Every Semiconductor Design Has To Be Compliant To Functional Safety

Every Semiconductor Design Has To Be Compliant To Functional Safety

1 SANJAY GUPTA atr apr14

A leader in embedded processing solutions for the automotive, consumer, industrial and networking markets, Freescale produces a wide portfolio of automotive MCUs, MPUs, analogue integrated circuits and sensor solutions. The largest Freescale design centre in Noida also boasts of contributing more than 60 % of the worldwide patent disclosures in the automotive domain. Sanjay Gupta, Director – Automotive & Industrial MCU Group, India Design Centre, Freescale Semiconductor India Pvt Ltd gives us a detailed insight into the company.

-------------------------------------------------

Having joined Motorola/ Freescale in 1996, Sanjay Gupta has had work assignments in the wireless business and digital networking, and as the Director – Automotive & Industrial MCU Group, India Design Centre, he has grown his organisation into the largest design centre for automotive MCU R&D with more than 340 engineers. Sanjay is responsible for execution of multiple product design efforts critical for ADAS, vehicle networking and DIS; global support for digital IP, validation and analogue & mixed signal; management support for MCU system, application and documentation teams in Noida and serving as the Automotive MCU focal point in India to drive synergy and efficiency across India. An engineering graduate in Electronics and Communication, Sanjay also earned an MBA degree from the Indian School of Business. He holds multiple US patents in the areas of state retention within a data processing system and miller cap tolerant special clock tree elements.

--------------------------------------------------

ATR _ Sanjay, the semiconductor industry in a way is driving most of the modern changes in our vehicles. Give us a sense of the current areas of work, and what could the future look like.

SANJAY GUPTA _ One of the ownerships that the Indian team carries from an R&D perspective is on advanced driver assistance systems (ADAS). We are currently working on complete end-to-end solutions on two unique systems under ADAS, which will eventually fuse together to provide the end solutions. One of them is a radar-based solution, while the other is a vision-based solution.

We have a portfolio of chips, where radar is going to be the technology used for driver assistance features. Radar-based communication with cars in front would dictate your protocols or software algorithms for the micro controllers to take relevant action. To avoid collision, for instance, there will be pre-programmed software that will dictate when the car will take an auto pilot or semi-auto pilot mode. The radar-based system would give relevant signals to the micro controllers to take charge and car will automatically start slowing down. I see a lot of cars featuring radar-based integrated technology in the next five years.

The parallel technology that we are developing is a vision-based technology. This technology features multiple cameras that will be integrated throughout the car, and will work real time. Data would be collected from, say, traffic signals and fed back to the car for the drivers to take appropriate decisions.

Significantly, the size of the chip we have done for these technologies would be one of the biggest in the automotive world. The previous biggest chip was one-third of this. This chip was kicked off in January this year and hopefully we will be delivering the first prototype next year.

2 SANJAY GUPTA atr apr14

But safety and security of data has remained a concern, especially for ADAS systems?

Indeed. In both these cases, the vehicle could take decisions that could be fatal, if they are not done properly. That's the reason we are now talking about data fusion, where we'd have one data coming in from the radar, and another coming in from the vision-based system. There would be a master controller that will take a decision only if both these mini computers convey the same message. For instance, if both these systems ask the car to park automatically, the car would do that overruling human interface.

The amount of algorithms and data we are looking at, have to be secure in nature. Wherever there is a good thing, there will be someone trying to inject a virus. The systems that we are trying to develop cannot have any virus injection possibilities; because we are talking about an era where data will be downloaded at speeds of 24 mbps in a car, speeds that are currently not even available in our homes.

In the semiconductor industry, every design has to be compliant to functional safety. This means that when an electronic function is supposed to kick in, at no point in time it can fail. There has to be a redundant system in place. There are solutions that already exist around orthogonal view or 360 degree view, for instance, but these technologies may not be fully functional compliant in terms of the ISO 26262 standards.

Miniaturisation of electronics is also fast catching up...

In the past, we used to do chips that require a board, which had two to three chips working together. We have moved on to develop single chip solutions that will help in reducing the BOM cost for our OEMs and Tier I suppliers. The size of the chip will be smaller and flexible for application based on the aesthetics of the car.

Talk to us about an interesting OEM project you were involved in.

We worked on a chip for BMW in the area of connected cars using Ethernet. BMW, along with Mercedes-Benz and Volkswagen, was part of a consortium on connected cars. There were competing technologies – Ethernet in a car or CAN-based protocols. The India Design Centre got the opportunity to design that chip for Ethernet in a car, and we were able to deliver a product that went from prototype stage to final production at one go. This also had the surround view capability. This, however, was not a single chip solution.

3 SANJAY GUPTA atr apr14

So, vehicles of the future would obviously need more on-board computing power. What sort of transition are we looking at?

We have been hearing about Internet of Things (IoT) for some time now. The latest buzzword within the Freescale R&D fraternity is Internet of Things in Cars, where an ecosystem is going to be developed for the automotive industry in parallel to the consumer systems.

Car-to-Car communication is one area that technology would drive very aggressively in the near-to-long term. We will see by the turn of this decade cars talking to each other. By that time, we also hope that features like speech recognition actually work to perfection. That is a necessity today considering the number of accidents that happen because of drivers trying to either type something on their smart phones or trying to search for something. This is going to take a very big shape in times to come, where speech is going to be a key driver for controls rather than touch buttons.

The second trend we foresee in the future is that of body languages getting recognised as actions. Gestures will play a huge role in performing actions in a car. Thirdly, within a few years we will see our cars becoming our home or office, which would allow downloading data at much higher speeds. Ecosystem partners like Google and Facebook are already conducting research on possible installation of mini robots in the sky, which will work as our Wi-Fi hotspots.

Lastly, there are instances of human lives being lost because of the lack of emergency response. In the next few years, we'll have systems in the car enabled by technology, where in the event of an accident there will be auto communication to pre-fed numbers oh family, hospital or police.

I understand the cost of electronics as a percentage of vehicle costs is forecast to rise to over 60 % in a standard car and to over 85 % in hybrid vehicles over the next decade. Your comments.

The industry has indeed moved from the early stages of MPFI engines in cars, when the electronics content in vehicles wasn't more than five per cent. Today, we have multiple chips in engines for better fuel economy and improved emissions. We have chips for infotainment, Bluetooth connectivity and navigation. Although the acceptance of many of these technologies is still nascent in emerging markets, the gap with matured markets is fast decreasing. I have no reasons to doubt that the cost of electronics as a percentage of vehicle costs would be somewhere in the range of 50 % in a standard car if all the innovations we are talking about find acceptance.

Battery management system is also an important area, especially for hybrids and electric vehicles. Talk about the new developments by Freescale in this area.

Today, we don't have batteries that are smart enough to tell us digitally how much of battery charge is left, or how fast it is degrading. Batteries should be smart enough to be able to talk to the vehicles. By Internet of Things in Cars, I meant the possibility of batteries telling the car that I'm not left with enough charge, so you should switch to an alternate mode.

If the next charging station is 10 km away, the communication network in the vehicle should enable the battery to decide and direct the driver to switch to a low-performing mode, or automatically tell the powertrain to reduce speed so as to conserve fuel for a longer time. The entire ecosystem – the gas station, batteries and cars – needs to talk in a secured manner, and that is what we are driving at Freescale. Technologically, battery management system is a tremendous growth opportunity for us. It's a great innovation area.

I understand you offer an AUTOSAR OS to your customers. While standardisation is important, I believe AUTOSAR is not suited for every embedded software architecture application due to its wide-ranging diversity. Is there a business case for a simpler, modular, and universal architecture?

I agree it's a little controversial. We see definite advantages because it gives commonality to at least a certain level and we can build on that. Having said that, I do feel that software is an area that will grow exponentially because for all these application we've talked about. Software will be the key enabler.

The slowdown in the automotive market certainly doesn't impact innovations and new product development for the future.

Cyclical blips are a part every business, and the semiconductor business is no exception. The auto industry, to my mind, is much more stable. As long as we continue to focus on innovation and technology, we don't expect to be effected much. There will be a time when we'll emerge much stronger from the downturn because we'll be ready with all cutting edge products. From an automotive standpoint, the Indian team is contributing more than 60 % of the worldwide patent disclosures, which amount to over 125 patents in the last five years.

--------------

Text: Deepangshu Dev Sarmah

Photo: Bharat Bhushan Upadhyay