IP providers make plans for the internet of things

By Chris Edwards |  No Comments  |  Posted: August 29, 2013
Topics/Categories: Blog - EDA, Embedded, IP  |  Tags: , , , , , , ,  | Organizations: ,

ARM’s purchase of Finland-based Sensinode sees the processor designer move further into its promotion of the Internet of Things (IoT), with the aim of using the acquired software to speed up the development of devices that need low-power wireless communication.

Sensinode was involved in the creation of the 6LowPAN and CoAP standards for low-power wireless sensor networks and has contributed to a variety of standardization efforts including that of Zigbee and the Open Mobile Alliance (OMA) protocols for machine-to-machine communications. ARM aims to provide access to the Sensinode protocol stacks for ARM processors alongside its open-source Mbed software – the idea being to speed up development and prototyping of IoT applications.

Mike Muller, CTO of ARM said in a press meeting in London, UK a couple of weeks ahead of the acquisition he sees IoT as a major market that will change the way that companies with physical infrastructure will use IT. He cited the example of SFPark, based in San Francisco, CA. As part of a trial, SFPark, part of the city’s municipal transportation authority, is deploying sensors across several thousand on-street and garage-based parking bays that will detect whether a car is parked – and relay that data in real time to the company’s servers, which in turn feed information on the closest parking bays to drivers using a smartphone app.

Parking projects

A similar project got underway several years in the Spanish city of Santander, burying magnetic sensors able to detect the body of a car parked over them in parking bays, with the aim of reducing traffic congestion – on the basis that a lot of it is down to drivers circling the city looking for somewhere to park.

SFPark also intends to use the real-time data to perform demand-based pricing on a finer level of granularity than it is doing already. When popular events are on in central San Francisco, expect the parking fees to increase there. But garages in the city that are further away might see falls.

“You can change the way you make money out of a car park. It’s about traditional businesses and the technology that could help make money out of their assets,” says Muller, arguing that much of the growth in IoT will be not so much in consumer-focused applications or large projects such as the smart-grid rollout but less publicly visible areas in the middle ground. “The ‘I’ might not always stand for ‘Internet’.”

Design tradeoffs

The question facing sensor-node designers looking at the IoT as a future market is the tradeoff between custom and off-the-shelf silicon. The number of cars that use multiple magnetic sensors to park is likely to exceed the number of public spaces that will benefit from sensors able to detect the vehicles arriving and leaving. As a result, the in-car sensors benefit from being custom designed on a reasonably advanced process whereas the parking-bay sensors are more likely to remain based on standard microcontrollers, many of which are now being designed to capture a chunk of the IoT business, that use much older processes than those at the leading edge today.

For its foray into IP for the IoT, Synopsys anticipates a requirement for custom sensor designs, particularly for systems that use multiple sensors. In the short term, this is more likely to be embedded in mobile phones where a variety of MEMS sensors will provide information to a central controller to help it make better judgments on how the device is being used and where it is. However, it is possible to see sensor fusion moving out from the phone into more industrially focused system designs.

A problem for the parking-bay sensor is how it copes with the problem of a full-sized parked in a compact bay. A simple magnetic sensor mounted in the middle of each bay may not detect this situation effectively and wind up guiding drivers to a space they cannot use. Employing multiple sensing techniques will demand more processing power but at low energy-consumption levels.

Process choices

Muller says he believes “at the microcontroller end of the spectrum you can afford to go to 65nm or 40nm and apply some innovation. Maybe there is some analog innovation that will tell you the make of car in your parking-bay sensor”. But he adds: “If you think your car-park sensor needs 14nm finFET, you are prematurely coming to market. You want to do 20nm, 16nm, 14nm finFET design for high-end mobile tablets as you are talking tens of millions of dollars to do a design. And it’s a complex design at that. Quite a lot of the value has been taken in mobile. But the seemingly more tedious embedded market probably accounts for most of the volume. You can look at the IoT as a rebranding of embedded.”

Many of today’s low-energy microcontrollers, a number of them based on the ARM Cortex M0 or M3, put the sensor interfaces on a peripheral bus. Some of the more advanced designs incorporate simple processors or state machines that allow data to be retrieved automatically without waking the host processor – keeping it in low-power sleep modes for longer. For its IoT offering, Synopsys reckons that designers will want higher-performance options and has developed a set of peripheral functions for its customizable ARC 32bit processor.

Bus-free design

Rich Collins, ARC segment marketing manager, claims: “We can eliminate the bus by pulling components into the CPU island, which reduces latency. Performance will be better and it cuts area and power. You can save 10 to 20 clocks per access by not going out on the bus.”

Custom logic added to the 32bit ARC processor then serves to perform hardware acceleration. “Processing with state machines is not going to be enough for performance,” Collins argues. “We feel that this level of performance is needed because there aren’t individual sensors anymore [in mobile phones]: they will be using sensor fusion.”

Hardware accelerators will be useful for improving performance and keeping energy consumption down, Collins says, pointing to the example of motion sensors used to determine whether a phone should be in portrait or landscape mode. “The algorithms are very square-root intensive. With hybrid processing, split between the processor and logic, you can get significant reductions in calculations and costs with 1000 gates total versus 5000 for a pure hardware implementation.”

Right now the main targets are sensor processing in mobile phones – an area that STMicroelectronics’ MEMS division chief Benedetto Vigna sees as crucial to the IoT – which are likely to be separate chips to the applications processor, although higher-volume IoT applications could justify the investment. “We think 28nm is the leading edge for this type of thing,” says Collins. “Most customers at the moment seem to be in 90nm and then planning to migrate to 65nm or 40nm.”

Comments are closed.


Synopsys Cadence Design Systems Siemens EDA
View All Sponsors