for marketing/sales.

Cellworks had done a small round of funding in January 2007 worth $1mn. Abbasi said, "We will do another round of funding in Q3-08 now that we have success stories and customers in place."